You are on page 1of 37

HOW OMEGA POINT HELPS INVESTORS OPTIMIZE THEIR

PORTFOLIOS WITH APACHE® SPARK™ ON DATABRICKS

© Omega Point ! ompnt.com 1


DATABRICKS

Why Databricks Product


Created Apache Spark to enable big Bring Spark to the enterprise: The
data use cases with a single engine just-in-time Data Platform

Contributes 75% of Spark’s code - Fully managed advanced analytics


10x more than others platform powered by Apache Spark

Trained 20K Spark users

© Omega Point ! ompnt.com 2


SPEAKERS

Omer Cedar
Omer Cedar Eran Cedar
Eran Cedar
CEO - Omega Point CTO - Omega Point

10+ years of algorithmic portfolio Product development in Silicon Valley for


management experience on Wall Street 12+ years

Built systematic quantitative trading Built Oracle’s Social Cloud after Oracle’s
strategies for Two Sigma Investments acquisition of Involver

Early user of Apache Spark for applying Developed systems that managed
data science in financial markets content from millions of online social
properties

© Omega Point ! ompnt.com 3


AGENDA

1.  The data challenge in portfolio management

2.  Omega Point’s vision and technology needs

3.  The Omega Point architecture on Databricks

4.  Lessons learned

© Omega Point ! ompnt.com 4


THE DATA CHALLENGE IN PORTFOLIO
MANAGEMENT

© Omega Point ! ompnt.com 5


A NEW INVESTMENT REALITY

© Omega Point ! ompnt.com 6


IDENTIFYING OPPORTUNITY

Fundamental active investors


represents 90% of the US market
& manage over $1.5 Trillion

SAME STRATEGY
Invests primarily based on company fundamentals and
industry expertise…

…yet they are experiencing divergence between their


investment strategy and their fund performance

NEW OUTCOME

BOTTOM LINE
Cannot identify all the market risks and implied bets in their portfolio.

© Omega Point ! ompnt.com 7


EXAMPLE
PATTERN COMPLEXITY IN UNDERSTANDING PORTFOLIO PERFORMANCE

© Omega Point ! ompnt.com 8


CASE STUDY: BREXIT
ACTIVE MANAGEMENT - Customers manage their portfolio exposures
during Brexit using Omega Point.

© Omega Point ! ompnt.com 9


OMEGA POINT’S VISION & TECHNOLOGY
NEEDS

© Omega Point ! ompnt.com 10


OUR TECHNOLOGY
QUANTITATIVE ENGINE

Nonlinear statistics augmented with User-drive optimization to control


multi-task learning and modulate risk/volatility

Distill key drivers of portfolio Automated, customized extraction of


performance from a high-order models from a synthetic sample of
number of features each individual client’s portfolio

Predict higher moment analysis, e.g. Broad and novel IP differentiating


volatility, directionality, etc. core quantitative technology – first
commercial application of core
innovations

© Omega Point ! ompnt.com 11


EMPOWERING OUR CUSTOMER TO WIN

Re-engineered core Satisfies emerging


quantitative engine for our reporting requirements
customer’s investment style

Visual delivery of simplified, Seamless integration


actionable information

© Omega Point ! ompnt.com 12


IMPACT
OMEGA POINT + DATABRICKS IN ACTION

Research

Customers utilize Omega Point to model new strategies

Reporting

Customers leverage Omega Point to provide on-demand


portfolio analytics to their investors.

© Omega Point ! ompnt.com 13


IDENTIFYING EXPOSURES
OUR CUSTOMER UTILIZES OMEGA POINT TO DISCOVER KEY IMBALANCES IN
THEIR PORTFOLIO

© Omega Point ! ompnt.com 14


IDENTIFYING EXPOSURES
OUR CUSTOMER UTILIZES OMEGA POINT TO DISCOVER KEY IMBALANCES IN
THEIR PORTFOLIO

© Omega Point ! ompnt.com 15


IDENTIFYING EXPOSURES
OUR CUSTOMER UTILIZES OMEGA POINT TO DISCOVER KEY IMBALANCES IN
THEIR PORTFOLIO

Core
Investment

© Omega Point ! ompnt.com 16


IDENTIFYING EXPOSURES
OUR CUSTOMER UTILIZES OMEGA POINT TO DISCOVER KEY IMBALANCES IN
THEIR PORTFOLIO

Non-Core
Investment

Core
Investment

© Omega Point ! ompnt.com 17


LINKING TO PERFORMANCE
THEY SEE HOW THOSE IMBALANCES ATTRIBUTE TO THEIR PORTFOLIO
PERFORMANCE

© Omega Point ! ompnt.com 18


ENABLING REAL-TIME PORTFOLIO INQUIRY – WHAT IF?
THEY USE OMEGA POINT TO RUN A SCENARIO ANALYSIS IN REAL-TIME, BY
MAKING HYPOTHETICAL ADJUSTMENTS TO THEIR PORTFOLIO

© Omega Point ! ompnt.com 19


ENABLING REAL-TIME PORTFOLIO INQUIRY – WHAT IF?
THEY USE OMEGA POINT TO RUN A SCENARIO ANALYSIS IN REAL-TIME, BY
MAKING HYPOTHETICAL ADJUSTMENTS TO THEIR PORTFOLIO

© Omega Point ! ompnt.com 20


ENABLING REAL-TIME PORTFOLIO INQUIRY – WHAT IF?
THEY USE OMEGA POINT TO RUN A SCENARIO ANALYSIS IN REAL-TIME, BY
MAKING HYPOTHETICAL ADJUSTMENTS TO THEIR PORTFOLIO

Adjustments to position
sizes automatically update
long, short and net decile
values for each selected
factor

© Omega Point ! ompnt.com 21


ENHANCING WITH OPTIMIZATION
THEY USE OMEGA POINT TO FIND OPTIMAL PORTFOLIOS THAT MEET THEIR
INVESTMENT CRITERIA

© Omega Point ! ompnt.com 22


THE OMEGA POINT ARCHITECTURE ON
DATABRICKS

© Omega Point ! ompnt.com 23


SUMMARY OF OMEGA POINT TECHNICAL REQUIREMENTS

Input Flexibility Scalability


Handle diverse input sources Simultaneously process
of data & communication market-relevant metrics across
protocols: REST, FTP, Parquet, thousands of securities
JSON, etc.

High Performance Reporting &


Queryability Visualization
Execution engine available for Sanity checking changes in
client input query (“scenario data or in comparing data
analysis”) sources

© Omega Point ! ompnt.com 24


HOW DATABRICKS ADDRESSES THE REQUIREMENTS

Input Flexibility Scalability


Scheduled jobs via notebooks Processing of parallelized
handle external API requests, Spark SQL queries to generate
FTP connections, NoSQL market-relevant metrics
interactions, & Parquet file
reads

High Performance Reporting &


Queryability Visualization
Final output data is stored for Multiple approaches to
ingest into webapp UI and visualize data (dashboards,
queryable via REST API pluggable libraries, etc..)

© Omega Point ! ompnt.com 25


DATA SANITY REPORTING

Easy-to-identify data
inconsistencies with Databricks
visual reporting tools

© Omega Point ! ompnt.com 26


DATA SANITY REPORTING

Easy-to-identify data
inconsistencies with Databricks
visual reporting tools

© Omega Point ! ompnt.com 27


DATA SANITY REPORTING

Easy-to-identify data
inconsistencies with Databricks
visual reporting tools

© Omega Point ! ompnt.com 28


DATA SANITY REPORTING

Easy-to-identify data
inconsistencies with Databricks
visual reporting tools

Houston, we have a problem!

© Omega Point ! ompnt.com 29


HOW WE DEPLOY WITH DATABRICKS
Applications

Node.js React Scala, Python NoSQL


Market Data

Databricks Notebooks AWS Persistent


Storage, S3

Feed

Client data

Monitoring
Raw Data
Feeds

Insights generation

DBFS
Spark

Jobs
Client data

Personalized Factors Features


Insights
Enriched data

Clusters

© Omega Point ! ompnt.com 30


REAL-TIME INTELLIGENCE WITH DATABRICKS REST API

Applications

Node.js React Scala, Python NoSQL


Market Data

AWS Persistent
Databricks REST API Storage, S3

Databricks Notebooks
Scenario
Feed
Analysis

Monitoring
Client data Attribution Raw Data
Feeds

Insights

DBFS
Optimization
generation

Spark

Jobs
Client data

Personalized Factors Features


Insights
Enriched data

Clusters

© Omega Point ! ompnt.com 31


REAL-TIME INTELLIGENCE WITH DATABRICKS REST API

Applications
Real-time Use Case
Node.js React Scala, Python NoSQL
Market Data

•  On-demand scenario analysis


•  Identify optimal AWS
exposure
Persistent
Databricks REST API •  Optimize performance attribution
Storage, S3

Databricks Notebooks
Scenario
Feed
Analysis

Monitoring
Client data Attribution Raw Data
Feeds

Insights

DBFS
Optimization
generation

Spark

Jobs
Client data

Personalized Factors Features


Insights
Enriched data

Clusters

© Omega Point ! ompnt.com 32


LESSONS LEARNED

1. Data management

2. Dynamic clusters

3. Performance optimizations

4. Limited version control

© Omega Point ! ompnt.com 33


SUMMARY

Design of a timely market risk system requires three key


ingredients:
1.  Integration to multiple sources of external market relevant data
2.  Simultaneous processing of 1000s of metrics
3.  On-demand processing / query capabilities from external web
applications

Databricks / Spark meets the above requirements, while


designed to provide maximum iterative flexibility for our data
engineers & data scientists

Our customers actively manage portfolio risk, test new


strategies, and improve their client relationships through
better reporting to their investors.

© Omega Point ! ompnt.com 34


Q&A

© Omega Point ! ompnt.com 35


END

© Omega Point ! ompnt.com 36


PARKING LOT

© Omega Point ! ompnt.com 37

You might also like