Testing The Engulfing
Bullish Pattern
Identifying chart patterns is often subjective.Is there a way
to objectively backtest chart patterns? It can be a challenge
but here's one way to doit.
by Pawel Kosinski
necessary componentofbacktesting trading strate-
pies isto have well-defined entries and exits. For
Tapanese candlesticks like the engulfing bullish
pattern, exits aren't really specified—traders de-
fine them in different ways, As an example, you
could look for a resistance well above a potential
entry price or sell the position ifa bearish price action occurs.
This is what makes backtesting a challenge.
On the other hand, traders would still like to backtest pat
terns suchas the engulfing bullish pattern to find out how well,
the pattern works, In this article I will show you a relatively
easy strategy to analyze this pattern
BACKTESTING CRITERIA
For my backtesting I used the following criteria (ee also
Figure 1)
+ OnDay0 (enclose), the engulfing bullish pattern cceurs.
This day is shown as B in Figure 1
ber 2018+ Teta Ants of STOCKS & COMMODITIES
* Price has decreased recently and this is detected by
demanding that both the five-day SMA and the 10-day
‘SMA are falling on Day 0 and on Days 1 and 2. This
‘way, you make sure the pattern occurs in a downtrend,
+ The general trend is, however, up, which I defined as the
100-day SMA above the 400-day SMA.
+ The entire stock market (represented by the S&P 500
index) is also bullish: The 100-day SMA of the index
is above the 400-day SMA.
+ On Day 0, high volume was detected—higher than the
average of the past 14 days.
+ Tirade stocks from the Russell 3000 inex. The advan-
tage of this is you don't necessarily focus on blue chips
or just good stocks. I also select stacks in which price
is in the $20-$90 per share range and volume on Day
iis higher than 100,000 shares.
I search for a local maximum in price that occurred some
days before the engulfing bullish pattern, Acbitrarily, I state
that this is the day where the high exceeds the surrounding
three days, In Figure 1, this is indicated by A. Nua, i8 the
umber of days when this happened (five days in the example
shown in Figure 1) and P,. is the high of this day In addition
Tassumed that N,, cannot exceed 30 days,ENTRY & EXIT
For the entry, I use three possible sce-
1. Buy the stock on the next day's open,
that is,on days.
2. Buy the stock on the open of day+2 if
the close of day+1 is above the high of
Day0 (thatis,the engulfing parternhas
‘been confirmed),
3. Buy the stock on the open of day+2 if
the close of day+1 is above the high of
Day 0 (butnotif the high of day+1 was
higher than the high of Day 0).
In Figure 1 you see an example of sce-
nario 2,
Regarding the exits, [will sell the stocksif
Profit target
138 @ 77,77
4
B ar’ ii
Go Lo
138 @ 72,72
price reaches P,.,.1place a stop-loss $2 be-
ow the high of Day0.Laterl will investigate
Another case since this requirement may be
somewhat artificial and many practitioners
‘may disagree with me.
Thisstrategy isnot perfect. Realistically,youshould consider
some other aspects, such as current price testing a support (or
evenaconfluence of supports) orthere may anotherresistance
Delon Pigs and 4 on. Thescfine, hi
a higher number of trades than what you can expect in real
life unless someone wishes to use fully automated trading
and blindly follow: the signals as in backtesting
For each trade you buy a number of shares so the total
‘price doesn't exceed $10,000. Alternatively (not done in this
research) you could test a strategy in which the total risk is
lower than some arbitrary valne.
jeseascls will scl it
RESULTS:
Iwilluse selected statistical results that are usedin NinjaTrader
software: () the total net profit in USD; (i) profit factor; (ii)
‘maxiowm drawdown; (iv) aumber of trades; (v) percent of
profitable trades; (vi) average trade in percent; (vi) winvToss
ratio; (vii) mamber of days a stock is held. I discussed these
parameters alsoin my September 2018 STOCKS & CoMMODI-
TIES article, “Double Bottoms Revisited.”
‘The backtesting results are shown in Figure 2. Tcompare
the three aforementioned scenarios. Note that scenario 2
results in the best performance. Actually, scenario 3 has a
slightly lower (that is, better) maximum drawdown but the
other parameters are rather worse.
‘Therefore, in the following section, I will pay attention to
Modification of scenario 2
First, make sure the take-profit level is high enough above a
potential entry price, so the rewardiiskratio would be decent.
We will doit by dividing potential gain defined by the price
Pag) by a potential loss ($2 below the high of Day 0, as men-
FIGURE 1: A SUCCESSFUL TRADE, Youbuy stock one dy fer theengulrg ulin pati occured
(town as 8). Thre sa lcal mairum afew days before (shown a A). Tis eel maxinum occured
‘Hp, 8878 999 and ts high hs day wllbe cad P, Tho profi target vei equlb
eT ET mT
ot net rf simist7_| $6507 12209)
Protec 132 1.58 129
Maxum arendoan —|—362% Prd 201%
{Muenber oft 2186 28 re
Poftblatades | 48706 wm sre
average ade] 087% 150% 75%
Waves 158 i 12
[Number of days ha 16 ‘ 8
FIGURE 2: BACKTESTING RESULTS 01/10/2000-12/31/2017. Scenario 2
‘uperorms, which means you sou wal or cantmaton atthe enguting
bulls pater cocured.
siondpeevicnay Laat
this rowardfiskratioto [radi pe om
beat leat 10, oti i
The results of this [Maximum drawdown 316%
‘modification are dis- [Munberottrades EA
Piyed in Figo 2. Tac eaters —| Sua
‘ins rato is clearly EB a
igher, aa expected. (Wueetyamcrag—-—
Unfortunately, the per~ IGOR S: MCREASWa THE POTENTIAL
‘REWARD /ISK RATIO. Hor, he patria
reardiskralshighor an 1.0, Theresuts
“aracomparedtoseenaro anayzeainFgue
2 (second coum). K wil be sight higher
centage of profitable
trades has decreased
accordingly. In my Sep-
tember2018S&Carticle, —(aimost 25%) The maximum dawdom isa
TusedtheKellycriterion _btworseferthesecondcaseandithe number
tw help me to decide in "ades has reduced
such cases:
K=P-(1-PyR
here P is the percent of profitable trades and R is the aver-
age win/loss ratio,
So, for my benchmark, the ease shown in Figure 2 in the
secondcolumn, Kwillbecome 24.5% Forthe second case Fi
Decemher 201 + Tecnica Anabirof STOCKS & COMMODITIES 15,er ee Em
Total ot sia Notatret ot CC) Total sett
Preto 12 Prttckr 12 195. Prot 182
axarun awiown—| 4.00% Mexinun rancoun | — 3.29% 220% Naini cewdenn—| 2.77%
Number vases 2200 Nanberfvades 2 o Number ol aces ma
Proteins] | — $5.02 Probie tades(@] | tare | e008 Pratl tedes a] | ave
| Average trade [%) 0.66% [Average trade [%) 027% 187% Average trade [%) 1
Wins 108 Wavess 1 13h Wines 128
Nunber oad 2 Number fay 8 z Nair oda Tad 6
FIGURE 4: EARLY EXIT. You et the wade
‘25 Sn a5 pice touches th hgh of Day 0
us he height ofthe ere Day O bar. You
hol the stock for fener dys. The resuts
‘are decent,
ure 3), K will be slightly higher (almost 26%). The maximum.
drawdown isa bit worse for the second case and the number
of trades has reduced so it depends on what you prefer:
Modification of scenario 1
Nest, I will return to seenario 1, where I enter on the first
day after the pattern occurred. In other words I don't wait
for any confirmation as studied in the first column from the
table in Figure 2.
This time I want to leave the trade auvch eatliee. I deine
a rather artificial take profit level that is the high of Day 0
plus the length of the entire candle of Day 0 (hati, high +
Ghigl- lon) uf Dey 0). The scsulls van be soon in Figuie 4.1
observed, perhaps as expected, there was a higher percentage
of profitable trades but lower gain from each trade, as well
as lower wingloss ratio. An advantage of this strategy is that
the number of days in a trade is less. This gives you am op-
portunity to use your capital for something else since your
capital isn't tied up.
In addition, I would do another modification, inspized by
the May 2018 S&C article by Ken Calhoun, “Moving Average
Hammer Pivots,” who, in the article, suggests combining. a
Japanese candlestick with some major moving average. In the
article, Calhoun uses the hammer candlestick and 200-day
SMA. Although I am studying the engulfing bullish pattern
(ota hammer), the idea is similar
I wanted to modify scenario I with a case where the lows
of Day 0 or Day 1 are below the 200-day SMA, but the eloses
are above it (in my computer program, I allow a few cents
deviation to inerease the number of trades). This mimics the
touching of this major moving average.
“The backtesting results for scenarios 1 and 2 are shown in
The backtesting confirms
the engulfing bullish pattern
is useful and can be applied
‘successfully in trading.
16 + Dacre 2018 * Techical Anas of STOCKS & COMMODITIES
FIGURE 5: WHAT IF THE ENGULFING BULLISH PATTER
‘TOUCHES THE 200-DAY SMA? The routs are deoent but
‘he numberof rade ies.
FIGURE 6: MODIFYING THE STOP-LOSS.
Herayuitodice aso spo, thls
‘os the postion manual stock closes
8030 below the ow of ay 0. The $2 hard
Sopossis ezine
Figure 5. It turns out that the results seem to be quite good—
note the high profit factor (especially for scenario 2) and low
‘maximum drawdowa, On the otherhand,the number of trades
is rather low—only 50 trades for scenario 2 (Compared to
428 trades where the pattern didu't touch the 200-day SMA.
‘Therefore, the statistical results may be questionable as there
‘may be too few trades to draw conclusions. Finally, this set-up
‘may seldom occur in real trading
‘MOVING THE STOP-Loss
So far I've shown results for scenarios
where the stop-Toss level was dofined by
the ugh uf Day Osun $2. Peay, thin
wasn’t the best choice so Tl add some
‘modifications —keep this stop-loss level
as itwas but also decide to leave the trade
if a daily close is $0.10 below the low of
Day 0. If this happens, just assume the
engulfing bullish pattern isn't valid and itis better to “es-
cape" before you lose more. In other words, don't wait for
the “hard” stop-loss
Figure 6 shows the results ofthis modification. Notice the
percentage of the profitable trades is lower, that is, we lose
some profitable trades. That the stock closed below the pat-
tem itn't“the end of the world” according tothe backtesting
results. On the other hand, the win/loss ratio, as expected. i
larger for the case when I used the “soft” stop-loss. Another
upside is the reduetion in maximum drawdown and sumber
cof days in the trade.
AND THE VERDICT ...
‘The backtesting confirms the engulfing bullish pattern is use-
ful and can be applied successfully in trading. The pattern
‘occurs often so you should have no problem spotting it. It
also Teads to quick profits and can be improved if combined
with other common strategies such as using trendlines and
pice action,
Pawel Kosinski, PhD, MEng, is a professor in process
technology at University of Bergen in Norway. His research
Continued on page 21Hispreferred vehicle is the so-called*Road Trip Trade,”
which I wrote about in my February 2017 S&C article,
“A Road Trip With Options Supertraders”
+ Trade short weekly condors with care or avoid
together, ATM rapid decay is also a reason to not do
a common and popular trade, Mark Sebastian says, “Tt
is also a reason nor to be short weekly condors. While
the Fikelihood of being hit is low, the need of the trader
to stay in the trade too close to or through expiration
makes the risk/reward very suspect.”
*+ Consider butterflies that are closer to expiration
versus farther. Mark Sebastian says: “Butterflios may
perform better closer to expiration instead of further
‘out. This could explain the success of weekly burterfes
‘beyond that ofthe simple pin (meaning cx
shoct strike middle). Thebuttrfy holderislongthe cheap
option that has a slow demise and shorttwo options that
are quickly losing their value when at the-money.It is
taking advantage of theta convergence.”
+ Take profits when targets are hit. Ifa short option,
Jong-theta trader achieves a profit, especially early ina
position, he should consider taking instead of holding
out for more. That “more” may turn out to be “less.”
‘especially if volatility moves against him. This is the
‘eternal ebb and flow of the market. Bulls can make
money, boars oan malso money, but you know what they
say about pigs.
+ And, finally, lower your expectations vis-i-vis that
highly optimisticrisk curve.Ttis picture, andapicture,
‘by definition, static. Itshows yourpositionat one instant
in time—this instant (see the sidebar “The Calendar
Spread: An Options Trader Staple” and sidebar Figure
1 for an explanation and example of a calendar spread
risk curve). In another second, minute, hour, or day, it
‘can and does change, and may or may not conform to
rules (such as ATMs decay faster or OTMs decay faster
and so on). As changes occur in market sentiment and
in supply & demand, and as the changes are expressed
KOSINSKI/ENGULFING BULLISH PATTERN
Continued from page 16
Interests Involve mathematical modeling of various physical
‘Phenomena, and he uses this experience for researching the
‘financial markets. He was the principal founder of the site
Tookintotrade.com, which offersbacktesting of various strate-
‘gies. He may be reached at pawel Kosinski@uib.no.
Buddha have been a reat options trader? Maybe yes, maybe
xno, but Iam thinking as we close this two-part article series
that his equanimity in the face of so much change would have
been unparalleled. I’s an essential asset that you surely need
alot of when trading options. Why? Because all the variables
and other things described here and last month in part 1 are
always affecting your options—they keep changing, moment
bby moment, all the time, right up to and through expiration
‘This means that what your risk curve told you earlier you'd
bbe getting may not be the case at all.
John A. Sarkett has written for Stocks & CoMMoDITIES
since 1995. He is also the author of Option Wizards: Real
Life Suwvess Stunies Pron The Ti
ket Mentors, volumes found on Amazon.com and at htp:i/
ception-wizard.com.
FURTHER READING
Sarkett, John A. (2017). “A Road Tri
FURTHER READING
Calhoun, Ken (2018). "Moving Average Hammer Pivots.”
Kosinski, Pawel [2018]. “Double Bottoms Revisited,” Tech-
NinjaTrader
480eFaitorial Resource Index
in the options greeks themselves, especially vega, your
risk curve will change. Sometimes it will change rather
dramatically; sometimes forbetter often for worse. There
are always zeasons for the change in the curve, whether
‘oF not they can be easily discerned.
\G) Apart to CHANGE
Buddha said something like, “Changs isthe
stufflifeis made out of” So,t00, with options.
(Change in profit potential ix ax mach m part
of the game as bid & ask, implied volatility,
nd yes, theta itself
Hewasforsomethe ultimate gura,bot would
ial Mathets, und Mas
With Options Super-
traders,” Technical Analysis of STOCKS & COMMODITIES,
‘Volume 35: February.
(2018}. “The Options Risk Curve, Part 1)" Techni
cal Analysis of STOCKS & COMMODITIES, Volume 36:
November.
qn)
Technical Analysis of STOCKS & COMMODITIES, Volume
36: May.
nical Analysis of STOCKS & COMMODITIES, Volume 36:
September,
eqnm=mm=)>
Deoeber201 + Technical Ana of STOCKS & COMMODITIES + 21