Market Makingand Risk Managementin Options Markets

Market Making and Risk Management in Options Markets
Naomi E. Boyd
Department of Finance, West Virginia University, Morgantown, WV 26505, USA
Abstract
This article examines the personal trading strategies of member proprietary traders in the
natural gas futures options market. Trading activity is found to mirror previous findings in
futures markets, specifically high frequency trading, with low risk exposure. The portfolio of
risk holdings by member proprietary traders are also examined to identify whether they are
instantaneously hedged using the underlying futures market, as well as to investigate how
they manage their inventory holding, rebalancing, and volatility risk exposures. Findings of
longer-term risk management practices by option markets indicate that instantaneous hedging
does not take place in this market. Exposure to price and volatility risks is actively managed,
while rebalancing risk exposure has a significant impact on profit for this trading group.
I would like to thank Peter Locke for his invaluable insights and comments, Li Sun, participants of the 2008
Financial Management Association doctoral consortium, the 2009 Financial Management Association and
Southern Finance Association meetings, seminar participants from West Virginia University, Kansas State
University, Clemson University and the University of Rhode Island for their helpful comments.
Naomi Boyd was a Consultant, Office of Chief Economist, Commodity Futures Trading Commission (CFTC),
Washington, D.C. when this research was conducted. The ideas expressed in this paper are those of the authors
and do not necessarily reflect those of Commodity Futures Trading Commission or its staff.
* Corresponding author: Tel.: +1-304-293-7891; fax: +1-304-293-5652.
1
INTRODUCTION
The services of liquidity production and price determination that market makers provide
serve a vital role in the proper and orderly functioning of the aggregate financial market
system. Without the presence of market makers, the system would be less efficient and more
costly to maintain. How and why market makers provide these services should be influenced
by the structure under which they operate. Market makers can be categorized within two
central structures: (1) a designated structure under which the market maker has an assigned
role and is obligated to perform certain duties including, but not limited to, providing
liquidity, filling orders, setting prices, and maintaining price continuity or (2) an open
structure under which a market maker is governed only by the rules set forth by the exchange
for all traders.
U.S. futures options (commodity options) operate under an open trading structure
which mimics the futures trading architecture, where members may broker orders or trade for
their own accounts with little constraint on their trading strategies. Floor traders (or their
screen trading equivalents) who choose to make a market with their trading are not required
to maintain inventory or price continuity and can enter and exit the market freely and without
constraint, other than risk controls imposed by their brokerage account. They have no
privileged access to pending orders, but at least on the floor have first-hand observation of
trading activity and the shouting and gesturing which predate a trade. Each trade must be
categorized as a trade for the executing members own account, the member’s clearing firm
account, another member’s own account, or a non-member account (customer). A member’s
income can be derived from proprietary trading and brokerage.
1
Research on futures markets has established that member proprietary trading serves
the market-making function in futures markets with trading behavior that is characterized by
low end-of-day inventory holdings, small trades, and large volumes (Silber 1984, Kuserk and
Locke, 1993, 1994). Related work shows highly volatile, short run profitability to futures
proprietary trading (Locke and Mann 2005). Proprietary options’ trading in natural gas
futures options is examined here to describe the institutional details of option market maker
behavior.
It is shown that member proprietary trading in futures-options markets serves the
market-making function just as was previously documented by Kuserk and Locke (1993) in
the corresponding futures markets. This finding documents one of the many similarities
between how these markets operate. Member proprietary traders are characterized by their
small trade sizes, high volumes, small amount of time between trades, and low end-of-day
inventory levels. These characteristics are indicative of a market maker performing the tasks
of providing liquidity and price setting. This analysis also shows that these traders are
profitable on average, albeit at very low levels. Therefore, market making is a viable,
profitable trading strategy for these traders.
Since there is an active group of market makers in the futures market supplying
liquidity and making that market, futures options market makers would not benefit from
similarly making a market indirectly in the futures price. Instead, as Figlewski (1989) points
out, a trader making a market in options will benefit from eliminating futures price risk and
concentrating on the other risk factors. Trading options, unlike futures, involves more than
univariate risk management. In addition to exposure to changes in the futures price, the
option value is also affected by changes in the other factors which influence the option value,
2
most notably the expectation of futures price volatility. Management of the exposure to the
futures price can be easily accomplished by trading futures. Management of the exposure to
volatility which arises in options trading requires trading in other options. In terms of
common option parameters, delta represents the exposure to futures price risk, and vega
represents exposure to volatility risk1. If option market makers use the underlying futures
market to hedge away their exposure to price risk, they could strategically manage their
residual risk and potentially earn higher profits than those who chose other trading strategies.
In particular, we focus on the risks taken by proprietary traders, in terms of
sensitivities to parameters in an option pricing model. The manner in which these traders
manage the various risks allow us to infer market making as a strategy and describe
particular characteristics of that strategy. We focus on the ability of option market makers to
dispel inventory holding risk through simultaneous participation in the underlying futures
market. The hedging practices of market makers have been shown to reduce the costs of
providing liquidity (Çetin, Protter, and Warachka, 2006) as well as have been shown to have
a direct impact on the size of bid-ask spreads in option markets (Huh et. al. 2012). Thus, how
option market markets hedge has direct implications into market frictions such as
transactions costs which will influence the price setting practices of these traders. We also
evaluate these traders exposure to gamma and vega risk to examine whether they maintain
level or changing levels of risk throughout the trading day on average.
This paper is the first known study to attempt to formally and empirically test the
theories posited by Figlewski (1989) and Cox & Rubinstein (1985) that option traders
maintain delta neutral positions by establishing corresponding and offsetting trades in the
1
For example, if the delta of an option is .5, then a $.10 change in the futures price will lead to an approximately
$.05 change in the option value, holding all else constant. By selling one futures contract for every two options
contracts, this will temporarily eliminate the futures price. The residual risk would be the other option pricing
factors, such as volatility.
3
underlying futures market. Thus, it is thought that option market makers engage in hedging
by (for example) purchasing a quantity of futures options while simultaneously executing a
certain number of contracts in the underlying futures contract, seeking instantaneous delta
neutrality. The degree to which these traders hedge by participating in both markets will
determine their vulnerability to the risk of holding positions in the option market. The short-
run exposure of the futures-option market maker to price risk, delta, is expected to be small if
they are instantaneously delta-hedged while the exposure to volatility, vega, may more
accurately represent the “inventory” that the market maker is carrying.
We find that futures-option market makers hedging practices do not coincide with
instantaneous hedging. Rather, their use of the underlying futures markets reflects a longer
term price risk management strategy. This type of strategy would be driven by option market
makers utilizing the underlying futures market to hedge when they cannot easily liquidate
their inventory in the options market or when they wish to hold onto a position for an
extended period of time (intraday) to allow prices to equilibrate or move into profitable
territory. We also note that large traders actively seek to hedge using the futures markets,
which is in line with their correspondingly larger inventory holdings; while smaller traders
tend to keep most of their trading centralized in the option market. On average, futures-
option market makers maintain very low levels of price risk over the course of a trading day
while their levels of rebalancing risk and volatility risk are much higher. Thus, futures-option
market makers’ primary exposure is to gamma and vega, or the speed at which delta changes
in response to a change in the underlying futures price and the response of the option price to
a change in the volatility of the futures price, respectively. In fact, analysis the impact of the
4
three risk parameters on daily profits reveals that gamma risk has the most substantial impact
on the profitability of an option market maker.
The remainder of this article is organized as follows: Section II provides information
regarding the data; Section III investigates the behavior of market makers in options markets
through a descriptive analysis; Section IV analyzes how market makers hedge their exposure
to price risk as well as details their intraday rebalancing and volatility risk exposures in the
option market; Section V concludes.
II. DATA
The data for this research consist of 20 months of transaction-level data in the natural-
gas futures and futures option markets traded on NYMEX (now part of the CME group),
spanning September 2005 through April 2007. This data set is maintained by the U.S.
Commodity Futures Trading Commission and comes from the computerized trade
reconstruction (CTR) records compiled and maintained by the agency from data feeds from
the exchanges.
Trading rules specify the obligations of the floor traders and how trades are to be
executed and recorded. When trades are executed on the floor, the trader on the sale side, for
either a futures or futures option, reports the transaction to the exchange for clearing. The
exchange records the relevant information regarding which futures or options expiration is
traded, the price or option premium, strike price, and information about the counterparties.
The information also requires an indication by the broker as to whether the trade was for a
proprietary account, another floor trader, the traders clearing member, or some outside
customer.
5
Our dataset includes the records of all options floor trades and provides information
including the price, quantity of the trade, trade date, trade time to the second, trade direction
(buy or sell), delivery month and year of the contract, customer type (trade for the member’s
account, his or her house’s account, another member on the floor, or a customer),
counterparty’s customer type, and the floor trader’s identification. We also use corresponding
futures and options daily settlement prices and daily interest rates (3-month T-bill).
As in previous research in this area, traders who executed personal options trades
infrequently are removed from the sample. We calculate a measure of incumbency as the
percentage of all possible days that each trader executed proprietary trades, and drop those
traders who participated on 5% or less of the days, reducing the number of options traders
from 144 to 91.
Natural gas futures and options trade in the physical environment from 9:00 am to
2:30 pm (ET). The futures contracts trade in units of 10,000 British thermal units (mmBtu)
and contracts which are not eventually offset require physical settlement. The minimum
price fluctuation is $0.001 per mmBtu, or $10 per contract. Typical natural gas prices range
from $5 to $15 per 10,000 mmBTU, yielding a notional contract value of $50,000 to
$150,000 dollars.
When delivery does occur it is a throughput at the Sabine Pipe Line Company’s
Henry Hub in Louisiana. This can take place no earlier than the first calendar day of the
delivery month and no later than the last calendar day of the delivery month. The Henry Hub
is a many natural-gas pipelines that serve markets throughout the U.S. east coast, the Gulf
and the Gulf Coast, and the Midwest up to the Canadian border. The futures seller is
6
responsible for the movement of the gas through the Hub and pays all Hub fees while the
buyer is responsible for movement from the Hub.
III. MARKET MAKER TRADING STRATEGIES
We follow previous research by Working (1967, 1977), Silber (1984), and Kuserk
and Locke (1993, 1994) to examine whether traders carrying out member proprietary trades
take on the role of market makers within the competitive framework of an option market.
Due to the link, both institutionally and through arbitrage conditions, of the futures and
futures-options markets, it is an empirical question as to whether the characteristics of market
makers in futures markets universally hold in the option market.
A. Market Making Activity
The positions and levels of market-making activity are determined through an
examination of several variables that have been shown to distinguish market makers from
other types of floor traders in the futures market. These descriptive statistics for active,
proprietary traders over the first three nearest contract months for options are provided in
Table 1, Panel A. Total trades documents how many different transactions were entered into
over the sample period. Similarly, total contracts are the total option contracts bought by
these traders. The average time between trades is calculated by first taking the average
number of minutes between each trade each day, and then averaging this number across
trader days. There are a total of 15,573 trader days in the sample, for our 91 traders over the
413 of days in the sample. As is shown in Panel A, on average option proprietary trading is
similar to that of proprietary trading in futures markets: traders maintain relatively low levels
of inventory and provide liquidity services to the market by trading frequently, in small
amounts, with high levels of overall volume.
7
[Insert Table 1 about here]
B. Market Making Revenue
Member proprietary trader’s income can be derived from proprietary trading and
brokerage. While the analysis that follows is meant to show whether member proprietary
trading generates positive levels of income for the trader on average, certain macroeconomic
and market specific factors may also influence the level of income during certain time
periods. For example, the relationship between commodity-equity cross-market linkages
could drive profitability for traders. This issue was examined by Buyuksahin and Robe
(2013) who documented that hedge funds that are relatively unconstrained are the only set of
traders who provide these linkages in participation. While member proprietary traders’
income from their brokerage functions may contain a portion stemming from trading with
hedge funds, it is not possible to disentangle who the outside customers are in our dataset.
Cross market linkages with other commodity markets may also impact member
proprietary trading activities and profitability. In the past, prices of crude oil and natural gas
followed closely with one another and major users viewed the products as close to perfect
substitutes (Onur, 2009). Research has shown that this relationship has diverged (Serletis
and Ricardo, 2004; Baschmeir and Griffin, 2006; and Serletis and Shahmoradi, 2006). In the
short term, prices of oil and natural gas are driven by different fundamental factors with
crude oil prices fluctuating in response to speculative activity in world oil markets and
natural gas prices responding to productive shocks. Azzarello et al. (2014) document a 330%
energy content price gap, which is primarily attributed an increase in production in natural
gas. Income stemming from brokerage functions would be influenced by cross-commodity
linkages, not necessarily income from proprietary trading.
8
Here we examine the distribution of income for active, proprietary trading in Panel A
of Table 2. Daily average income for options (in dollars) for each proprietary trade across
the nearest three expirations is found by marking to market each trade over the course of a
trader day, summing the income for each individual trader, and averaging the income for
each trader over all trader days by contract expiration. If the trade is a sell, the income is
found by taking the difference between the trade price and the settlement price and
multiplying by the quantity. If the trade is a buy, the income is found by taking the difference
between the settlement price and the trade price and multiplying by the quantity. The
quartiles of daily income are found from the total daily income levels for each trader. Thus,
the minimum corresponds to the lowest level of income made by an individual trader during
the sample period for a specified contract expiration. Panel B in Table 2 on the other hand
details the distribution of average daily proprietary income across all traders each day. Thus,
we have two views of income: among traders and across traders.
Both tables provide similar results: 25% of traders are unprofitable, while the mean
profit levels, albeit small on average, are positive across traders and days; however Table 2
documents a slightly skewed distribution with more traders taking a positive profit than a loss
on average. The low levels of profitability beg the question of why member proprietary
traders are willing to provide liquidity to the market, if one average, they are only making
low levels of returns. A negatively skewed return distribution will increase the loss
probability, while a positively skewed return distribution will increase the probability of
gaining. A preference for positive skewness will cause investors to require lower rates of
returns on these assets (Barberis and Huang, 2008; Boyer, Mitton, and Vorkink, 2010), and,
as such, a preference for skewness represents a desire to gamble. Given that option traders
9
often trade based on skew, the relatively low levels of profit seen on average could be the
result of member proprietary traders being willing to take lower average daily returns for the
possibility of big upside potential of these highly skewed assets.
C. Market Making Competition
Futures and option markets have systems that can be identified as having market-
making competition, which is in stark contrast to the designated system of the NYSE
specialist. Competition among dealers has been found to lower spreads (Stoll, 1978; Benston
& Hagerman, 1974; Tinic & West, 1972, Wahal, 1997; Klock & McCormick, 1999).
Understanding the dynamics of the price setting and trading behaviors of market makers
requires an evaluation of the competitive forces among this group of traders. The average
number of traders is documented across trade type and contract expiration in Table 3 to
determine the extent of competitive forces in each of the various trade categories. Traders in
these markets can conduct more than one type of trade. The vast majority of traders are
classified as making CTI = 1 trades, which indicates they own 10% or more in the trading
account for which they are trading. On average there are 51 traders executing trades of any
type and maturity (first three nearest to expiration contracts) per day.
D. Interdealer Trading
In inventory microstructure models market makers face exogenous demands to buy
and sell and profit from buying and selling at bid and ask prices respectively, with the spread
dependent on the underlying risk of the asset. Bid and offer placement are adjusted to
manage inventory risk (e.g. Ho and Stoll, 1983; Manaster and Mann, 1996). Further, as
10
discussed in Locke and Sarajoti (2004), interdealer trades are typically more costly to initiate
than trades conducted with other trade groups; thus, the only rational explanation for the
existence of a significant percentage of interdealer trades is inventory control. These traders
would rather immediately transfer their unwanted inventory than face the uncertainty of
waiting for customer orders. Table 4 details the percentage of trades by customer type in the
options market through a frequency analysis of trade combinations across the three nearest
expiration contracts to examine the extent of interdealer trading in the options market. The
levels of interdealer trading range from 16.33% in the second deferred contract to 22.83% in
the nearby contract.
E. Summary of Market Making Activities
The institutional characteristics of futures-options markets indicate that on average,
member proprietary traders in this market trade often, in small amounts, with very little time
in between trades, and are responsible for one of the highest levels of activity in terms of
volume. Thus, these traders serve similar market making roles in both the futures and futures
options markets. Further, it is shown that the levels of competition among these traders are
higher than other trade type categories and member proprietary traders are prone to engaging
in interdealer trading. Competition will lower the average levels of profitability for these
traders, as will engaging in interdealer trades. These two factors may be the driving forces
behind the low levels of profitability within this trade type. However, even given low levels
of profitability, market making in options futures markets is a profitable trading strategy.
These summary measures shed first light on the institution framework under which member
proprietary traders make their market. In order to understand more fully the constraints that
11
these traders face and how these frictions can help explain the behavior and price setting
practices of market makers in futures-option markets, the risk structure of member
proprietary traders follows.
IV. RISK MANAGEMENT
Analysis of market-making risk dynamics in options markets has received very little
attention in the literature and has primarily focused on the overall risk that option market
makers bear. The risk exposure of the option market maker will determine how and whether
the market maker is able to dispel certain portions of risk which will ultimately drive his or
her trading and price setting behavior. Agarwal and Narayan (2004) characterize risk
exposures and portfolio decisions involving hedge funds, and they find that investors wishing
to earn risk premia associated with different risk factors need to differentiate hedging
strategies. Research has shown that the risk that option market makers are exposed to greatly
depends on the stochastic nature of the underlier’s returns.
Ho and Stoll (1983) showed that, if the stock return volatility is constant, the market
maker’s risk exposure per dollar of investment is nonstochastic over the interval during
which the inventory is held. However, unless the option market maker can trade
continuously, an option transaction’s contribution to the dealer’s risk exposure will be
stochastic because the volatility of an option is a function of both its (stochastic) hedge ratio
and the underlying asset’s return volatility.
Jameson and Wilhelm (1992) evaluated the risks that options market makers face and
provided empirical evidence that their risk factors are unique to option markets due to the
stochastic volatility of the stock return and the inability to rebalance the option position
continuously. Specifically, by measuring the impact of delta, gamma, and vega on bid-ask
12
spread, they found that gamma and vega are significant determinants of spreads. These risks
can be reduced through diversification; however, the authors’ finding of significant influence
of the risks on spreads indicates that diversification does not completely eliminate the risk
due to discrete rebalancing and stochastic volatility on the option market makers’ portfolio.
O’Hara and Oldfield (1986) utilize a dynamic framework of market makers facing
uncertainty with future order flow and future value of underlying equity, and demonstrated
that inventory and risk preferences have a pervasive influence on the market makers’ pricing
policy, influencing both the size and placement of the spread. These results were further
examined by Giannetti, Zhong, and Wu (2004) when they developed an inventory-based
approach to study market-making behavior in option markets. They posit that the hedging
practices of option market makers have a substantial impact on the setting of bid-ask spreads
and optimal inventory control. By adding hedging mechanisms to the standard inventory-
control model, the authors derived the market makers optimum option quote setting and
inventory-control policies. Huh, et al. (2012) further develops a theoretical model which
evaluates the relationship between hedging practices of option market makers and the size of
the bid ask spread. Here it is posited that the bid-ask spread increases when option market
makers use the underlying market to hedge visa-a-via using the options market to hedge.
Of vital importance to studying market maker behavior in option markets is
evaluation of these traders’ exposure to and management of risk. Option market makers’
primary exposure to risk comes from price movement in the option and underlying asset
markets, rebalancing needs, and volatility of the underlying assets. As noted above, much of
the previous research in the literature has focused on the price movement and stochastic
nature of the underlying assets. However, to determine the overall extent of option market
13
makers exposure to risk, inventory positions must be examined in terms of a vector of risk
measures, which include delta, gamma, and vega. Exposure to these risks is due to the
influence of certain variables on the option price. Several variables are known to affect
option prices: the price of the underlying asset, the options strike price, the time until
expiration, the volatility of the price of the underlying asset, the risk free rate, and the value
of the dividends expected during the life of the option.
The primary purpose of this article is to explore the activities and risk management of
market makers in options markets. Here, the sensitivity of the option price to both the price
of the underlying asset and the volatility of the price of the underlying asset are evaluated
through an examination of delta, gamma, and vega. The underlying volatility not only plays
a role in the valuation of the options being traded, but also in the ability of market makers to
eliminate their exposure to price and volatility risk. The predominant issue is whether
market makers in options futures markets maintain instantaneous delta neutrality as posited
in the theoretical literature. If market makers establish positions in both the underlier and in
options that are hedged with respect to fluctuations in the price of the underlying asset, their
portfolios will be delta neutral. It is found that the member proprietary traders examined in
this article do not hedge instantaneously, which exposes them to fluctuations in the value of
their portfolios when the underlier fluctuates.
Delta measures the degree to which an option price will move given a change in the
underlying asset or, in this case, the sensitivity of the option to the futures price. The delta is
often called the neutral hedge ratio; with a portfolio of n shares of a stock, n divided by delta
gives the number of calls needed to be written to create a neutral hedge. A positive delta
indicates that the option position will rise in value as the stock price rises and drop in value
14
as the stock price falls. A negative delta, on the other hand, means that the options position
should rise in value if the stock price falls and drop in value if the stock price rises. The delta
of a call option can range from 0 to 1, whereas the delta of a put option can range from -1 to
0. Thus, a short call has a negative delta, and the long call has a positive delta with these
values reversed for puts and the same for stocks. The closer an option’s delta is to either -1 or
+1, the more the price of the option will respond like an actual long or short stock when the
stock price fluctuates.
Gamma indicates how much the delta changes for a $1.00 change in the stock price. It
is the second partial derivative of the option price with respect to the underlier. If gamma is
small, delta changes slowly, and to keep a portfolio of options delta neutral one can rebalance
the portfolio less frequently. If gamma is large, delta is extremely sensitive to changes in the
underlying price, and, therefore, the portfolio will have to be adjusted more frequently. Both
long calls and puts have positive gamma. That is, long call positions will have deltas that
become more positive and move towards 1 when the underlying price changes but move
toward zero when the underlying price falls. Long puts will have deltas that move toward -1
when the stock price falls and move toward 0 when the stock price rises. Short calls and puts
have negative gamma; thus, the opposite effects take place. Futures will always have a
gamma of zero because the delta value is always 1.0; thus, it never changes.
The vega of an option indicates how much the price of the option will change as the
volatility of the underlying asset changes. Vega is calculated to show the theoretical price
change for every 1% point change in implied volatility. Long calls and puts both have
positive vega, which indicates that the value of the option will increase as the volatility
increases and decrease when volatility decreases. Short calls and puts both have negative
15
vega, which means that the value of the option will increase when volatility decreases and
decrease when volatility increases. Vega is the greek which has the most impact on option
prices second to delta. Jameson and Wilhelm (1992) showed that an options gamma and vega
were important in the determination of option market makers’ bid-ask spreads and provided
an indication of the inventory-risk exposure these traders faced.
The risk characteristics of a portfolio of options can be described by the sum of the
risk characteristics of the portfolio components. Central to the evaluation of the risk
characteristic that provides the most information about exposure to inventory-holding risk is
whether market makers in option markets maintain delta neutrality by hedging their trades in
the option market with offsetting positions in the underlying futures market. While many
have speculated this to be the case, due to data limitations, it has not been formally tested.
This unique data set facilitates the evaluation of traders’ positions in both the futures and
option markets, which allows for incorporation of information about whether these traders
are maintaining delta neutral positions by conducting simultaneous, off-setting trades. Two
issues are addressed in the analysis that follows: (1) Do option market makers use the
underlying futures market to maintain delta neutrality? and (2) How are market makers
managing their exposure to sources of rebalancing and volatility risk? These questions are
addressed by evaluating the market maker’s risk holdings at varying intervals intraday.
A. Position Risk Parameters
If a trader simultaneously trades in both the option and futures markets, position delta
will reflect the extent to which the trader is maintaining a delta-neutral portfolio at the end of
each day by creating offsetting trades from participation in both markets. If the trader is only
participating in the option market, the position delta is calculated using only the trades from
16
the option market and will also provide information about end-of-day delta neutrality. The
gamma and vega of futures are zero; thus, those values incorporate only information about
trades in the option market.
In order to facilitate the analysis of the position parameter levels, several simplifying
assumptions must be made. First, it is assumed that traders conducting member proprietary
trades begin each trading day with an inventory level of zero. Manaster and Mann (1996)
provided evidence that daily changes in inventory are concentrated around zero, so we follow
the previous literature which assumes that all traders begin the day with a zero-inventory
position. Second, while the analysis for the option market is performed by broker
identification numbers, which are unique for a particular trader, in order to track trades from
the option to the futures market, account numbers must be used when matching trades in both
markets2. The trades are matched by account number and time; thus, it is assumed that if a
trade occurs within a specified time frame for the same account, it is instigated by the same
broker. There may be more than one broker per account number; therefore, noise will be
introduced into the matching process. Finally, trades in the options market will contain only
those performed by traders conducting CTI 1 trades, whereas they are matched with both CTI
1 and CTI 3 trades from the futures market because a trader in either category of trade in the
futures market could in practice be executing offsetting trades for the option market makers.
In order to calculate the parameter estimates, a price series must be formed for the
option and futures markets. There are two issues to address in forming a matched price series
for the option and futures markets: (a) which contract expiration to use and (b) how to
address nonsynchronous trading issues. The first issue arises because the contract with the
2
A subsample was also studied that evaluated the position risk positions matched by executing broker IDs after
electronic trading in futures markets began because in theory option market makers could simultaneously trade in
both markets using hand held devices. Our findings were insignificantly different and robust to the matching
procedure using account numbers.
17
highest volume may not be the nearby contract. Volume is well known to be a proxy for
information and is highly related to open interest; thus, we use both the nearby and first
deferred contracts which contain the highest overall levels of volume for this analysis. The
second issue arises because futures markets are much more active than the relatively illiquid
option market; thus, issues involving nonsynchronous trading must be addressed. NYMEX
requires that trades be reported within one minute of execution, so we aggregate prices over a
one-minute time span in order to form a price series that reflects the volume weighted
average price for a minute for both the futures and options markets.
Volume-weighted average prices for observations from proprietary trades (CTI 1
trades) for the nearby and first deferred contracts are computed over one minute interval for
both the option and the futures markets. The volume weighted average prices are found by
multiplying the trade price for a given observation by the quantity traded at that price with
the average taken over all observations in a minute. The last observation at each strike, for
each option type (put and call) is taken along with the futures-settlement price in an
increment, where the last volume-weighted average futures price in the increment is used as a
proxy for the settlement price. These values are used in a binomial pricing model to estimate
the option premiums.
The binomial option-pricing models the underlying instrument over time, as opposed
to a particular point in time; thus, it is used to allow for the early exercise component of
futures contracts. Also known as American-style options, which can be exercised at any point
in time, early exercise is a unique feature of options on futures contracts that stems from the
minimal time value associated with in-the-money futures options. It is advantageous to
18
exercise in-the-money futures options early and reinvest the proceeds at the risk free rate in
order to earn a higher overall rate of return, which unique to these types of options.
An implied standard deviation is used as a proxy forσ F which, along with the time
duration of a step t, measured in years, is used to calculate the probability that the price of the
underlying asset will move up or down at each step in the binomial tree. This implied
standard deviation is calculated from the most actively traded, near-the-money option for the
settlement futures price3. A grid search is used to find the implied standard deviation, which
minimizes the mean-squared error over the trading day by comparing the average option
premium to the observed premium for each hypothetical sigma.
This method ensures that the tree is recombinant, which reduces the number of tree
nodes and speeds up the computation of the option price. This property also allows for the
underlying price to be calculated directly from a formula at each node rather than from
having to build the entire tree. It is well known that option valuations cycle from high to low
as the number of steps increases, holding time to maturity constant; therefore, two separate
steps are used, 30 and 31, to calculate the average option value for each hypothetical sigma.
Once the premiums, futures prices, and implied standard deviation have been found,
the delta, gamma, and vega are calculated as in Hull (2000). An estimate of delta, gamma,
and vega are computed for each strike price and option type (put and call) in each increment.
The estimated risk parameters are merged back with the trader level data to compute position
parameter values for each trade. This is done by summing the quantity of trade for a
particular strike and option type for each trader in an increment, which is then multiplied by
the corresponding (option type and strike) estimated parameter values to compute the
3
Bloomberg implied volatilities were also used with no significant changes in the results. Thus, our estimation of
the implied volatility is robust to the methodology described.
19
position parameter value for each trader in each increment. The position levels are marked to
market at the end of each increment to account for open positions (either long or short) in the
computation of the position parameter levels.4 These position parameter values are examined
in greater detail to examine option market makers exposure to various sources of intraday
risk in the sections that follow.
B. Delta Neutrality
The theoretical literature on option risk management postulates that option market
makers may hedge inventory risk exposures by maintaining delta-neutral positions
(Figlewski, 1989; Cox & Rubinstein, 1985). This would require option market makers to
engage in hedging by (for example) purchasing a quantity of futures options while
simultaneously executing a certain number of contracts in the underlying futures contract,
seeking instantaneous delta neutrality. The degree to which these traders hedge by
participating in both markets will determine their vulnerability to the risk of holding
positions in the option market. The short-run exposure of the futures-option market maker to
price risk, delta, is expected to be small if they are instantaneously delta-hedged while the
exposure to volatility, vega, may more accurately represent the “inventory” that the market
maker is carrying.
As specified above, the futures and options data are matched by account numbers and
time to determine whether (1) instantaneous hedging is taking place and (2) if instantaneous
hedging is not being used, examining the process by which option market makers use futures
markets to hedge during the course of a trading day. Table 5 reports the results that depict
whether option market makers are instantaneously delta neutral by matching the options and
4
Marking the position parameter levels to market each increment entails summing the positions of each increment to
carry forward the balance (if the trader is net long in the increment) or debit (if the trader is net short in the
increment) of trades. The balance is then added to the first trade in the increment and multiplied by the increment’s
parameter estimate to calculate the increment’s parameter position level.
20
futures data by account number and time; where we make the assumption that if a trade takes
place under the same account number in the same minute the futures trade corresponds to the
options trade that was executed in that same time frame. We find that contrary to theory,
option market makers do not maintain instantaneous delta neutrality.
Due to this finding, an additional analysis is performed to determine whether option
market makers tend to specialize in hedging in a particular market. To evaluate whether
option market makers specialize in hedging only in options or in futures the first month of the
sample, September 2005, is evaluated. A frequency analysis of the number of traders
engaging trades in both the futures and option markets versus the option market only is
performed. The frequency provides a count of the number of trades for a particular trader. Of
the 65 option market makers who were trading in September 2005, 25, or 38.64%, traded
only in the option market and did not use the futures market to hedge. Of these traders, their
overall trading activity is very low, capturing only 3.8% of the overall number of trades
conducted that month. Forty, or 61.54%, of the options market makers engage in trading in
both markets. The number of trades in the futures market far surpasses the number of trades
in the options market with 3020 and 4452 trades in options and futures respectively.
Thus, hedging options trades in the futures market is not a one-for-one strategy.
Trading in options captures 38.88% of the overall number of trades for the month, whereas
trades in the futures market are at about 57.32%. Higher amounts of trading in the futures
market may be one explanation for why the levels of position delta are higher when both
option and futures trades are evaluated than when only the options trades are evaluated.
Market makers who are using both markets to hedge may be overestimating their exposure to
21
price risk, resulting in holding (or selling) too many of the underlying futures contracts to
offset their positions in the options market.
There are other explanations as well however: first, option market makers trade
frequently and in small amounts as is documented in Table 1; therefore, many of their option
trades are liquidated quickly which would eliminate the need for hedging in the futures
market. Further, if they are unable to unwind these positions efficiently and quickly, it may
be an impetus to eventually go into the futures market to offset their inventory holding risk
exposure. The desire to keep bid-ask spreads to a minimum as postulated in Huh et. al.
(2012) in order to keep liquidity high may also be a reason for the above findings.
The ability of market makers to liquidate their inventory holdings quickly and
efficiently using futures markets is examined by widening the interval by which options are
matched with the futures trades to better capture the hedging activity being employed by the
option market makers. Two primary filters are used: (1) 600 seconds and (2) the trading
increment which consists of one hour of trading, except the initial increment that consists of
the first 2 hours of trading.5 The results from this analysis are presented in Table 6, with
Panel A presenting the merge base of 600 seconds and Panel B containing the results from
the increment merge base. From this analysis it appears that market makers actively seek to
utilize the underlying futures market to hedge their nearby contracts more than the further to
expiration contracts. Further when comparing the alternative merge base specifications of
600 seconds and an increment with the instantaneous hedging (within one minute), it appears
that lengthening the time frame captures greater amounts of market maker hedging activities.
This analysis supports the notion that while market makers in options markets do not
maintain instantaneous delta neutrality, they do utilize the underlying futures market to hedge
5
Due to the lack of trading in the first hour, the first 2 hours of trade are combined for the incremental analysis.
22
inventory that either was not easily liquidated quickly or is being held onto for a period of
time to possibly allow for prices to adjust to some acceptable level.
We also examine whether there are differences between large and small traders in
their risk management. We split the sample into two categories of traders: large traders, or
those who maintain an absolute value of quantity traded during an increment of 30 contracts
or more and small traders, or those who fall below 30 contracts in any given increment. The
results of this analysis are presented in Table 7, which indicates that large traders are the
primary users of the futures market for hedging purposes. Small traders do not maintain delta
neutrality as shown by the larger levels of position risk when including the futures trades
than when not. It may be that larger traders are more adept at managing their inventory
holding risk due to their frequent needs to dispel this risk because of their much larger
holdings on average. The levels for the position risk holdings are significantly lower when
including the futures trades corresponding to the options transactions made within the same
increment across both maturity spectrums (nearby and first deferred). Small traders on the
other hand are primarily trading frequent, small amounts in the options market with limited
trading taking place in the futures market, thus eliminating the need for futures market
hedging.
C. Intraday Analysis of Gamma and Vega
Of further importance is whether and how the market maker’s other position risks are
changing over the course of the trading day. Examining the intraday values of gamma and
vega allows for a decomposition of the characteristics that option market makers manage in
23
order to mitigate their exposure to various sources of risk. By evaluating the distribution of
the risk characteristics over the trading day, it can be determined whether any intraday
patterns in risk management exist for market makers in the options market.
Intraday market-maker gamma and vega risk is evaluated over five time increments in
Table 8. Panel A denotes the position risk parameters averaged over all traders. Both gamma
and vega exhibit a u-shaped pattern across the increments indicating increased levels of risk
on average at the beginning and end of the trading day when volume is also the highest.
Significant differences for position gamma correspond to the higher volumes at the beginning
and end of the trading day, while position vega has a significant drop at midday when
volumes are typically at their lowest levels. Thus, on average the risk management practices
of market makers seem to follow typical trends in volume.
Panel B (C) of Table 8 provides the position gamma and vega by increment for large
(small) traders. For large traders both gamma and vega experience relatively flat levels over
the course of the trading day with a significant increase in the last increment. Interesting
differences can be noted between the contract expirations where the first deferred levels of
position vega are of three orders of magnitude larger than the nearby contract. It appears that
large option market makers focus their volatility risk management on the nearby contract.
For small traders, position gamma decreases with contract expiration while position vega is
two orders of magnitude larger in the first deferred contract.
D. The Microstructure of Risk Management
In Table 9 we partition the increment position risk parameters based on number of
trades, trade size, and volume to determine whether any of these market microstructure
24
characteristics play a role in the management of intraday risk by option market makers.
Across all of the sample partitions we see a very large amount of dispersion among the
quartiles especially between the lower 50% and upper 50%, which again suggests that
significant differences in trading and risk characteristics are present among large versus small
traders. Those with the highest levels of trading activity, with the largest trades, and those
who generate the highest amounts of volume, all exhibit the greatest amounts of risk
management: delta risk appears to exhibit an inverted U-shape across the trading increments;
vega risk is highest in the first two increments and virtually flat across the latter three time
increments; gamma risk seems to fluctuate within some acceptable range. No discernible
pattern exists for the lowest quintile for vega risk, but delta risk for this group also exhibits
an inverted u-shape across the trading day.
E. Does Moneyness Matter?
Due to the various costs in trading different types of options market makers may tend
to trade in particular categories of moneyness. This issue is evaluated with the results
presented in Table 10. It is reasonable to assume that certain traders may choose to specialize
in a group of options determined by moneyness due to the relative differences in cost and
structure of the various option types. For instance, deep in-the-money options are almost
perfect substitutes for the underlying security, while in-the-money options are the cheapest. If
option market makers have certain trading strategies based on the differences between
moneyness categories, patterns in trading certain options should emerge.
25
A moneyness level of 3% is chosen because it offers the greatest range of
observations in each moneyness group. A frequency analysis is performed to determine
whether market makers specialize in moneyness groups. This analysis reveals that the vast
majority of options market makers trade in all types of moneyness with only 11 of the 65
trading in only one or two categories of option moneyness. For those 11 traders, all but two
conduct only one trade during the sample period. Of the remaining 54 traders who participate
in trading across all levels of moneyness, 66.17% of their trades are in out-of-the-money
options, 18.87% are in at-the-money options, and 14.12% of their trades are in in-the-money
options as shown in Table 10. Thus, it does not appear that market makers focus on only one
category of moneyness, but instead trade across moneyness groups, with the majority of their
trading focused on out-of-the-money options, probably due to their cost-relative to at-the-
money or in-the-money options.
F. Risk and Return
Previous research has documented the impact of the position greeks on bid-ask spread
but since we are interested here in market making as a trading strategy, we evaluate the
impact of each of the position risk parameters on profit. If the risk/return model holds true
we expect to see significant levels of risk being related to profit. The univariate analysis
above indicates that option market makers actively seek to manage their exposure to delta
risk over both the nearby and first deferred contracts, albeit not instantaneously, focus their
vega hedging on primarily the nearby contract, and are subject to significant levels of gamma
risk. A simple regression of daily profit for each trader on the position risk parameters
(Delta, Vega, and Gamma respectively) yielded the following results for the nearby contract
26
where Position Gamma was the only risk parameter found to have a significant effect on
market maker profit:
Pi ,t =4.31−14.05 δ i ,t −6.14 ϑ i , t +81.73 γ i ,t + ε i ,t
These results are in line with those of Jameson and Wilhelm (1992) who found Gamma to have a
positive and significant effect on the spread and also correspond to the theoretical design
constructed in Huh et. al. (2012) with respect to the ability of a market maker to rebalance
increasing the costs associated with trading.
V. CONCLUSION
The institutional characteristics of traders behind four different trade classifications
are evaluated for the futures option NYMEX natural-gas market in order to decompose trade-
type characterization. It is found that traders conducting member proprietary trading in the
natural-gas option market behave as though they are market makers, on average, trading
often in small amounts with very little time in between trades, and are responsible for the
highest levels of activity in terms of volume. They also end the trading day with very low
levels of inventory in order to mitigate their exposure to overnight inventory-holding risk.
Evaluation of the extent of competitive forces in each trade category and the use of
interdealer trades to expel unwanted inventory are also conducted in order to provide more
information on the institutional details of option market making. It is shown that traders who
conduct member proprietary trading are one of the largest trader groups and engage in
significant amounts of interdealer trading in order to maintain their preferred inventory
levels.
27
The portfolios of option market makers are examined in terms of their exposure to
daily levels of risk as measured by delta, gamma, and vega. It is found that end-of day
positions are very small, a result that supports the hypothesis that market makers try to
mitigate their exposure to overnight risk. Intraday, position delta and vega are found to be
relatively constant. Position vega has a significant drop at midday (increment 3) but has
insignificant changes and small levels throughout most of the trading day. Gamma has
significant changes between increments 1 and 2 and then again at the end of the trading day
between increment’s 4 and 5, which likely results from higher volumes at the beginning and
end of the trading day. These results lend support to the hypothesis that market makers in
options markets work to maintain their exposure to both price and volatility risk, and are
primarily exposed to the effects of rebalancing risk. Analysis of the relationship between the
position risk parameters and profits shows a significant and positive relationship between
profitability and position gamma risk exposure.
When comparisons are made between large and small traders it is found that large
traders utilize the underlying futures market to hedge price risk, but only at longer time
horizons. One explanation for this is that the underlying futures market is used by option
market makers wanting to dispel their inventory holding risk that cannot be eliminated in the
option market; indicating a preference for managing risk using options. The exposure of large
traders to rebalancing and volatility risk is significantly higher than that of smaller traders, as
larger traders inventories are more cumbersome to manage throughout the trading day.
This article provides an in-depth, descriptive analysis of how market makers in option
markets make their market and lays the foundation for a wealth of future research paths.
Future research directly stemming from this analysis should evaluate how changes in risk
28
holdings affect the prices that market makers maintain. Patterns in bid-ask spreads are well
documented; thus, the intraday changes in risk holdings and the movement of traders into and
out of the market may serve as additional measures to help explain their U-shaped patterns.
Other issues that deserve further examination include how option market makers are using
the option market to mitigate their exposure to price risk, the impact of a market event on the
number and ability of traders providing market-making services, as well as the extent to
which interdealer trading impacts risk levels and, ultimately, market prices. These are largely
unaddressed areas in the literature and warrant further investigation. This paper serves as the
basis for a fruitful stream of future research surrounding market making in option markets.
References
Agarwal, Vikas, and Naik, Narayan Y. (2004). Risks and Portfolio Decisions Involving Hedge
Funds. The Review of Financial Studies, Vol. 17, No. 1, 63-98.
Azzarello, S. (2014). Energy price spread: Natural Gas vs. Crude Oil in the US. CME Group
Market Insights Publication.
Barberis, N. and Huang, M. (2008). Stocks as lotteries: The implications of probability weighting
for security prices, American Economic Review, Vol. 98, pp. 2066-2100.
29
Baschmeir, L. and Grifﬁn, J. (2006). Testing for market integration: crude oil, coal and natural
gas. The Energy Journal, Vol. 27, 2.
Benston, G., and Hagerman, R. (1974). Determinants of bid-ask spreads in the over the counter
market. Journal of Financial Economics, 1(4), 353.
Boyer, B., Mitton, T., and Vorkink, K. (2010). Expected idiosyncratic skewness, Review
of Financial Studies, Vol. 23, pp. 169-202.
Buyuksahin, B. and Robe, M. (2014). Speculators, Commodities, and Cross-Market Linkages.
Journal of International Money and Finance, Vol. 42, pp. 38-70.
Çetin, U., Jarrow, R., Protter, P., and Warachka, M. (2006). Pricing Options in an Extended
Black Scholes Economy with Illiquidity: Theory and Empirical Evidence. The Review of
Financial Studies, Vol. 19, No. 2, 493-529.
Cox, J. C., and Rubinstein, M. (1985). Options markets. Upper Saddle River, NJ: Prentice Hall.
Cox, J. C., Ross, S. A., and Rubinstein, M. (1979). Option pricing: A simplified
approach. Journal of Financial Economics, 7(3), 229.
Figlewski, S. (1989). Options arbitrage in imperfect markets. Journal of Finance, 44, 1289.
Giannetti, A., Zhong, R., and Wu, L. (2004). Inventory hedging and option market making.
International Journal of Theoretical and Applied Finance, 7(7), 853.
Ho, T. S. Y., and Stoll, H. R (1983.) The dynamics of dealer markets under competition. Journal
of Finance, 38, 1053.
Huh, S., Lin, H. and Mello, A. (2012). Hedging by option market makers: Theory and evidence.
Working paper.
Hull, J. (2000). Options, futures and other derivative securities. Upper Saddle River, NJ:
Prentice-Hall.
30
Jameson, M., and Wilhelm, W. (1992). Market making in the options markets and the costs of
discrete hedge rebalancing. The Journal of Finance, 47(2), 765.
Klock, M., and McCormick, D. T. (1999). The impact of market maker competition on
NASDAQ spreads. Financial Review, 34, 55.
Kuserk, G., and Locke, P. (1993). Scalper behavior in futures markets: An empirical analysis.
Journal of Futures Markets, 13, 409.
Kuserk, G., & Locke, P. (1994). Market maker competition on futures exchanges. Journal of
Derivatives, Summer, 56.
Locke, P.R., and Steven C. Mann (2005). Professional Trader Discipline and Trade Disposition.
Journal of Financial Economics, 76, 2005.
Locke, P. R., and Sarajoti, P. (2004). Interdealer trading in futures markets. Journal of Futures
Markets, 24, 923.
Manaster, S., and Mann, S. C. (1996). Life in the pits: Competitive market making and inventory
control. Review of Financial Studies, 9, 953.
O'Hara, M., and Oldfield, George S. (1986). The Microeconomics of Market Making. The
Journal of Financial and Quantitative Analysis, Vol. 21, No. 4, 361-376.
Onur, I. (2009). Natural gas markets: How sensitive are they to crude oil price changes? OPEC
Energy Review.
Serletis, A. and Ricardo, R., 2004. Testing for common features in North American Energy
Markets. Energy Economics, Vol. 26, pp. 401–414.
Serletis, A. and Shahmoradi, A., 2006. Returns and volatility in the NYMEX Henry Hub natural
gas futures market. OPEC Review: Organization of the Petroleum Exporting Countries,
Vol. 30, pp. 171–189.
31
Silber, W. (1984). Marketmaking behavior in an auction market: An analysis of scalpers in
futures markets. Journal of Finance, 39, 937.
Stoll, H. R. (1978). The pricing of security dealer services: An empirical study of NASDAQ
stocks. Journal of Finance, 33, 1153.
Tinic, S., and West, R. (1972). Competition and the pricing of dealer service in the over-the-
counter stock market. The Journal of Financial and Quantitative Analysis, 7(3), 1707.
Wahal, S. (1997). Entry, exit, market makers, and the bid-ask spread. Review of Financial
Studies, 10, 871.
Working, Holbrook (1977). Price Effects of Scalping and Day Trading, Selected Writings of
Holbrook Working, Chicago Board of Trade.
Working, Holbrook (1967). Tests of a Theory Concerning Floor Trading on Commodity
Exchanges, Food Research Institute Studies: Supplement.
32
Table 1: Summary Statistics for NYMEX Natural-Gas Options Trading
Table 1 displays summary statistics for the most active traders for all trade categories over the first three nearest contract months for
options. The level of analysis used to conduct the testing of whether member proprietary trader behavior is indicative of that of market
makers in futures options is meant to provide an indication of how an average trader conducting a certain type of trade behaves and
the characteristics of each type of trade. The total number of trades each day is determined through a frequency analysis that provides
a count of the number of trades every day by each trade group across the three nearest contract expirations. The daily average number
of trades is found by taking the average of the total number of daily trades obtained from the frequency analysis (the total number of
trades divided by the number of trader days). The daily average volume is found by first summing the total quantity of purchases
traded in a day (buy observations only) by an individual trader for a trade type and contract expiration. This provides the total sum of
quantity traded for each trader on every day for a trade type and contract expiration. This total is averaged over the total trader days by
trade category and contract expiration to obtain a daily average level of trading volume. The average trade size is found by evaluating
the average quantity traded for each trade category and expiration. The average time between trades is found by evaluating the average
time between each trade for each trade category and expiration.
Summary Statistics for NYMEX Natural-Gas Options

Daily average Total number Daily average Total Average Average time
CTI number of trades of trades volume volume trade size between trades
3
3
Nearby contract
1 89 36,646 167 2,154,761 29.43 18.42
2 5 1,511 263 226,920 66.13 15.06
3 4 1,499 145 176,675 55.23 15.07
4 55 22,617 259 1,928,015 43.39 15.06
First deferred contract
1 47 19,179 121 1,232,891 32.13 15.93
2 4 864 324 189,577 95.27 12.42
3 2 501 133 63,144 60.52 17.31
4 33 33 225 1,281,987 48.52 15.92
Second deferred contract
1 26 10,789 99 691,003 31.23 12.76
2 3 470 272 99,440 95.26 14.74
3 2 266 154 43,036 80.16 15.37
4 20 8,010 193 837,020 53.95 14.01
Table 2: Distribution of Proprietary Trader Income
Panel A in Table 2 displays the distribution of income for active, proprietary trading. Daily average income for options (in dollars) for each proprietary trades
across the nearest three expirations is found by marking to market each trade over the course of a trader day, summing the income for each individual trader, and
averaging the income for each trader over all trader days by contract expiration. If the trade is a sell, the income is found by taking the difference between the
trade price and the settlement price and multiplying by the quantity. If the trade is a buy, the income is found by taking the difference between the settlement
price and the trade price and multiplying by the quantity. The quartiles of daily income are found from the total daily income levels for each trader. Thus, the
minimum corresponds to the lowest level of income made by an individual trader during the sample period for contract expiration. Panel B in Table 2 displays
the distribution of daily income where each day a proprietary trader’s income is calculated by marking to market all of their trades at daily settlement prices. An
average across all traders is taken to obtain a daily average income for each day in the sample. This table represents the distribution of the daily average incomes
across 413 days with the top row containing all trades and the next three incomes broken out by expiration.
Panel A: Income Distribution

Mea
Contract N Minimum 25% Median 75% Maximum
n
- $50
All trades 15,573 $228 -$1,799,279 $60 $1,779,416
$162 0
- $80
Nearby 13,503 $245 -$422,450 $50 $445,018
$375 0
- $47
4
3
First Deferred 10,944 $267 -$5,396,627 $25 $5,333,350

$150 0
$25
Second Deferred 7,760 $331 -$122,190 -$40 $0 $697,990
0
Panel B: Daily Income Distribution
Mea
Contract N Minimum 25% Median 75% Maximum
n
- $58
All trades 413 $224 -$8,842 $145 $10,820
$175 5
- $97
Nearby 413 $229 -$20,966 $188 $12,717
$350 0
- $73
First Deferred 413 $254 -$6,864 $113 $16,068
$289 2
- $44
Second Deferred 413 $331 -$12,777 $49 $53,248
$195 4
Table 3: Number of Traders
The table presents the daily average number of traders executing the various types of trades in options market across
our sample. A trader trades a CTI=1 trade when they own 10% or more in the trading account for which they are
trading. CTI=2 executed trades are for the traders clearing member account. CTI=3 trades are executing for other
floor traders who are present on the floor. A trader executes a CTI=4 trade when the principal behind the trade is a
non-member, or a customer. Traders may execute all 4 of the trade types for all contract maturities. There are on
average 51 traders executing trades of any type and any maturity per day.
CTI Average number of traders

Nearby contract
1 31
2 2
3 3
4 18
First deferred contract
1 25
2 2
3 2
4 14
Second deferred contract
1 17
2 1
3 1
4 10
35
Table 4: Interdealer Trading Options
The percentage of trades by customer type in the options market is determined through a frequency analysis of trade
combinations across the nearest three expiration contracts to examine the extent of interdealer trading in the options
market. Interdealer trades are identified when both the initiator of the trade and the opposite trader are both trading
for their personal accounts.
Trader Opposite trader Percentage of trades by customer type

Nearby Contract
Personal Personal 22.83%
Personal House 3.30%
Personal Other floor 3.96%
Personal Customer 64.79%
House House 0.05%
House Other floor 0.17%
House Customer 1.29%
Other floor Other floor 0.03%
Other floor Customer 0.61%
Customer Customer 2.97%
First Deferred Contract
House House 0.07%
Second Deferred Contract
House House 0.07%
36
Table 5: Delta Risk Analysis Merge Base of 60 Seconds
Table 5 provides evidence testing the hypothesis that option market makers maintain instantaneous delta neutral
positions. The trading day is partitioned into five increments. Using the last trade for both options and futures in a
time increment, an implied standard deviation is found for each time increment, which minimizes the sum of
squared errors between the options price estimated by the binomial option pricing model and the observed options
incremental settlement price. This implied standard deviation is then used to compute the delta for all option strikes
and types (puts and calls) in each increment. For each trader, the quantity of trade is summed over the increment and
multiplied by the estimated parameter values to compute the trader’s exposure to portfolio risk as measured by
position delta. The variable options and futures delta include futures trades placed within 60 seconds of the options
trade under the same account number. The absolute value of each trader’s position risk parameter for each trade is
taken and averaged over all traders trading in a given increment. The positions are marked to market each increment
by summing the quantity traded over a particular increment for a trader and using that level as the beginning
inventory level for the next increment.
Delta Risk Analysis: Merge Base of 60 Seconds

Nearby Contract
Incremen Options t- p-
t Delta Options and Futures Delta Difference Value DF Value
1 13.80 15.57 -1.77 -6.69 9224 <.0001
2 -1.68 -5.67 1125 <.0001
18.31 19.99 7
3 -1.98 -5.73 1220 <.0001
21.83 23.81 0
4 -2.49 -6.72 1289 <.0001
23.72 26.21 2
5 -3.21 -7.81 1326 <.0001
26.11 29.32 5
Incremen Options t- p-
t Delta Options and Futures Delta Difference Value DF Value
1 7.67 9.11 -1.43 -5.92 6598 <.0001
2 10.49 12.02 -1.53 -4.77 8595 <.0001
3 13.04 14.44 -1.40 -4.1 9584 <.0001
4 -1.52 -3.93 1033 <.0001
14.53 16.04 3
5 -2.19 -4.96 1078 <.0001
16.04 18.23 7
37
Table 6: Delta Risk Analysis with Alternative Merge Base Specifications
Table 6 explores alternate matching specifications of futures and options trades to explore option market markets
position delta risk management strategies. The trading day is partitioned into five increments. Using the last trade
for both options and futures in a time increment, an implied standard deviation is found for each time increment,
which minimizes the sum of squared errors between the options price estimated by the binomial option pricing
model and the observed options incremental settlement price. This implied standard deviation is then used to
compute the delta for all option strikes and types (puts and calls) in each increment. For each trader, the quantity of
trade is summed over the increment and multiplied by the estimated parameter values to compute the trader’s
exposure to portfolio risk as measured by position delta. The variable options and futures delta include futures
trades placed within either (1) 600 seconds (Panel A) or (2) increment (Panel B) of the executed options trade under
the same account number. The absolute value of each trader’s position risk parameter for each trade is taken and
averaged over all traders trading in a given increment. The positions are marked to market each increment by
summing the quantity traded over a particular increment for a trader and using that level as the beginning inventory
level for the next increment.
Panel A: Delta Risk Analysis: Merge Base of 600 Seconds

Nearby Contract
t- p-
Increment Options Delta Options and Futures Delta Difference DF
Value Value
1 19.68 15.41 4.28 6.1 3738 <.0001
8
3
2 18.75 13.10 5.65 7.57 3069 <.0001

3 20.15 14.84 5.32 6.12 2545 <.0001
4 17.60 14.29 3.31 4.27 2435 <.0001
5 16.43 18.75 -2.32 -2.73 2216 0.0063
t- p-
Value Value
1 11.99 13.48 -1.49 -2.42 2054 0.0157
2 13.54 14.59 -1.06 -1.22 1558 0.2232
3 13.32 14.72 -1.4 -1.62 1175 0.1061
4 13.70 15.71 -2.01 -2.31 1129 0.0208
5 17.16 20.63 -3.46 -2.02 1069 0.0435
Panel B: Delta Risk Analysis: Merge Base of Increment

Nearby Contract
t- p-
Value Value
1 20.61 19.49 1.12 1.72 4517 0.0859
2 19.38 16.92 2.46 3.69 3732 0.0002
3 20.74 18.47 2.28 2.92 3125 0.0035
4 18.16 17.98 0.18 0.24 2981 0.8083
5 17 21.22 -4.22 -4.92 2548 <.0001
t- p-
Value Value
1 12.1 16.19 -4.09 -6.99 2658 <.0001
2 13.17 16.45 -3.28 -4.28 2036 <.0001
3 13.08 16.21 -3.12 -3.95 1570 <.0001
4 14.18 18.35 -4.17 -5.41 1499 <.0001
9
3
Table 7: Delta Risk Analysis By Trader Size with a Merge Base of Increment
Table 7 explores differences between large and small traders with regards to the management of their position delta
risk. A large trader is defined as one whose absolute value of quantity of trade in a given increment is 30 contracts
or more, while a small trader is one who trades below this same threshold. The trading day is partitioned into five
increments. Using the last trade for both options and futures in a time increment, an implied standard deviation is
found for each time increment, which minimizes the sum of squared errors between the options price estimated by
the binomial option pricing model and the observed options incremental settlement price. This implied standard
deviation is then used to compute the delta for all option strikes and types (puts and calls) in each increment. For
each trader, the quantity of trade is summed over the increment and multiplied by the estimated parameter values to
compute the trader’s exposure to portfolio risk as measured by position delta. The variable options and futures delta
include futures trades placed within an increment of the executed options trade under the same account number. The
absolute value of each trader’s position risk parameter for each trade is taken and averaged over all traders trading in
a given increment. The positions are marked to market each increment by summing the quantity traded over a
particular increment for a trader and using that level as the beginning inventory level for the next increment.
Panel A: Large Traders with Merge Base of Increment

Nearby Contract
t- p-
Value Value
1 84.13 54.84 29.28 9.46 782 <.0001
0
4
2 83.53 48.79 34.74 10.29 600 <.0001

3 91.93 55.92 36.01 10.48 510 <.0001
4 82.42 53.7 28.71 7.26 457 <.0001
5 89.61 68.87 20.73 4.68 336 <.0001
t- p-
Value Value
1 71.48 58.57 12.91 3.07 256 0.0024
2 85.87 61.59 24.28 3.94 187 0.0001
3 78.26 56.82 21.44 3.85 155 0.0002
4 81.2 62.87 18.33 4.24 171 <.0001
5 100.97 68.89 32.08 3.86 146 0.0002
Panel B: Small Traders with Merge Base of Increment

Nearby Contract
t- p-
Value Value
1 7.29 12.08 -4.79 -12.7 3734 <.0001
2 7.07 10.8 -3.74 -10.19 3131 <.0001
3 6.83 11.15 -4.31 -7.65 2614 <.0001
4 6.5 11.5 -4.99 -10.63 2523 <.0001
5 5.94 13.96 -8.02 -11.67 2211 <.0001
t- p-
Value Value
1 5.74 11.65 -5.91 -13.09 2401 <.0001
2 5.78 11.86 -6.08 -11.59 1848 <.0001
3 5.9 11.73 -5.83 -9.96 1414 <.0001
4 5.5 12.58 -7.09 -11.3 1327 <.0001
5 5.59 14.64 -9.04 -7.74 1161 <.0001
1
4
Table 8: Intraday Gamma and Vega Risk Position Levels by Increment
Table 8 evaluates the option market maker’s intraday exposure to Position Gamma and Position Vega, or
rebalancing and volatility risk respectively. The trading day is partitioned into five increments. Using the last trade
for both options and futures in a time increment, an implied standard deviation is found for each time increment,
which minimizes the sum of squared errors between the options price estimated by the binomial option pricing
model and the observed options incremental settlement price. This implied standard deviation is then used to
compute the gamma and vega for all option strikes and types (puts and calls) in each increment. For each trader, the
quantity of trade is summed over the increment and multiplied by the estimated parameter values to compute the
trader’s exposure to portfolio risk as measured by position gamma and vega. The absolute value of each trader’s
position risk parameter for each trade is taken and averaged over all traders trading in a given increment. The
positions are marked to market each increment by summing the quantity traded over a particular increment for a
trader and using that level as the beginning inventory level for the next increment. Bolded values indicate a
significant difference from the previous increment’s value.
Panel A: All Position Gamma and Vega

Nearby Contract
Increment Gamma Vega
1 54.95 14.17
2 51.78 14.73
3 45.57 12.99
2
4
4 45.86 13.49
5 64.40 15.02
1 33.03 27.00
2 27.29 25.82
3 27.20 25.87
4 29.36 26.72
5 36.60 31.30
Panel B: Large Traders Position Vega and Gamma
Nearby Contract
1 157.35 37.53
2 150.82 41.90
3 126.75 39.77
4 130.96 42.64
5 242.15 54.60
1 155.33 114.53
2 121.01 124.70
3 103.55 112.92
4 116.89 105.74
5 161.43 139.49
Panel C: Small Traders Position Vega and Gamma
Nearby Contract
1 34.27 9.46
2 32.12 9.34
3 29.51 7.70
4 30.58 8.25
5 37.52 9.03
1 20.08 17.73
2 17.61 15.61
3 18.64 16.11
4 17.93 16.40
5 20.10 17.01
3
4
Table 9: Subsample Analysis of the Intraday Risk Parameter Position Levels Over Five Time Increments
Table 9 evaluates whether the option market maker’s intraday exposure to their portfolio of position risk holdings is influenced by their number of trades (Panel
A), trade size, or volume traded. The trading day is partitioned into five increments. Using the last trade for both options and futures in a time increment, an
implied standard deviation is found for each time increment, which minimizes the sum of squared errors between the options price estimated by the binomial
option pricing model and the observed options incremental settlement price. This implied standard deviation is then used to compute the delta, gamma, and vega
for all option strikes and types (puts and calls) in each increment. For each trader, the quantity of trade is summed over the increment and multiplied by the
estimated parameter values to compute the trader’s exposure to portfolio risk. The absolute value of each trader’s position risk parameter for each trade is taken
and averaged over all traders trading in a given increment. The positions are marked to market each increment by summing the quantity traded over a particular
increment for a trader and using that level as the beginning inventory level for the next increment.
Panel A: Quartiles Based on the Number of Trades

Variable Increment 1 Increment 2 Increment 3
1 2 3 4 1 2 3 4 1 2 3 4
Position Delta 2.22 2.84 17.38 19.94 5.28 3.58 25.30 33.29 4.23 4.58 28.48 34.86
Position Delta without Futures 2.19 2.92 15.40 18.18 5.10 3.42 19.16 23.83 4.26 3.85 20.85 23.91
Position Gamma 2.16 2.79 30.58 36.16 10.30 5.89 39.20 52.04 7.64 4.55 37.65 41.95
Position Vega 1.28 2.17 13.04 16.25 4.27 3.47 19.08 24.91 3.80 3.84 18.27 24.80
4
4
Increment 4 Increment 5
1 2 3 4 1 2 3 4
Position Delta 4.46 4.38 25.77 30.13 2.31 4.98 34.68 28.19
Position Delta without Futures 4.72 3.59 19.28 19.39 2.41 4.45 25.48 20.41
Position Gamma 9.12 6.13 44.80 49.12 4.53 9.51 48.49 53.27
Position Vega 5.62 2.91 18.75 21.04 2.60 3.71 16.64 23.89
Panel B: Quartiles Based on Trade Size
1 2 3 4 1 2 3 4 1 2 3 4
Position Delta 2.91 5.15 13.28 30.82 3.82 7.55 20.98 51.26 4.26 7.60 21.87 53.35
Position Gamma 4.29 4.58 22.55 58.43 7.97 9.12 32.01 79.84 6.21 7.62 23.26 69.88
Position Vega 2.47 2.97 10.72 24.96 3.42 6.28 15.47 37.53 3.47 5.15 13.57 37.67
1 2 3 4 1 2 3 4
Position Delta 4.22 6.83 18.27 45.21 3.03 6.75 17.09 47.25
Position Gamma 6.86 9.48 29.58 76.52 7.18 11.23 29.58 83.90
Position Vega 3.84 4.59 11.61 31.36 3.00 4.18 12.66 34.08
Panel C: Quartiles Based on Volume

1 2 3 4 1 2 3 4 1 2 3 4
Position Delta 1.30 3.47 7.74 27.70 1.60 4.80 11.39 46.69 2.23 6.23 11.01 48.86
Position Gamma 1.73 3.22 12.83 51.76 3.39 7.51 19.58 70.19 3.77 6.18 14.33 59.67
Position Vega 1.50 2.13 5.92 23.78 2.07 4.14 9.15 35.89 2.72 4.48 7.26 32.98
1 2 3 4 1 2 3 4
5
4
Position Delta 3.43 5.23 10.74 42.88 2.90 5.05 9.10 44.03
Position Gamma 4.03 7.73 19.13 70.06 10.54 9.90 19.92 73.72
Position Vega 2.32 3.86 7.33 32.93 3.42 3.55 7.05 32.89
Table 10: Percentage of Trades in Each Moneyness Category
Table 10 presents the percentage of trades in each category of option moneyness for traders who trade in all
categories, where moneyness is defined by a 3% range. In other words, for a range of 3%, an at-the-money (ATM)
option is one whose strike price is within 3% of the price of the futures settlement price, an out-of-the-money
(OTM) option is one whose strike price is above 3% of the futures settlement price, and an in-the-money (ITM)
option is one whose strike price is below 3% of the futures settlement price.
Category Percentage of trades
OTM 66.17%
ATM 18.87%
ITM 14.12%
6
4

Market Makingand Risk Managementin Options Markets

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Market Makingand Risk Managementin Options Markets

Uploaded by

Copyright:

Available Formats

Market Making and Risk Management in Options Markets

for all traders.

account, another member’s own account, or a non-member account (customer). A member’s

income can be derived from proprietary trading and brokerage.

It is shown that member proprietary trading in futures-options markets serves the

profitable trading strategy for these traders.

In particular, we focus on the risks taken by proprietary traders, in terms of

level or changing levels of risk throughout the trading day on average.

by (for example) purchasing a quantity of futures options while simultaneously executing a

accurately represent the “inventory” that the market maker is carrying.

on the profitability of an option market maker.

The remainder of this article is organized as follows: Section II provides information

option market; Section V concludes.

from 144 to 91.

buyer is responsible for movement from the Hub.

III. MARKET MAKER TRADING STRATEGIES

futures-options markets, it is an empirical question as to whether the characteristics of market

makers in futures markets universally hold in the option market.

A. Market Making Activity

The positions and levels of market-making activity are determined through an

amounts, with high levels of overall volume.

B. Market Making Revenue

periods. For example, the relationship between commodity-equity cross-market linkages

gas. Income stemming from brokerage functions would be influenced by cross-commodity

linkages, not necessarily income from proprietary trading.

we have two views of income: among traders and across traders.

possibility of big upside potential of these highly skewed assets.

[Insert Table 2 about here]

C. Market Making Competition

[Insert Table 3 about here]

In inventory microstructure models market makers face exogenous demands to buy

existence of a significant percentage of interdealer trades is inventory control. These traders

the nearby contract.

[Insert Table 4 about here]

E. Summary of Market Making Activities

The institutional characteristics of futures-options markets indicate that on average,

of profitability, market making in options futures markets is a profitable trading strategy.

practices of market makers in futures-option markets, the risk structure of member

proprietary traders follows.

IV. RISK MANAGEMENT

depends on the stochastic nature of the underlier’s returns.

continuously, an option transaction’s contribution to the dealer’s risk exposure will be

and the underlying asset’s return volatility.

examined by Giannetti, Zhong, and Wu (2004) when they developed an inventory-based

Of vital importance to studying market maker behavior in option markets is

of the dividends expected during the life of the option.

their portfolios when the underlier fluctuates.

stock price fluctuates.

an indication of the inventory-risk exposure these traders faced.

A. Position Risk Parameters

trades in the option market.

Volume-weighted average prices for observations from proprietary trades (CTI 1

the option premiums.

minimal time value associated with in-the-money futures options. It is advantageous to

premium to the observed premium for each hypothetical sigma.

risk in the sections that follow.

makers may hedge inventory risk exposures by maintaining delta-neutral positions

engage in hedging by (for example) purchasing a quantity of futures options while

simultaneously executing a certain number of contracts in the underlying futures contract,

option market makers do not maintain instantaneous delta neutrality.

[Insert Table 5 about here]

Due to this finding, an additional analysis is performed to determine whether option

market makers tend to specialize in hedging in a particular market. To evaluate whether