SlideShare a Scribd company logo
BlackArbs LLC
Brian Christopher, CFA
April 25, 2017
Brian Christopher
Quantitative researcher, Python developer, CFA charterholder, and founder of Blackarbs LLC, a quantitative research firm.
Six years ago he learned to code using Python for the purpose of creating
algorithmic trading strategies. Four years ago he decided to self publish his
research with a focus on practical, reproducible application.
Now he continues his open research initiatives for a growing community of
traders, researchers, developers, engineers, architects and practitioners
across various industries.
He attained a BSc in Economics from Northeastern University in Boston,
MA and received the Chartered Financial Analyst (CFA) designation in 2016.
About the Speaker
2
 Return prediction is difficult.
 In Time Series Analysis (TSA) we observe multiple phenomena
that violate the rules of stationarity (non-linear processes,
volatility clustering, autocorrelation, etc.)
 Traditional TSA techniques don’t generalize well for asset
return prediction but may be useful in volatility prediction
(ARIMA, ARCH, GARCH)
The Problem with Traditional Time Series Analysis
(Defining the Problem)
3
Traditional Issues with Asset Return Prediction
4
Blackarbs credit: SeanAbu.com
What Can We Do?
5
 We need a model that doesn’t require stationary data.
 We need a model that can approximate non-linear
distributions.
Are mixture models a solution?
6
 Mixture models build on several established concepts.
 Markov models - Sequences where future state
depends only upon current state aka memoryless
processes.
 Hidden Markov models – The true state is unobserved
(hidden) but there are observable factors that give us
information about the true state.
 Expectation-Maximization (E-M) – Class of algorithms
that iterates between computing class parameters and
evaluating the likelihood of the data given those
parameters.
What are mixture models and how do they work?
7
 We know that time series exhibit temporary periods where
the expected means and variances are stable through time.
These periods or regimes can be likened to hidden states.
 If we can better estimate an asset's most likely regime or
hidden state, including the associated means and variances,
then our predictive models become more adaptable and will
likely improve.
 If that's the case, then all we need are observable variables
whose behavior allows us to infer the true hidden state(s).
What are mixture models and how do they work?
Summary
8
 Think of an asset’s returns as a sequence of states or
“regimes”.
 Each regime has its own descriptive statistics including mean
and volatility.
 Examples include “low-vol” vs “high-vol” regimes.
 The model assumes that over time, the asset’s returns will
transition between regimes based on probability.
 Mixture models try to estimate the sequence of regimes,
each regime’s descriptive attributes, and the transition
probabilities between regimes.
 The most well known is the Gaussian Mixture Model (GMM).
What are mixture models and how do they work?
9
 GMM assumes each regime (aka component, class, state) is
generated by a Gaussian process with discoverable
parameters.
 GMM uses E-M under the hood alternating between
estimating the regime parameters (mean, variance) and
likelihood that those parameters could generate the data,
until convergence or another stopping criteria is met.
Can GMMs model non-stationary data sets?
10
 Using SPY return data from Yahoo Finance I fit the GMM with
3 components representing low-volatility, neutral, and high-
volatility regimes.
Can GMMs model non-stationary data sets?
11
 Notice the 2th hidden state or “regime” has a negative mean
return, and the largest variance, which represents our high-
vol regime. The 1th hidden state looks like our low-vol
regime, and the 0th hidden state is somewhere in between.
 The GMM accounts for non-stationarity by design. It chops
the data into regimes and provides us with an estimate of the
mean and variance for each one.
Can GMMs approximate nonlinear data
distributions?
12
 Short answer: Yes
 The top image shows the
GMM trying to fit 2
Gaussians to the data with
predictably bad results.
 The bottom image shows
the GMM trying to fit 16
Gaussians to the data with
more success.
 You can approx. many
complex shapes with
enough Gaussian
components but you risk
overfitting.
Strategy Design
13
Now that we have determined that GMMs can overcome the
weaknesses of the TSA models, we can design a trading
strategy to test its application:
 The strategy uses a total of 4 factors to estimate the
sequence of regimes and their parameters.
• The asset’s returns (Yahoo Finance)
• TEDRATE (FRED)
• T10Y2Y (FRED)
• T10Y3M (FRED)
Strategy Design
14
 The strategy works like so.
 Feed market data to the GMM.
 Call predict method to extract model estimate of last regime
“label”.
 Once we have the “label” we can get the model estimate of
the mean and variance for that regime.
 We feed the parameter estimates to the JohnsonSU
distribution and draw N samples to construct our 99%
confidence intervals (CIs).
 Any returns that fall outside the estimated CIs are defined as
outliers and assumed to mean revert. This strategy assumes
a “buy” of the etf for M number of steps (days) after the
outlier event of a return falling below the lower CI.
Evaluate the Strategy
15
 For a single ETF with a set of parameters, for the walk-
forward period, we can plot the model accuracy as well as an
“equity” plot.
 Below is an example output for SPY:
Evaluate the Strategy
16
To really dive in we have to analyze the aggregated results. By
organizing the data properly...
 We can evaluate each ETF individually across lookbacks, steps
or ks
 We can evaluate the ETFs collectively across lookbacks, steps
or ks
 We’ll look at a few examples in the following slides...
Evaluate the Strategy
17
We can see that across lookbacks SPY, QQQ, TLT have positive
median returns per event. EFA and EEM have more symmetric
distributions.
Evaluate the Strategy
18
We can see a similar distribution pattern here using the
sum_ratio statistic. The sum_ratio statistic is the sum of returns
> 0 divided by the sum of returns < 0. Values greater than 1 are
considered a “success”.
Evaluate the Strategy
19
In this example we evaluate the median return for each ETF by
lookback and across steps. Again we see SPY, QQQ, and TLT
dominating. EFA is also mostly > 0 across steps.
Evaluate the Strategy
20
In this plot we see the sum_ratio metric is > 1 across multiple
dimensions for SPY, QQQ, TLT, and EEM. GLD’s sum_ratio
depends on lookback. EFA is weak, and EEM suffers from
exposure at 21 steps.
Evaluate the Strategy
21
Here we can evaluate the median returns across k (# of
components). Again we see broad success across k for SPY,
QQQ, TLT, and surprisingly EFA. GLD and EEM are more varied
and dependent on lookback.
Evaluate the Strategy
22
Here we can evaluate the median returns across k (# of
components). Again we see broad success across k for SPY,
QQQ, TLT, and surprisingly EFA. GLD and EEM are more varied
and dependent on lookback.
Evaluate the Strategy
23
Here we see a similar pattern among the ETFs with the
sum_ratio for SPY, QQQ, TLT are > 1 across multiple
dimensions. EEM and GLD show similar variability based on
lookback but with passable performance overall.
Conclusions
24
 The secret sauce for GMM accuracy when predicting the
return distribution is choosing the proper sampling
distribution (JohnsonSU vs. Normal)
 US based ETFs showed consistently positive results across
many parameter combinations.
 There was more variability with GLD, EFA, and EEM depending
on metric evaluated, steps (exposure), and k which warrants
an expanded study.
 Must note that there is a risk that these results are simply the
result of being long the US market post 2009!
Further Areas to Explore
25
 GMM with large cross section of single-issuer stocks.
 GMM testing with multiple time frames (intraday, weekly,
monthly).
 GMM as basis for volatility strategy (testing volatility trends
post outlier event.
 Are the outlier events themselves randomly distributed? Or are
they clustered?
 GMM as a basis for multi-asset class strategy (combinations of
equity, options, futures)
Resources
26
* Introduction to Hidden Markov Models with Python, Networkx
and Sklearn - http://blkarb.se/2l6Lndw
* Intro to Expectation-Maximization, K-Means, Gaussian Mixture
Models with Python, Sklearn - http://blkarb.se/2n1A7kn
* Python Data Science Handbook by Jake Vanderplas -
https://github.com/jakevdp/PythonDataScienceHandbook
* Blackarbs blog – Blackarbs.com/blog
* Blackarbs Github -
https://github.com/BlackArbsCEO/Mixture_Models
Contact
27
* Website: Blackarbs.com
* Twitter: @blackarbsCEO
* Linkedin: www.linkedin.com/in/bcrblackarbs/
* Email: bcr@blackarbs.com
SINGAPORE
30 Cecil Street, #19-08,
Prudential Tower,
Singapore – 049712
Phone: +65-9057-8301
INDIA
A-309, Boomerang,
Chandivali Farm Road, Powai,
Mumbai - 400 072
Phone: +91-22-61691400
Learn Automated Trading
Join EPAT™
The Executive Programme in
Algorithmic Trading
Enquire Now!

More Related Content

What's hot

How to design quant trading strategies using “R”?
How to design quant trading strategies using “R”?How to design quant trading strategies using “R”?
How to design quant trading strategies using “R”?
QuantInsti
 
Leveraging artificial intelligence to build algorithmic trading strategies
Leveraging artificial intelligence to build algorithmic trading strategiesLeveraging artificial intelligence to build algorithmic trading strategies
Leveraging artificial intelligence to build algorithmic trading strategies
QuantInsti
 
"Quantitative Trading as a Mathematical Science" by Dr. Haksun Li, Founder an...
"Quantitative Trading as a Mathematical Science" by Dr. Haksun Li, Founder an..."Quantitative Trading as a Mathematical Science" by Dr. Haksun Li, Founder an...
"Quantitative Trading as a Mathematical Science" by Dr. Haksun Li, Founder an...
Quantopian
 
Statistics - The Missing Link Between Technical Analysis and Algorithmic Trad...
Statistics - The Missing Link Between Technical Analysis and Algorithmic Trad...Statistics - The Missing Link Between Technical Analysis and Algorithmic Trad...
Statistics - The Missing Link Between Technical Analysis and Algorithmic Trad...
Quantopian
 
Risk Management: Maximising Long-Term Growth Presentation
Risk Management: Maximising Long-Term Growth PresentationRisk Management: Maximising Long-Term Growth Presentation
Risk Management: Maximising Long-Term Growth Presentation
QuantInsti
 
"A Framework for Developing Trading Models Based on Machine Learning" by Kris...
"A Framework for Developing Trading Models Based on Machine Learning" by Kris..."A Framework for Developing Trading Models Based on Machine Learning" by Kris...
"A Framework for Developing Trading Models Based on Machine Learning" by Kris...
Quantopian
 
Algorithmic trading and Machine Learning by Michael Kearns, Professor of Comp...
Algorithmic trading and Machine Learning by Michael Kearns, Professor of Comp...Algorithmic trading and Machine Learning by Michael Kearns, Professor of Comp...
Algorithmic trading and Machine Learning by Michael Kearns, Professor of Comp...
Quantopian
 
Beware of Low Frequency Data by Ernie Chan, Managing Member, QTS Capital Mana...
Beware of Low Frequency Data by Ernie Chan, Managing Member, QTS Capital Mana...Beware of Low Frequency Data by Ernie Chan, Managing Member, QTS Capital Mana...
Beware of Low Frequency Data by Ernie Chan, Managing Member, QTS Capital Mana...
Quantopian
 
Using Java & Genetic Algorithms to Beat the Market
Using Java & Genetic Algorithms to Beat the MarketUsing Java & Genetic Algorithms to Beat the Market
Using Java & Genetic Algorithms to Beat the Market
Matthew Ring
 
Modeling Transaction Costs for Algorithmic Strategies
Modeling Transaction Costs for Algorithmic StrategiesModeling Transaction Costs for Algorithmic Strategies
Modeling Transaction Costs for Algorithmic Strategies
Quantopian
 
"Active Learning in Trading Algorithms" by David Fellah, Head of the EMEA Lin...
"Active Learning in Trading Algorithms" by David Fellah, Head of the EMEA Lin..."Active Learning in Trading Algorithms" by David Fellah, Head of the EMEA Lin...
"Active Learning in Trading Algorithms" by David Fellah, Head of the EMEA Lin...
Quantopian
 
Pairs Trading from NYC Algorithmic Trading Meetup November '13
Pairs Trading from NYC Algorithmic Trading Meetup November '13Pairs Trading from NYC Algorithmic Trading Meetup November '13
Pairs Trading from NYC Algorithmic Trading Meetup November '13
Quantopian
 
Stock Market Price Prediction Using Technical Analysis
Stock Market Price Prediction Using Technical AnalysisStock Market Price Prediction Using Technical Analysis
Stock Market Price Prediction Using Technical Analysis
ASHEESHVERMA6
 
The QuantCon Keynote: "Counter Trend Trading – Threat or Complement to Trend ...
The QuantCon Keynote: "Counter Trend Trading – Threat or Complement to Trend ...The QuantCon Keynote: "Counter Trend Trading – Threat or Complement to Trend ...
The QuantCon Keynote: "Counter Trend Trading – Threat or Complement to Trend ...
Quantopian
 
Should You Build Your Own Backtester? by Michael Halls-Moore at QuantCon 2016
Should You Build Your Own Backtester? by Michael Halls-Moore at QuantCon 2016Should You Build Your Own Backtester? by Michael Halls-Moore at QuantCon 2016
Should You Build Your Own Backtester? by Michael Halls-Moore at QuantCon 2016
Quantopian
 
From Backtesting to Live Trading by Vesna Straser at QuantCon 2016
From Backtesting to Live Trading by Vesna Straser at QuantCon 2016From Backtesting to Live Trading by Vesna Straser at QuantCon 2016
From Backtesting to Live Trading by Vesna Straser at QuantCon 2016
Quantopian
 
Pairs Trading
Pairs TradingPairs Trading
Pairs Trading
Vincent Chen
 
A Guided Tour of Machine Learning for Traders by Tucker Balch at QuantCon 2016
A Guided Tour of Machine Learning for Traders by Tucker Balch at QuantCon 2016A Guided Tour of Machine Learning for Traders by Tucker Balch at QuantCon 2016
A Guided Tour of Machine Learning for Traders by Tucker Balch at QuantCon 2016
Quantopian
 
DIY Quant Strategies on Quantopian
DIY Quant Strategies on QuantopianDIY Quant Strategies on Quantopian
DIY Quant Strategies on Quantopian
Jess Stauth
 

What's hot (20)

How to design quant trading strategies using “R”?
How to design quant trading strategies using “R”?How to design quant trading strategies using “R”?
How to design quant trading strategies using “R”?
 
Leveraging artificial intelligence to build algorithmic trading strategies
Leveraging artificial intelligence to build algorithmic trading strategiesLeveraging artificial intelligence to build algorithmic trading strategies
Leveraging artificial intelligence to build algorithmic trading strategies
 
"Quantitative Trading as a Mathematical Science" by Dr. Haksun Li, Founder an...
"Quantitative Trading as a Mathematical Science" by Dr. Haksun Li, Founder an..."Quantitative Trading as a Mathematical Science" by Dr. Haksun Li, Founder an...
"Quantitative Trading as a Mathematical Science" by Dr. Haksun Li, Founder an...
 
Statistics - The Missing Link Between Technical Analysis and Algorithmic Trad...
Statistics - The Missing Link Between Technical Analysis and Algorithmic Trad...Statistics - The Missing Link Between Technical Analysis and Algorithmic Trad...
Statistics - The Missing Link Between Technical Analysis and Algorithmic Trad...
 
Risk Management: Maximising Long-Term Growth Presentation
Risk Management: Maximising Long-Term Growth PresentationRisk Management: Maximising Long-Term Growth Presentation
Risk Management: Maximising Long-Term Growth Presentation
 
"A Framework for Developing Trading Models Based on Machine Learning" by Kris...
"A Framework for Developing Trading Models Based on Machine Learning" by Kris..."A Framework for Developing Trading Models Based on Machine Learning" by Kris...
"A Framework for Developing Trading Models Based on Machine Learning" by Kris...
 
Algorithmic trading and Machine Learning by Michael Kearns, Professor of Comp...
Algorithmic trading and Machine Learning by Michael Kearns, Professor of Comp...Algorithmic trading and Machine Learning by Michael Kearns, Professor of Comp...
Algorithmic trading and Machine Learning by Michael Kearns, Professor of Comp...
 
Beware of Low Frequency Data by Ernie Chan, Managing Member, QTS Capital Mana...
Beware of Low Frequency Data by Ernie Chan, Managing Member, QTS Capital Mana...Beware of Low Frequency Data by Ernie Chan, Managing Member, QTS Capital Mana...
Beware of Low Frequency Data by Ernie Chan, Managing Member, QTS Capital Mana...
 
Using Java & Genetic Algorithms to Beat the Market
Using Java & Genetic Algorithms to Beat the MarketUsing Java & Genetic Algorithms to Beat the Market
Using Java & Genetic Algorithms to Beat the Market
 
Modeling Transaction Costs for Algorithmic Strategies
Modeling Transaction Costs for Algorithmic StrategiesModeling Transaction Costs for Algorithmic Strategies
Modeling Transaction Costs for Algorithmic Strategies
 
"Active Learning in Trading Algorithms" by David Fellah, Head of the EMEA Lin...
"Active Learning in Trading Algorithms" by David Fellah, Head of the EMEA Lin..."Active Learning in Trading Algorithms" by David Fellah, Head of the EMEA Lin...
"Active Learning in Trading Algorithms" by David Fellah, Head of the EMEA Lin...
 
Pairs Trading from NYC Algorithmic Trading Meetup November '13
Pairs Trading from NYC Algorithmic Trading Meetup November '13Pairs Trading from NYC Algorithmic Trading Meetup November '13
Pairs Trading from NYC Algorithmic Trading Meetup November '13
 
Stock Market Price Prediction Using Technical Analysis
Stock Market Price Prediction Using Technical AnalysisStock Market Price Prediction Using Technical Analysis
Stock Market Price Prediction Using Technical Analysis
 
The QuantCon Keynote: "Counter Trend Trading – Threat or Complement to Trend ...
The QuantCon Keynote: "Counter Trend Trading – Threat or Complement to Trend ...The QuantCon Keynote: "Counter Trend Trading – Threat or Complement to Trend ...
The QuantCon Keynote: "Counter Trend Trading – Threat or Complement to Trend ...
 
Should You Build Your Own Backtester? by Michael Halls-Moore at QuantCon 2016
Should You Build Your Own Backtester? by Michael Halls-Moore at QuantCon 2016Should You Build Your Own Backtester? by Michael Halls-Moore at QuantCon 2016
Should You Build Your Own Backtester? by Michael Halls-Moore at QuantCon 2016
 
From Backtesting to Live Trading by Vesna Straser at QuantCon 2016
From Backtesting to Live Trading by Vesna Straser at QuantCon 2016From Backtesting to Live Trading by Vesna Straser at QuantCon 2016
From Backtesting to Live Trading by Vesna Straser at QuantCon 2016
 
Pairs Trading
Pairs TradingPairs Trading
Pairs Trading
 
A Guided Tour of Machine Learning for Traders by Tucker Balch at QuantCon 2016
A Guided Tour of Machine Learning for Traders by Tucker Balch at QuantCon 2016A Guided Tour of Machine Learning for Traders by Tucker Balch at QuantCon 2016
A Guided Tour of Machine Learning for Traders by Tucker Balch at QuantCon 2016
 
DIY Quant Strategies on Quantopian
DIY Quant Strategies on QuantopianDIY Quant Strategies on Quantopian
DIY Quant Strategies on Quantopian
 
SAPM
SAPMSAPM
SAPM
 

Similar to Can we use Mixture Models to Predict Market Bottoms? by Brian Christopher - 25th April 2017

Testing and extending the capital asset pricing model
Testing and extending the capital asset pricing modelTesting and extending the capital asset pricing model
Testing and extending the capital asset pricing model
Gabriel Koh
 
Federico Thibaud - Capital Structure Arbitrage
Federico Thibaud - Capital Structure ArbitrageFederico Thibaud - Capital Structure Arbitrage
Federico Thibaud - Capital Structure ArbitrageFederico Thibaud
 
Oriskany' TCS Multi-Factor Timing - Factor Investing White Paper
Oriskany' TCS Multi-Factor Timing - Factor Investing White PaperOriskany' TCS Multi-Factor Timing - Factor Investing White Paper
Oriskany' TCS Multi-Factor Timing - Factor Investing White Paper
Dess Kabakova
 
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
ijscmcj
 
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
ijscmcj
 
Volatility Forecasting - A Performance Measure of Garch Techniques With Diffe...
Volatility Forecasting - A Performance Measure of Garch Techniques With Diffe...Volatility Forecasting - A Performance Measure of Garch Techniques With Diffe...
Volatility Forecasting - A Performance Measure of Garch Techniques With Diffe...
ijscmcj
 
Volatility forecasting a_performance_mea
Volatility forecasting a_performance_meaVolatility forecasting a_performance_mea
Volatility forecasting a_performance_mea
ijscmcj
 
Xue paper-01-13-12
Xue paper-01-13-12Xue paper-01-13-12
Xue paper-01-13-12
Yuhong Xue
 
Advance IQ Capital Quantitative Models
Advance IQ Capital Quantitative ModelsAdvance IQ Capital Quantitative Models
Advance IQ Capital Quantitative Models
Andrew Leon Palashewsky, CMT
 
Advance iq capital quantitative models
Advance iq capital quantitative modelsAdvance iq capital quantitative models
Advance iq capital quantitative models
Andrew214
 
Financial Economics Essay Maxwell Mayhew 2015
Financial Economics Essay Maxwell Mayhew 2015Financial Economics Essay Maxwell Mayhew 2015
Financial Economics Essay Maxwell Mayhew 2015Max Mayhew
 
All that Glitters Is Not Gold_Comparing Backtest and Out-of-Sample Performanc...
All that Glitters Is Not Gold_Comparing Backtest and Out-of-Sample Performanc...All that Glitters Is Not Gold_Comparing Backtest and Out-of-Sample Performanc...
All that Glitters Is Not Gold_Comparing Backtest and Out-of-Sample Performanc...justinlent
 
Net net-graham -paper-june-2010
Net net-graham -paper-june-2010Net net-graham -paper-june-2010
Net net-graham -paper-june-2010vidyanshu
 
Machine learning in credit risk modeling : a James white paper
Machine learning in credit risk modeling : a James white paperMachine learning in credit risk modeling : a James white paper
Machine learning in credit risk modeling : a James white paper
James by CrowdProcess
 
Capital asset pricing Model (CAPM-Asif).pptx
Capital asset pricing Model (CAPM-Asif).pptxCapital asset pricing Model (CAPM-Asif).pptx
Capital asset pricing Model (CAPM-Asif).pptx
AsifHasan395472
 
Capm
CapmCapm
Trading System Seminar Handout
Trading System Seminar HandoutTrading System Seminar Handout
Trading System Seminar HandoutDean Hoffman
 

Similar to Can we use Mixture Models to Predict Market Bottoms? by Brian Christopher - 25th April 2017 (20)

Testing and extending the capital asset pricing model
Testing and extending the capital asset pricing modelTesting and extending the capital asset pricing model
Testing and extending the capital asset pricing model
 
Federico Thibaud - Capital Structure Arbitrage
Federico Thibaud - Capital Structure ArbitrageFederico Thibaud - Capital Structure Arbitrage
Federico Thibaud - Capital Structure Arbitrage
 
Statistical Arbitrage
Statistical Arbitrage Statistical Arbitrage
Statistical Arbitrage
 
Oriskany' TCS Multi-Factor Timing - Factor Investing White Paper
Oriskany' TCS Multi-Factor Timing - Factor Investing White PaperOriskany' TCS Multi-Factor Timing - Factor Investing White Paper
Oriskany' TCS Multi-Factor Timing - Factor Investing White Paper
 
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
 
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
VOLATILITY FORECASTING - A PERFORMANCE MEASURE OF GARCH TECHNIQUES WITH DIFFE...
 
Volatility Forecasting - A Performance Measure of Garch Techniques With Diffe...
Volatility Forecasting - A Performance Measure of Garch Techniques With Diffe...Volatility Forecasting - A Performance Measure of Garch Techniques With Diffe...
Volatility Forecasting - A Performance Measure of Garch Techniques With Diffe...
 
Volatility forecasting a_performance_mea
Volatility forecasting a_performance_meaVolatility forecasting a_performance_mea
Volatility forecasting a_performance_mea
 
Xue paper-01-13-12
Xue paper-01-13-12Xue paper-01-13-12
Xue paper-01-13-12
 
Advance IQ Capital Quantitative Models
Advance IQ Capital Quantitative ModelsAdvance IQ Capital Quantitative Models
Advance IQ Capital Quantitative Models
 
Advance iq capital quantitative models
Advance iq capital quantitative modelsAdvance iq capital quantitative models
Advance iq capital quantitative models
 
Financial Economics Essay Maxwell Mayhew 2015
Financial Economics Essay Maxwell Mayhew 2015Financial Economics Essay Maxwell Mayhew 2015
Financial Economics Essay Maxwell Mayhew 2015
 
All that Glitters Is Not Gold_Comparing Backtest and Out-of-Sample Performanc...
All that Glitters Is Not Gold_Comparing Backtest and Out-of-Sample Performanc...All that Glitters Is Not Gold_Comparing Backtest and Out-of-Sample Performanc...
All that Glitters Is Not Gold_Comparing Backtest and Out-of-Sample Performanc...
 
Net net-graham -paper-june-2010
Net net-graham -paper-june-2010Net net-graham -paper-june-2010
Net net-graham -paper-june-2010
 
HEHEH
HEHEHHEHEH
HEHEH
 
HEHEH
HEHEHHEHEH
HEHEH
 
Machine learning in credit risk modeling : a James white paper
Machine learning in credit risk modeling : a James white paperMachine learning in credit risk modeling : a James white paper
Machine learning in credit risk modeling : a James white paper
 
Capital asset pricing Model (CAPM-Asif).pptx
Capital asset pricing Model (CAPM-Asif).pptxCapital asset pricing Model (CAPM-Asif).pptx
Capital asset pricing Model (CAPM-Asif).pptx
 
Capm
CapmCapm
Capm
 
Trading System Seminar Handout
Trading System Seminar HandoutTrading System Seminar Handout
Trading System Seminar Handout
 

More from QuantInsti

ChatGPT and Machine Learning in Trading
ChatGPT and Machine Learning in TradingChatGPT and Machine Learning in Trading
ChatGPT and Machine Learning in Trading
QuantInsti
 
Introduction to Quantitative Factor Investing
Introduction to Quantitative Factor InvestingIntroduction to Quantitative Factor Investing
Introduction to Quantitative Factor Investing
QuantInsti
 
Machine Learning for Options Trading
Machine Learning for Options TradingMachine Learning for Options Trading
Machine Learning for Options Trading
QuantInsti
 
Portfolio Assets Allocation with Machine Learning
Portfolio Assets Allocation with Machine LearningPortfolio Assets Allocation with Machine Learning
Portfolio Assets Allocation with Machine Learning
QuantInsti
 
Price Action Trading - An Introduction
Price Action Trading - An IntroductionPrice Action Trading - An Introduction
Price Action Trading - An Introduction
QuantInsti
 
Introduction to Systematic Options Trading
Introduction to Systematic Options TradingIntroduction to Systematic Options Trading
Introduction to Systematic Options Trading
QuantInsti
 
Competitive Edges in Algorithmic Trading
Competitive Edges in Algorithmic TradingCompetitive Edges in Algorithmic Trading
Competitive Edges in Algorithmic Trading
QuantInsti
 
Volatility Trading: Trading The Fear Index VIX
Volatility Trading: Trading The Fear Index VIXVolatility Trading: Trading The Fear Index VIX
Volatility Trading: Trading The Fear Index VIX
QuantInsti
 
Big Data And The Future Of Retail Investing
Big Data And The Future Of Retail InvestingBig Data And The Future Of Retail Investing
Big Data And The Future Of Retail Investing
QuantInsti
 
Backtest of Short Straddles on SPX Index
Backtest of Short Straddles on SPX IndexBacktest of Short Straddles on SPX Index
Backtest of Short Straddles on SPX Index
QuantInsti
 
Pairs Trading In the Brazilian Stock Market
Pairs Trading In the Brazilian Stock MarketPairs Trading In the Brazilian Stock Market
Pairs Trading In the Brazilian Stock Market
QuantInsti
 
How To Set Up Automated Trading
How To Set Up Automated TradingHow To Set Up Automated Trading
How To Set Up Automated Trading
QuantInsti
 
How To Set Up Automated Trading
How To Set Up Automated TradingHow To Set Up Automated Trading
How To Set Up Automated Trading
QuantInsti
 
Quantitative Data Analysis of Cryptocurrencies
Quantitative Data Analysis of CryptocurrenciesQuantitative Data Analysis of Cryptocurrencies
Quantitative Data Analysis of Cryptocurrencies
QuantInsti
 
Introduction to Quantitative Trading - Investment Management Club of Yale Uni...
Introduction to Quantitative Trading - Investment Management Club of Yale Uni...Introduction to Quantitative Trading - Investment Management Club of Yale Uni...
Introduction to Quantitative Trading - Investment Management Club of Yale Uni...
QuantInsti
 
How to automate an options day trading strategy
How to automate an options day trading strategyHow to automate an options day trading strategy
How to automate an options day trading strategy
QuantInsti
 
Predict daily stock prices with random forest classifier, technical indicator...
Predict daily stock prices with random forest classifier, technical indicator...Predict daily stock prices with random forest classifier, technical indicator...
Predict daily stock prices with random forest classifier, technical indicator...
QuantInsti
 
How Pandemics Impact the Financial Markets - A Quantitative Analysis
How Pandemics Impact the Financial Markets - A Quantitative AnalysisHow Pandemics Impact the Financial Markets - A Quantitative Analysis
How Pandemics Impact the Financial Markets - A Quantitative Analysis
QuantInsti
 
Webinar on Algorithmic Trading - Why make the move? with Vivek Krishnamoorthy...
Webinar on Algorithmic Trading - Why make the move? with Vivek Krishnamoorthy...Webinar on Algorithmic Trading - Why make the move? with Vivek Krishnamoorthy...
Webinar on Algorithmic Trading - Why make the move? with Vivek Krishnamoorthy...
QuantInsti
 
Backtesting And Live Trading With Interactive Brokers Using Python With Dr. H...
Backtesting And Live Trading With Interactive Brokers Using Python With Dr. H...Backtesting And Live Trading With Interactive Brokers Using Python With Dr. H...
Backtesting And Live Trading With Interactive Brokers Using Python With Dr. H...
QuantInsti
 

More from QuantInsti (20)

ChatGPT and Machine Learning in Trading
ChatGPT and Machine Learning in TradingChatGPT and Machine Learning in Trading
ChatGPT and Machine Learning in Trading
 
Introduction to Quantitative Factor Investing
Introduction to Quantitative Factor InvestingIntroduction to Quantitative Factor Investing
Introduction to Quantitative Factor Investing
 
Machine Learning for Options Trading
Machine Learning for Options TradingMachine Learning for Options Trading
Machine Learning for Options Trading
 
Portfolio Assets Allocation with Machine Learning
Portfolio Assets Allocation with Machine LearningPortfolio Assets Allocation with Machine Learning
Portfolio Assets Allocation with Machine Learning
 
Price Action Trading - An Introduction
Price Action Trading - An IntroductionPrice Action Trading - An Introduction
Price Action Trading - An Introduction
 
Introduction to Systematic Options Trading
Introduction to Systematic Options TradingIntroduction to Systematic Options Trading
Introduction to Systematic Options Trading
 
Competitive Edges in Algorithmic Trading
Competitive Edges in Algorithmic TradingCompetitive Edges in Algorithmic Trading
Competitive Edges in Algorithmic Trading
 
Volatility Trading: Trading The Fear Index VIX
Volatility Trading: Trading The Fear Index VIXVolatility Trading: Trading The Fear Index VIX
Volatility Trading: Trading The Fear Index VIX
 
Big Data And The Future Of Retail Investing
Big Data And The Future Of Retail InvestingBig Data And The Future Of Retail Investing
Big Data And The Future Of Retail Investing
 
Backtest of Short Straddles on SPX Index
Backtest of Short Straddles on SPX IndexBacktest of Short Straddles on SPX Index
Backtest of Short Straddles on SPX Index
 
Pairs Trading In the Brazilian Stock Market
Pairs Trading In the Brazilian Stock MarketPairs Trading In the Brazilian Stock Market
Pairs Trading In the Brazilian Stock Market
 
How To Set Up Automated Trading
How To Set Up Automated TradingHow To Set Up Automated Trading
How To Set Up Automated Trading
 
How To Set Up Automated Trading
How To Set Up Automated TradingHow To Set Up Automated Trading
How To Set Up Automated Trading
 
Quantitative Data Analysis of Cryptocurrencies
Quantitative Data Analysis of CryptocurrenciesQuantitative Data Analysis of Cryptocurrencies
Quantitative Data Analysis of Cryptocurrencies
 
Introduction to Quantitative Trading - Investment Management Club of Yale Uni...
Introduction to Quantitative Trading - Investment Management Club of Yale Uni...Introduction to Quantitative Trading - Investment Management Club of Yale Uni...
Introduction to Quantitative Trading - Investment Management Club of Yale Uni...
 
How to automate an options day trading strategy
How to automate an options day trading strategyHow to automate an options day trading strategy
How to automate an options day trading strategy
 
Predict daily stock prices with random forest classifier, technical indicator...
Predict daily stock prices with random forest classifier, technical indicator...Predict daily stock prices with random forest classifier, technical indicator...
Predict daily stock prices with random forest classifier, technical indicator...
 
How Pandemics Impact the Financial Markets - A Quantitative Analysis
How Pandemics Impact the Financial Markets - A Quantitative AnalysisHow Pandemics Impact the Financial Markets - A Quantitative Analysis
How Pandemics Impact the Financial Markets - A Quantitative Analysis
 
Webinar on Algorithmic Trading - Why make the move? with Vivek Krishnamoorthy...
Webinar on Algorithmic Trading - Why make the move? with Vivek Krishnamoorthy...Webinar on Algorithmic Trading - Why make the move? with Vivek Krishnamoorthy...
Webinar on Algorithmic Trading - Why make the move? with Vivek Krishnamoorthy...
 
Backtesting And Live Trading With Interactive Brokers Using Python With Dr. H...
Backtesting And Live Trading With Interactive Brokers Using Python With Dr. H...Backtesting And Live Trading With Interactive Brokers Using Python With Dr. H...
Backtesting And Live Trading With Interactive Brokers Using Python With Dr. H...
 

Recently uploaded

The Roman Empire A Historical Colossus.pdf
The Roman Empire A Historical Colossus.pdfThe Roman Empire A Historical Colossus.pdf
The Roman Empire A Historical Colossus.pdf
kaushalkr1407
 
1.4 modern child centered education - mahatma gandhi-2.pptx
1.4 modern child centered education - mahatma gandhi-2.pptx1.4 modern child centered education - mahatma gandhi-2.pptx
1.4 modern child centered education - mahatma gandhi-2.pptx
JosvitaDsouza2
 
Polish students' mobility in the Czech Republic
Polish students' mobility in the Czech RepublicPolish students' mobility in the Czech Republic
Polish students' mobility in the Czech Republic
Anna Sz.
 
Francesca Gottschalk - How can education support child empowerment.pptx
Francesca Gottschalk - How can education support child empowerment.pptxFrancesca Gottschalk - How can education support child empowerment.pptx
Francesca Gottschalk - How can education support child empowerment.pptx
EduSkills OECD
 
A Strategic Approach: GenAI in Education
A Strategic Approach: GenAI in EducationA Strategic Approach: GenAI in Education
A Strategic Approach: GenAI in Education
Peter Windle
 
The approach at University of Liverpool.pptx
The approach at University of Liverpool.pptxThe approach at University of Liverpool.pptx
The approach at University of Liverpool.pptx
Jisc
 
Synthetic Fiber Construction in lab .pptx
Synthetic Fiber Construction in lab .pptxSynthetic Fiber Construction in lab .pptx
Synthetic Fiber Construction in lab .pptx
Pavel ( NSTU)
 
Mule 4.6 & Java 17 Upgrade | MuleSoft Mysore Meetup #46
Mule 4.6 & Java 17 Upgrade | MuleSoft Mysore Meetup #46Mule 4.6 & Java 17 Upgrade | MuleSoft Mysore Meetup #46
Mule 4.6 & Java 17 Upgrade | MuleSoft Mysore Meetup #46
MysoreMuleSoftMeetup
 
2024.06.01 Introducing a competency framework for languag learning materials ...
2024.06.01 Introducing a competency framework for languag learning materials ...2024.06.01 Introducing a competency framework for languag learning materials ...
2024.06.01 Introducing a competency framework for languag learning materials ...
Sandy Millin
 
Home assignment II on Spectroscopy 2024 Answers.pdf
Home assignment II on Spectroscopy 2024 Answers.pdfHome assignment II on Spectroscopy 2024 Answers.pdf
Home assignment II on Spectroscopy 2024 Answers.pdf
Tamralipta Mahavidyalaya
 
How libraries can support authors with open access requirements for UKRI fund...
How libraries can support authors with open access requirements for UKRI fund...How libraries can support authors with open access requirements for UKRI fund...
How libraries can support authors with open access requirements for UKRI fund...
Jisc
 
The Accursed House by Émile Gaboriau.pptx
The Accursed House by Émile Gaboriau.pptxThe Accursed House by Émile Gaboriau.pptx
The Accursed House by Émile Gaboriau.pptx
DhatriParmar
 
678020731-Sumas-y-Restas-Para-Colorear.pdf
678020731-Sumas-y-Restas-Para-Colorear.pdf678020731-Sumas-y-Restas-Para-Colorear.pdf
678020731-Sumas-y-Restas-Para-Colorear.pdf
CarlosHernanMontoyab2
 
Model Attribute Check Company Auto Property
Model Attribute  Check Company Auto PropertyModel Attribute  Check Company Auto Property
Model Attribute Check Company Auto Property
Celine George
 
Language Across the Curriculm LAC B.Ed.
Language Across the  Curriculm LAC B.Ed.Language Across the  Curriculm LAC B.Ed.
Language Across the Curriculm LAC B.Ed.
Atul Kumar Singh
 
special B.ed 2nd year old paper_20240531.pdf
special B.ed 2nd year old paper_20240531.pdfspecial B.ed 2nd year old paper_20240531.pdf
special B.ed 2nd year old paper_20240531.pdf
Special education needs
 
How to Make a Field invisible in Odoo 17
How to Make a Field invisible in Odoo 17How to Make a Field invisible in Odoo 17
How to Make a Field invisible in Odoo 17
Celine George
 
Palestine last event orientationfvgnh .pptx
Palestine last event orientationfvgnh .pptxPalestine last event orientationfvgnh .pptx
Palestine last event orientationfvgnh .pptx
RaedMohamed3
 
Introduction to AI for Nonprofits with Tapp Network
Introduction to AI for Nonprofits with Tapp NetworkIntroduction to AI for Nonprofits with Tapp Network
Introduction to AI for Nonprofits with Tapp Network
TechSoup
 
Instructions for Submissions thorugh G- Classroom.pptx
Instructions for Submissions thorugh G- Classroom.pptxInstructions for Submissions thorugh G- Classroom.pptx
Instructions for Submissions thorugh G- Classroom.pptx
Jheel Barad
 

Recently uploaded (20)

The Roman Empire A Historical Colossus.pdf
The Roman Empire A Historical Colossus.pdfThe Roman Empire A Historical Colossus.pdf
The Roman Empire A Historical Colossus.pdf
 
1.4 modern child centered education - mahatma gandhi-2.pptx
1.4 modern child centered education - mahatma gandhi-2.pptx1.4 modern child centered education - mahatma gandhi-2.pptx
1.4 modern child centered education - mahatma gandhi-2.pptx
 
Polish students' mobility in the Czech Republic
Polish students' mobility in the Czech RepublicPolish students' mobility in the Czech Republic
Polish students' mobility in the Czech Republic
 
Francesca Gottschalk - How can education support child empowerment.pptx
Francesca Gottschalk - How can education support child empowerment.pptxFrancesca Gottschalk - How can education support child empowerment.pptx
Francesca Gottschalk - How can education support child empowerment.pptx
 
A Strategic Approach: GenAI in Education
A Strategic Approach: GenAI in EducationA Strategic Approach: GenAI in Education
A Strategic Approach: GenAI in Education
 
The approach at University of Liverpool.pptx
The approach at University of Liverpool.pptxThe approach at University of Liverpool.pptx
The approach at University of Liverpool.pptx
 
Synthetic Fiber Construction in lab .pptx
Synthetic Fiber Construction in lab .pptxSynthetic Fiber Construction in lab .pptx
Synthetic Fiber Construction in lab .pptx
 
Mule 4.6 & Java 17 Upgrade | MuleSoft Mysore Meetup #46
Mule 4.6 & Java 17 Upgrade | MuleSoft Mysore Meetup #46Mule 4.6 & Java 17 Upgrade | MuleSoft Mysore Meetup #46
Mule 4.6 & Java 17 Upgrade | MuleSoft Mysore Meetup #46
 
2024.06.01 Introducing a competency framework for languag learning materials ...
2024.06.01 Introducing a competency framework for languag learning materials ...2024.06.01 Introducing a competency framework for languag learning materials ...
2024.06.01 Introducing a competency framework for languag learning materials ...
 
Home assignment II on Spectroscopy 2024 Answers.pdf
Home assignment II on Spectroscopy 2024 Answers.pdfHome assignment II on Spectroscopy 2024 Answers.pdf
Home assignment II on Spectroscopy 2024 Answers.pdf
 
How libraries can support authors with open access requirements for UKRI fund...
How libraries can support authors with open access requirements for UKRI fund...How libraries can support authors with open access requirements for UKRI fund...
How libraries can support authors with open access requirements for UKRI fund...
 
The Accursed House by Émile Gaboriau.pptx
The Accursed House by Émile Gaboriau.pptxThe Accursed House by Émile Gaboriau.pptx
The Accursed House by Émile Gaboriau.pptx
 
678020731-Sumas-y-Restas-Para-Colorear.pdf
678020731-Sumas-y-Restas-Para-Colorear.pdf678020731-Sumas-y-Restas-Para-Colorear.pdf
678020731-Sumas-y-Restas-Para-Colorear.pdf
 
Model Attribute Check Company Auto Property
Model Attribute  Check Company Auto PropertyModel Attribute  Check Company Auto Property
Model Attribute Check Company Auto Property
 
Language Across the Curriculm LAC B.Ed.
Language Across the  Curriculm LAC B.Ed.Language Across the  Curriculm LAC B.Ed.
Language Across the Curriculm LAC B.Ed.
 
special B.ed 2nd year old paper_20240531.pdf
special B.ed 2nd year old paper_20240531.pdfspecial B.ed 2nd year old paper_20240531.pdf
special B.ed 2nd year old paper_20240531.pdf
 
How to Make a Field invisible in Odoo 17
How to Make a Field invisible in Odoo 17How to Make a Field invisible in Odoo 17
How to Make a Field invisible in Odoo 17
 
Palestine last event orientationfvgnh .pptx
Palestine last event orientationfvgnh .pptxPalestine last event orientationfvgnh .pptx
Palestine last event orientationfvgnh .pptx
 
Introduction to AI for Nonprofits with Tapp Network
Introduction to AI for Nonprofits with Tapp NetworkIntroduction to AI for Nonprofits with Tapp Network
Introduction to AI for Nonprofits with Tapp Network
 
Instructions for Submissions thorugh G- Classroom.pptx
Instructions for Submissions thorugh G- Classroom.pptxInstructions for Submissions thorugh G- Classroom.pptx
Instructions for Submissions thorugh G- Classroom.pptx
 

Can we use Mixture Models to Predict Market Bottoms? by Brian Christopher - 25th April 2017

  • 1. BlackArbs LLC Brian Christopher, CFA April 25, 2017
  • 2. Brian Christopher Quantitative researcher, Python developer, CFA charterholder, and founder of Blackarbs LLC, a quantitative research firm. Six years ago he learned to code using Python for the purpose of creating algorithmic trading strategies. Four years ago he decided to self publish his research with a focus on practical, reproducible application. Now he continues his open research initiatives for a growing community of traders, researchers, developers, engineers, architects and practitioners across various industries. He attained a BSc in Economics from Northeastern University in Boston, MA and received the Chartered Financial Analyst (CFA) designation in 2016. About the Speaker 2
  • 3.  Return prediction is difficult.  In Time Series Analysis (TSA) we observe multiple phenomena that violate the rules of stationarity (non-linear processes, volatility clustering, autocorrelation, etc.)  Traditional TSA techniques don’t generalize well for asset return prediction but may be useful in volatility prediction (ARIMA, ARCH, GARCH) The Problem with Traditional Time Series Analysis (Defining the Problem) 3
  • 4. Traditional Issues with Asset Return Prediction 4 Blackarbs credit: SeanAbu.com
  • 5. What Can We Do? 5  We need a model that doesn’t require stationary data.  We need a model that can approximate non-linear distributions.
  • 6. Are mixture models a solution? 6  Mixture models build on several established concepts.  Markov models - Sequences where future state depends only upon current state aka memoryless processes.  Hidden Markov models – The true state is unobserved (hidden) but there are observable factors that give us information about the true state.  Expectation-Maximization (E-M) – Class of algorithms that iterates between computing class parameters and evaluating the likelihood of the data given those parameters.
  • 7. What are mixture models and how do they work? 7  We know that time series exhibit temporary periods where the expected means and variances are stable through time. These periods or regimes can be likened to hidden states.  If we can better estimate an asset's most likely regime or hidden state, including the associated means and variances, then our predictive models become more adaptable and will likely improve.  If that's the case, then all we need are observable variables whose behavior allows us to infer the true hidden state(s).
  • 8. What are mixture models and how do they work? Summary 8  Think of an asset’s returns as a sequence of states or “regimes”.  Each regime has its own descriptive statistics including mean and volatility.  Examples include “low-vol” vs “high-vol” regimes.  The model assumes that over time, the asset’s returns will transition between regimes based on probability.  Mixture models try to estimate the sequence of regimes, each regime’s descriptive attributes, and the transition probabilities between regimes.  The most well known is the Gaussian Mixture Model (GMM).
  • 9. What are mixture models and how do they work? 9  GMM assumes each regime (aka component, class, state) is generated by a Gaussian process with discoverable parameters.  GMM uses E-M under the hood alternating between estimating the regime parameters (mean, variance) and likelihood that those parameters could generate the data, until convergence or another stopping criteria is met.
  • 10. Can GMMs model non-stationary data sets? 10  Using SPY return data from Yahoo Finance I fit the GMM with 3 components representing low-volatility, neutral, and high- volatility regimes.
  • 11. Can GMMs model non-stationary data sets? 11  Notice the 2th hidden state or “regime” has a negative mean return, and the largest variance, which represents our high- vol regime. The 1th hidden state looks like our low-vol regime, and the 0th hidden state is somewhere in between.  The GMM accounts for non-stationarity by design. It chops the data into regimes and provides us with an estimate of the mean and variance for each one.
  • 12. Can GMMs approximate nonlinear data distributions? 12  Short answer: Yes  The top image shows the GMM trying to fit 2 Gaussians to the data with predictably bad results.  The bottom image shows the GMM trying to fit 16 Gaussians to the data with more success.  You can approx. many complex shapes with enough Gaussian components but you risk overfitting.
  • 13. Strategy Design 13 Now that we have determined that GMMs can overcome the weaknesses of the TSA models, we can design a trading strategy to test its application:  The strategy uses a total of 4 factors to estimate the sequence of regimes and their parameters. • The asset’s returns (Yahoo Finance) • TEDRATE (FRED) • T10Y2Y (FRED) • T10Y3M (FRED)
  • 14. Strategy Design 14  The strategy works like so.  Feed market data to the GMM.  Call predict method to extract model estimate of last regime “label”.  Once we have the “label” we can get the model estimate of the mean and variance for that regime.  We feed the parameter estimates to the JohnsonSU distribution and draw N samples to construct our 99% confidence intervals (CIs).  Any returns that fall outside the estimated CIs are defined as outliers and assumed to mean revert. This strategy assumes a “buy” of the etf for M number of steps (days) after the outlier event of a return falling below the lower CI.
  • 15. Evaluate the Strategy 15  For a single ETF with a set of parameters, for the walk- forward period, we can plot the model accuracy as well as an “equity” plot.  Below is an example output for SPY:
  • 16. Evaluate the Strategy 16 To really dive in we have to analyze the aggregated results. By organizing the data properly...  We can evaluate each ETF individually across lookbacks, steps or ks  We can evaluate the ETFs collectively across lookbacks, steps or ks  We’ll look at a few examples in the following slides...
  • 17. Evaluate the Strategy 17 We can see that across lookbacks SPY, QQQ, TLT have positive median returns per event. EFA and EEM have more symmetric distributions.
  • 18. Evaluate the Strategy 18 We can see a similar distribution pattern here using the sum_ratio statistic. The sum_ratio statistic is the sum of returns > 0 divided by the sum of returns < 0. Values greater than 1 are considered a “success”.
  • 19. Evaluate the Strategy 19 In this example we evaluate the median return for each ETF by lookback and across steps. Again we see SPY, QQQ, and TLT dominating. EFA is also mostly > 0 across steps.
  • 20. Evaluate the Strategy 20 In this plot we see the sum_ratio metric is > 1 across multiple dimensions for SPY, QQQ, TLT, and EEM. GLD’s sum_ratio depends on lookback. EFA is weak, and EEM suffers from exposure at 21 steps.
  • 21. Evaluate the Strategy 21 Here we can evaluate the median returns across k (# of components). Again we see broad success across k for SPY, QQQ, TLT, and surprisingly EFA. GLD and EEM are more varied and dependent on lookback.
  • 22. Evaluate the Strategy 22 Here we can evaluate the median returns across k (# of components). Again we see broad success across k for SPY, QQQ, TLT, and surprisingly EFA. GLD and EEM are more varied and dependent on lookback.
  • 23. Evaluate the Strategy 23 Here we see a similar pattern among the ETFs with the sum_ratio for SPY, QQQ, TLT are > 1 across multiple dimensions. EEM and GLD show similar variability based on lookback but with passable performance overall.
  • 24. Conclusions 24  The secret sauce for GMM accuracy when predicting the return distribution is choosing the proper sampling distribution (JohnsonSU vs. Normal)  US based ETFs showed consistently positive results across many parameter combinations.  There was more variability with GLD, EFA, and EEM depending on metric evaluated, steps (exposure), and k which warrants an expanded study.  Must note that there is a risk that these results are simply the result of being long the US market post 2009!
  • 25. Further Areas to Explore 25  GMM with large cross section of single-issuer stocks.  GMM testing with multiple time frames (intraday, weekly, monthly).  GMM as basis for volatility strategy (testing volatility trends post outlier event.  Are the outlier events themselves randomly distributed? Or are they clustered?  GMM as a basis for multi-asset class strategy (combinations of equity, options, futures)
  • 26. Resources 26 * Introduction to Hidden Markov Models with Python, Networkx and Sklearn - http://blkarb.se/2l6Lndw * Intro to Expectation-Maximization, K-Means, Gaussian Mixture Models with Python, Sklearn - http://blkarb.se/2n1A7kn * Python Data Science Handbook by Jake Vanderplas - https://github.com/jakevdp/PythonDataScienceHandbook * Blackarbs blog – Blackarbs.com/blog * Blackarbs Github - https://github.com/BlackArbsCEO/Mixture_Models
  • 27. Contact 27 * Website: Blackarbs.com * Twitter: @blackarbsCEO * Linkedin: www.linkedin.com/in/bcrblackarbs/ * Email: bcr@blackarbs.com
  • 28. SINGAPORE 30 Cecil Street, #19-08, Prudential Tower, Singapore – 049712 Phone: +65-9057-8301 INDIA A-309, Boomerang, Chandivali Farm Road, Powai, Mumbai - 400 072 Phone: +91-22-61691400 Learn Automated Trading Join EPAT™ The Executive Programme in Algorithmic Trading Enquire Now!