DEEP LEARNING FOR LONG-TERM Jonathan Masci VALUE INVESTING - PowerPoint PPT Presentation

DEEP LEARNING FOR LONG-TERM Jonathan Masci 
 VALUE INVESTING Co-Founder of NNAISENSE 
 General Manager at Quantenstein

COMPANY STRUCTURE Joint Venture between 
 ‣ Large-scale NN solutions for ‣ Asset manager since 1994 superhuman perception and ‣ Value philosophy motor control ‣ ultimate goal of marketing AGI ‣ Funds outperform on the long ‣ leverages 25-year track record run of IDSIA , one of the leading ‣ AuM 3.7bn EUR (Feb. 2017) research teams in AI: ‣ recipient of the NVIDIA AI pioneers award

EVOLUTIONARY-RL DEMO Learned parking behavior at NIPS conference Learned behavior from driver perspective RL to the real world 
 Without a teacher, no supervision

WHAT WE DO ‣ Fully automated portfolio manager ‣ Long-Term Vision ‣ Build custom portfolios directly from fundamental data ‣ No human in the loop: ‣ Deep Learning and Reinforcement Learning ‣ Less biased

MAJOR DIFFERENCES BETWEEN FINANCE AND OTHER DOMAINS ‣ Rules of the game change over time : how to avoid forgetting what worked and not mixing things up? ‣ Lot of “ state aliasing ”: similar market configurations lead to opposite developments, state is only partially observable ‣ Limited history , and only one history ‣ No clear single objective, not as simple as classifying cats and dogs ‣ Rules for neural network design don’t transfer to finance as straightforwardly as it may seem

KEEP A LONG-TERM VIEW ON THINGS

DATABASE OF ‣ LSTM, CNN, etc. FUNDAMENTAL DATA ‣ What supervised signal to use, 
 and how to optimize for it? PREPROCESSING SINGLE INSTRUMENT AI ALPHA GENERATOR STOCK PICKER MODELS SUPERVISED SIGNAL

ARE WE REALLY IN THE BIG DATA REGIME? ▸ Data : 10K companies, 20 years, new signal every month ▸ 240 data points per company, 2.4M data points in total ▸ Using sequences reduces the number of samples, what’s the sweet spot? ▸ Only one history and the rules of the game change over time ▸ Data augmentation : ▸ If good prior, one can try to augment the training data ▸ In finance if you have a good prior you don’t need AI

BACKTESTING 300 Expected 250 200 Real 150 100 Training Testing 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 WALK-FORWARD TESTING

300 WFT Step 1 250 200 TRAIN 150 100 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 300 250 200 TEST 150 100 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016

300 WFT Step 2 250 200 TRAIN 150 100 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 300 250 200 TEST 150 100 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016

300 WFT Step 5 250 200 TRAIN 150 100 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 300 250 200 TEST 150 100 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016

WALK FORWARD TESTING ▸ Tries to minimize “double dipping” as much as possible ▸ Can involve training a very large number of models ▸ e.g. monthly retraining for 10 years produces 120 training stages ▸ Tradeoff between retraining periods and target horizon not easy to determine, many models will have to tick at different time-scales

PLENTY OF DATA WHEN GOING END-TO-END ▸ Given a set of companies and their corresponding series of fundamental data produce a set of portfolios, optimized over a given time horizon, that maximize criteria such as SharpeRatio and InformationRatio ▸ Select a random start date ▸ Select a sub-universe of K companies out of the N ▸ this gets us a choose ( K , N )- fold increase in the amount of data ▸ Issue with current systems is that they try to get alpha from fundamental data, what we want is conditional alpha . No prior on what is a good signal to be extracted, the system implicitly learns features that work for portfolio construction. This is the foundation of Deep Learning

Universe of companies DATABASE OF FUNDAMENTAL DATA FEATURES 0 FEATURES N Company 0 Company N PREPROCESSING PREPROCESSING AI RISK PORTFOLIO AI ALPHA AI ALPHA CONSTRAINTS GENERATOR GENERATOR BUILDER 
 LOSS FEATURES 0 FEATURES N Optimized portfolio No supervision on what signal to extract

SYSTEM TRAINING

Each EXPERIMENT INSTANCE 
 runs a full WFT training EXPERIMENT INSTANCE GPU#0 EXPERIMENT RESULTS DATABASE INSTANCE FRONTEND EXPERIMENT REPORTING AND CONFIGURATION GPU#1 ANALYSIS MANAGER Pool of experiments 
 EXPERIMENT scales linearly with number 
 INSTANCE of GPUs, but no speedup 
 GPU#N for single experiment

Each WFT step runs on a 
 separate GPU in a 
 MAP-REDUCE fashion WFT STEP 0 GATHER RESULTS AND PACK THEM INTO RESULT OBJECT GPU#0 WFT STEP 1 FRONTEND EXPERIMENT REPORTING AND GPU#1 ANALYSIS Experiment execution 
 WFT STEP T scales linearly with number 
 GPU#N of GPUs.

RESULTS ANALYSIS AND VISUALIZATION

Cumulative Performance Outperformance Heat Map

Performance Heat Map Rolling Performance

BAYERNINVEST ACATIS 
 MSCI WORLD INDEX KI AKTIEN GLOBAL 50 1654 #positions 251.4% 104.7% Performance 12.0% 6.7% Performance p.a. 13.9% 13.0% Volatility p.a. Return/Volatility 0.9 0.5 5.3% — Outperformance p.a. 1.0 — Information Ratio -49.1% -48.5% Maximum Drawdown 2.5% 2.4% Dividend yield 12M 1.92 1.22 Calmar Ratio L36M

MASTER FACTS Investment Company BayernInvest, München Custodian BayernLB, München Manager ACATIS Investment GmbH, Frankfurt AI Model Developer Quantenstein GmbH, Frankfurt ISIN DE000A2AMP25 (Institutional class) Bloomberg Ticker BIAKIAK GR Equity Minimum Investment 50,000 Euro (institutional class) Investment Focus Equity Global Domicile Germany Currency EUR Benchmark MSCI World NDR (EUR) Inception March 23rd, 2017 Fiscal Year-End Dec. 31st Front End Fee Max 5% Ongoing Costs 1.03% Performance Fee At present, starting at 3% outperformance 25% of yield generated by the fund during the settlement period is above the reference value MSCI World NDR (EUR). Permission for Public Distribution D Distribution Distributed

DISCLAIMER ▸ This document is only intended for information purposes. It is solely directed at professional clients or suitable counterparties in terms of the Securities Trading Act, and is not intended for distribution to retail customers. ▸ Past performance does not guarantee future results. Quantenstein accepts no liability that the market forecasts will be achieved. The information is based on carefully selected sources which Quantenstein deems to be reliable, but Quantenstein makes no guarantee as to its correctness, completeness or accuracy. Holdings and allocations may change. The opinions promote understanding of the investment process and are not intended as a recommendation to invest. ▸ The investment opportunity discussed in this document may be unsuitable for certain investors depending on their specific investment objectives and depending on their financial situation. Furthermore, this document does not constitute an offer to persons to whom it may not be distributed under the respectively prevailing laws. ▸ The information does not represent an offer nor an invitation to subscription for shares and is intended solely for informational purposes. Private individuals and non-institutional investors should not buy the funds directly. Please contact your financial adviser for additional information. The information may not be reproduced or distributed to other persons.

THANK YOU