EFTA00556664.pdf

4 June 2012 Wit II [email protected] Confidential EFTA00556664 Table of Contents 3 Introduction 4 Executive Summary 5 Competitive Advantages 6 Investment Process 7 Strategies Description 11 Performance 13 Appendix 33 4ARMIN EFTA00556665 Introduction Founded in October 2010, QARMIN is a small Paris-based Prop Trading House that specializes in quantitative systematic fully automatized strategies with a focus on listed and highly liquid instruments (Europe and US) and medium frequency strategies. First stage of development was dedicated to development and live-testing of our proprietary trading platform including automated strategies, backtesting tools, risk management engine and execution algorithms. We have gone live with the founders' private funds and are now looking for a business partner to operate on a larger scale. QARMIN June 2012 - Confidential EFTA00556666 Executive Summary J Primary objective is to achieve consistent risk-adjusted returns throughout different market cycles, taking advantage of a flexible dynamic allocation process relying on 3 largely uncorrelated investment axes: multi-asset directional, market neutral and volatility arbitrage. We typically target a minimum Sharpe Ratio of 2, with an annualized Return to Max Drawdown of 3. J Custom trading platform: Research and Development, backtesting and deployment run on a unified platform, hence making possible process streamlining. We're maintaining an extensive historical database of prices and fundamentals for 6000+ instruments listed on US and European exchanges, going back more than 15 years. QARMIN is a team of experienced prop traders and developers who have committed a significant amount of their wealth in the project, and have worked extensively together in the past. Each team member has a strong academic background in mathematical finance and has extensive experience in the derivatives market. A diversified strategy portfolio with high pre-leverage target returns QARMIN June 2012 - Confidential it ARMIN EFTA00556667 Competitive Advantages ❑ Strong methodology and flexible platform allow industrialization of idea generation and testing: time to market from R&D to production is low ❑ Capacity to create and manage several uncorrelated bets by creating nonstandard coherent underlyings (basket, synthetic instruments) ❑ Large existing (and growing) library of uncorrelated single models/strategies displaying low pairwise correlation (-5% average correlation over 10yr period) within each investment axis Extensive menu of allotment between each of those models benefiting from our proprietary allocation model therefore increasing Sharpe Ratio compared to any given single strategy's Sharpe Total liberty to switch off strategies with low prospective returns ❑ Active hedging method for tail risks via quantitative process involving long only cheap OTM long-term options; focus on macro picture, top down approach and exogenous stimuli analysis therefore departing from the historical bias and enhancing the capital preservation capacity of the portfolio QARMIN June 2012 - Confidential EFTA00556668 Investment Process Strategy Development (cf. Appendix p15) • Underlying idea stemming from market experience 77 Strategy • (risk premium location) Historical validation (backtesting, choice of relevant i Development parameters, observed returns) I Alpha Discovery • Data-mining bias correction (reality check, expected Historical Validation Expected Returns returns, coherence with market fundamentals — is the edge still in place?) Portfolio Allocation (cf. Appendix p16) • Bottom-up approach using customized allocation algorithms at each allocation step • Top-Down approach adjusting marginal weights of single models and investment axes via our own proprietary risk Portfolio index Allocation Bottom-Up Top•Down Risk Management (cf. Appendix p18) • Single model dedicated risk management (stop-loss, volatility/VaR adjusted size by instrument) • Specific risk manager for each investment style (directional, market neutral, volatility arbitrage) • Global aggregation for ultimate capital preservation constraint Execution Management (cf. Appendix p21) • Specific algorithms for each investment style • Transaction costs analysis (brokerage, slippage, rebates) for best execution algorithm selection and trades/performance reconciliation Identification of alpha at each step of the process QARMIN June 2012 - Confidential ARMIN EFTA00556669 Strategies Description Overview • 3 Investment axes whose relative weights are controlled via a proprietary portfolio allocation model; directional, market neutral and volatility arbitrage are effectively uncorrelated through different market cycles Investment Axis Description Asset Class Horizon Equities, Equity Index, Fixed Income, Commodity, Technical and fundamental short/medium Currency Futures and ETFs. Options on these asset term pattern From hours behavioural coherent 2 to with Directional classes. finance, market macro/micro structure and several weeks Geographic positioning: Europe, US and Emerging directional events Markets (via ETF) Adaptive statistical analysis to capitalize on Equities, Equity Index, Fixed Income Futures From 10 minutes Market Neutral local divergence and mean reversion nature Geographic positioning: Europe, US to 2 weeks of risk neutral baskets Equities: main indices and their stock components, VIX, Arbitrage of the volatility curve mispricing Volatility V2X, Main Currencies, Commodities, ETF Volatilities (local vol, skew, kurtosis, term structure) of From 1 week to VXX, VXZ, Listed vanilla options Arbitrage an underlying compared to its peers, both 1 month Geographic positioning: Europe, US and Emerging using fundamental and statistical approach Markets (via ETF) ARMIN EFTA00556670 Strategies Description Directional Directional Strategy Sample* Strategy Description Underlyings Horizon Take advantage of short term negative Large Cap Equities 1 to 5 days Mean Reversion autocorrelation using specific volatility analysis Equity Futures Exploit long term trend following nature of capital Equity Index Futures 5 days to markets; overlay via tactical options writing Commodity Futures several weeks Trend Following FX Futures Fixed Income Futures Emerging Market ETFs Capitalize on over/under-reaction of various indices Equity Index Futures 2 to 10 hours Macro Events around macroeconomic announcements Fixed Income Futures Exploit statistical patterns of gaps at open Equity Index Futures 2 to 10 hours Gap Benefit from intraday and extraday seasonality due Equities 2 hours to 5 Seasonality to structural imbalances Equity Index Futures days non-exhaustive list of currently available live strategies QARMIN June 2012 - Confidential '')QARMIN EFTA00556671 Strategies Description Market Neutral Market Neutral Strategy Sample* Strategy Description Underlyings Horizon Exploit local relative mispricing of equities and Volatility Index Futures 10 minutes to Equity vs. Volatility volatility as an asset class Equity Index Futures 1 day Arbitrage Take advantage of temporary divergence within a Equities 1 day to 2 bespoke basket constructed via various statistical weeks Equity Statistical Arbitrage methods Take advantage of temporary divergence within a Equity Index Futures 2 hours to 3 Cross Asset Futures bespoke basket of instruments from different asset Commodity Futures days Arbitrage classes constructed via adequate statistical methods FX Futures Fixed Income Futures ' non-exhaustive list of currently available live strategies QARMIN June 2012 - Confidential ARMIN EFTA00556672 Strategies Description Volatility Arbitrage Volatility Arbitrage Strategy Sample* Strategy Description Underlyings Horizon Benefit from implied volatility curve mispricing of an Equity Options 1 week to 1 Single Stocks Relative underlying compared to its peers (L/S volatility, Equity Index Options month Value Dispersion) Take advantage of large time decay effect for short Equity Options 1 day to 1 Short term options term options around expiration; identify behaviours of week expiration effect underlyings on expiration date deriving from large options positions hedging Exploit volatility curve dislocation around idiosyncratic Equity Options 1 week to 1 Event Driven events month • non-exhaustive list of currently available live strategies QARMIN June 2012 - Confidential 4ARMIN EFTA00556673 Performance 1/2 Consistent risk-adjusted returns over a 10yr backtest period with no sign of abating in the recent past Directional/Market Neutral Historical Drawdown e e Sharpe 3.92 Yearly Return to Max DD 5.34 Avg Yearly Return 22.8% Profitable Days % 59% Avg Daily Return 0.08% StDev Daily Return 0.37% Best Daily Return 2.91% Worst Daily Return -2.24% Directional/Market Neutral Equity Curve Worst DD -4.27% Max DD Duration 136 days Avg Worst 10 DD Duration 62.9 days Ct ARMIN EFTA00556674 Performance 2/2 Badctested Yearly Returns (Post-Transaction Costs, Unleyeraged) • 1 Margin of safety built into target return expectations 45% moving forward; 15% yearly 40% target performance has been reached 9 years out of 35% the last 10; it is a large 7.8% 30% (or 0.85 std dev) below 25% average over the entire period 20% 15% 15% target return is coherent with 4.27% max backtested drawdown, hence allowing us to construct a portfolio with 2007 2008 2009 2010 2011 2012 an expected Return to max drawdown higher than 3 Yearly Returns — — — Average Backtested Returns 23.5% — — — Target Return 15% Strategy can be leveraged up Y2012 performance has been frozen as of May 31,1 and extrapolated for the remainder to 5 times (margin to equity < of the calendar year 100%) to achieve an iso The backtest is presented with an equal-weighted allocation method; live allocation will return-to-max-drawdown tweak the relative weights of each strategies and each investment style, hence improving the risk profile of the portfolio; weight of any given axis will never represent couple with no significant more than 50% at any given time with the exception of Volatility Arbitrage which will be foreseeable loss of limited to 25% relative weight performance (besides Volatility Arbitrage investment axis is not represented on this chart as backtesting marginal funding costs) remains difficult to achieve; however the expected performance is in line with the above-presented result and should add an extra layer of diversification, therefore improving the overall profile of the portfolio QARMIN June 2012 - Confidential EFTA00556675 Appendix 14 Team Bio 15 Strategy Development 16 Portfolio Allocation 17 Risk Management 20 Execution Management 21 Technical Platform 28 Performance Details QARMIN June 2012 - Confidential 4ARMIN EFTA00556676 Team Bio Li Bertrand LOUVARD, Founding Partner Bertrand joined Societe Generale in 2000 as a trader on the US Indices Derivatives Trading Book in NY. In 2003, he became Head of the US Indices Flow Business Trading Book before being appointed Head of the US Equity Flow Business Trading Book in 2005. He expanded the Hedge Fund Client Business in the US for vanilla derivatives products. He then joined the Lyxor Structured Product Solutions team in 2009 to develop the structured products active management offer within the Quantitative Asset Management team of Lyxor. Bertrand Louvard is a graduate of the tcole Centrale Paris and has a Master's degree in Mathematics, Probability and Statistics from the University of Jussieu (Paris VI). Li Francois-Charles SCAPULA, Founding Partner Francois-Charles was a fellow associate at the CEREG between 2001 and 2004 and a professor of Econometrics and Quantitative Techniques at University Paris IX Dauphine. He has spent the last 7 years as a prop trader for Fortis and Societe Generale focusing on quant equity derivatives strategies, where he traded listed and OTC stock and index options, variance swaps, futures from all asset classes. He was also responsible for setting up the trading platform at Lyxor for the quant asset management department. Francois-Charles is a graduate of the Ecole Normale Superieure and ESSEC. He has an « Agregation » in Economics, a Master's degree in Applied Mathematics and a Master's degree in Finance from Paris IX Dauphine. He was a PhD candidate in Mathematical Finance at the University of Paris IX Dauphine. Sylvain Rey, Founding Partner As a graduate of Telecom SudParis with a specialty in parallel and distributed system, Sylvain has been a Software Consultant in the San Francisco Bay Area from 2000, then in Paris, France from 2003. As a seasoned Application and Systems Architect, he has designed and developed many solutions for various banking and financial institutions such as BNP Paribas, AXA Investment Managers, Banque de France and the European System of Central Banks. Each Partner has committed a substantial amount of his personal wealth to the venture. Sylvain and Bertrand have known each other for more than 15 years. Bertrand and Francois Charles were working together at Societe Generale. Sylvain, Bertrand and Francois Charles have founded QARMIN in October 2010 and have worked extensively since then to build a cutting edge systematic platform. QARMIN June 2012 - Confidential EFTA00556677  IDEA GENERATION Strategy Risk Premium Intuited Asset Class/Universe/Instruments defined Development Systematic Rule Formulated 1. Entry / Exit 2. Specific Risk Management (Size, Stop Loss, Overlay) Yes Are observed results Config coherent with market experience? Top-down process Database Selection Criterion Bottom-up process 1. Prices 2. Volumes 8acktesting Data-Mining 1 Average Return 2 Sharpe 3. Fundamentals Engine Engine 3 Ret on drawdown 4. Derivatives 5. Proprietary Data 4. Ret. on VaR Yes Observed uputrun parameters set dnu sysemauc rule ITOM weights performance set observed data sets. satisfactory? Data Mining Correction Yes SINGLE STRATEGY READY Integrate Strategy In Library °IQARMIN EFTA00556678  STEPWISE BOTTOM UP CONTROL PORTFOLIO ALLOCATION TOP DOWN CONTROL PROCESS DIRECTIONAL LIBRARY MARKET NEUTRAL LIBRARY VOL ARB UBRARY Single Strategy Generation Process Single Strategy Generation Process INITIAL STEP 0 Direct. I Direct. Direct. Neut. CI Strategy Development 1 n 1 q. Strategy Development i STATISTICAL EQUITY CURVES ANALYSIS EXOGENEOUS CRITERIA ANALYSIS Including: Including: 1. Mean Variance Optimization over 1. Strength of Risk Premia underlying various horizons each of the single strategies (e.g. IV 2. Maximum diversification under Alloc. Alloc. Alloc. %Ile vs. RV, frequency of channel STEP 1 performance constraint borders crossing, high volume in bear 3. Maximization of various performance market, etc.) measures in walk forward process 2. Liquidity of traded instruments INTERMEDIATE RESULT 1 Directional Market Neutral STATISTICAL EQUITY CURVES ANALYSIS EXOGENEOUS CRITERIA ANALYSIS Kelly Criterion, Markowitz Portfolio Including: Allocation 1. Macro/Flow Environment (e.g. Economic Indicators) 2. Volatility/Credit Conditions (VIX, iTraxx) Allocation 3. Fear/Greed Index, Deal Index STEP 2 (Proprietary index) 4. Crowding out trades (recent HF performance vs. LT mean) Decision to tilt allocation towards better historically performing strategies in identified conditions virec:Lionai volatility mroitrage INTERMEDIATE RESULT 2 aggregation aggregation aggregation EFTA00556679 Risk Management 1/2 Ex-ante and ex-post approaches Ex-ante approach ❑ Portfolio allocation taking decisions based on past performance analysis under risk constraints • Variance minimization of the portfolio for a given performance target via Var-Covar matrix for strategies equity curves and underlyings historical return • Minimum ex-ante diversification required for inclusion of a given strategy in the meta-portfolio ❑ Sentiment Index adjusting notional at risk based on fundamental and technical data keeping the margin to equity relatively constant (and targeting stable risk return profile) ❑ Hedging macro picture with "long-only" cheap long term OTM options via fundamental and quantitative screening of underlyings with available derivative markets (see focus) Ex-post approach ❑ Single position aggregation and real-time dedicated risk metrics for each business line... • Multi Directional: Tenor notional exposure (via Var-Covar Matrix for underlying return) • Market Neutral: Notional Replication, Spread Risks • Volatility Arbitrage: Volatility, Kurtosis/Skew Exposure (Tail Risks) ❑ ...and at the global level for ultimate risk control and capital preservation • Greeks, beta-adjusted notional exposure, dispersion risks ❑ Stress-testing of all positions via adverse relevant / historical scenarios, VaR, etc. QARMIN Ju 2012 - nfidential EFTA00556680 Risk Management 2/2 Macro Picture - Options ❑ « Buy only» long dated out-of-the-money options ❑ For a wide variety of asset classes and instruments, identify cheap options in terms of current implied volatility vs. universe (universe-relative cheapness at a given time) and long term average (self-relative cheapness over time) ❑ Analyze technicals, fundamentals and sentiments, leveraging our platform and proprietary database via a quantitative process that allows us to screen hundreds of assets Come up with a list of instruments in a position to capitalize on high potential global macro situations ❑ « Be fearful when others are greedy and greedy when others are fearful » (W. Buffet) • Market psychology is the motor of the performance: we try and benefit from fear, greed, hysteria and mania • Options are often mispriced during periods of irrationality hence providing great risk/reward opportunities ❑ Actively manage options • Long term only options in order to reduce adverse time decay impact: we do not keep options with time to maturity lower than 1 year • Profit-taking/Stop-loss methodology: we seek to return 5/10 times the original investment on any given bet; as soon as intermediary targets are reached, we deleverage part of the bet ❑ Risk management/performance enhancement process • Our general investment philosophy is to look for statistical anomalies and capitalize on it by designing systematic strategies; although an adaptive process, it is bound to be historically biased • These strategies are often - not always- based on cashing in risk premia (implicit or explicit) • Buying cheap OTM options allows us to hedge both biases (historical and short risk/premium) via a deductive approach that covers any unpredictable events that June fall out of historical scope QARMIN June 2012 - Confidential EFTA00556681  STEPWISE BOTTOM UP CONTROL RISK MANAGEMENT TOP DOWN CONTROL PROCESS Market Neutral Volatility Arbitrage INTERMEDIATE RESULT 2 aggregation aggregation NO CONTROL RISK CONTROL BASED ON CALCULATED Risk ex-ante has been measured and MEASURES calibrated at STEP 2 Including: 1. VaR, Capital at risk under adverse Directional Market Neutral Volatility Arbitrage scenarios Risk Engine Risk Engine Risk Engine 2. Greeks STEP 3 3. Notional replication, spread risks 4. Tenor Notional on uncorrelated instruments Each axis is managed separately as they are incurring different types of risk INTERMEDIATE RESULT 3 Global Aggregation NO CONTROL RISK CONTROL BASED ON CALCULATED Risk ex-ante has been measured and MEASURES allocated at STEP 2 Risks are aggregated at the top level to Global Risk Engine ensure global capital preservation STEP 4 INTERMEDIATE Throat Dacitinne RESULT 4 EFTA00556682  STEPWISE BOTTOM UP CONTROL EXECUTION MANAGEMENT TOP DOWN CONTROL PROCESS INTERMEDIATE RESULT 4 Target Positions SLIPPAGE & TCA FOR ALL EXECUTIONS LIQUIDITY, LEVEL 2, BOOK ANALYSIS METHODS Execution Algorithm Library 1. Simulated Analysis of book order, market 2. Real microstructure, number of trades at the Directional Market Neutral Volatility Arbitrage bid/offer over last relevant bar We allocate more to the best recent library library library execution method while keeping minimum diversification. STEP 5 Those measures give us a hint as to what the street is doing in regards to our trading signals. Combination of algorithmic orders for single instruments Pair trading and basket algorithms Aggressive vs. passive algorithm for directional trades FINAL ORDERS EFTA00556683 Technical Platform Architecture and flows overview Prime Broker Custom API / FIX FIX Monitoring and Reporting UHF Trader Engine tools Q Strategy Packages Reporting Database L- I IIIICLIdSt QARMIN Trading Framework fi 11 Pricing, Corporate Actions and Real-time Data Aggregator Fundamental Data Aggregator I QARMIN I I Bloomberg Activ Financials Additional Providers Custom indicators Pricing Real-time data Real-time data Fundamental data Corporate Actions QARMIN June 2012 - Confidential ARMIN EFTA00556684 Technical Platform Features 1/2 ❑ Custom integrated Platform • Fully mastered internally • Modern software methodologies • Every team member is a developer • Code base is managed, factorized and peer reviewed • Fast development cycles using agile methodologies ❑ Based on QuantServer / QuantOffice from Deltix La • Well renown software, powering various Institutionals, Funds and Proprietary Trading Houses C • Complex Event Processing architecture a) • Heavily parallelized, high performance system a • Modern foundation with highly optimized, managed processes • Tight technical partnership with Deltix ❑ Modern tools for business development • Instrument / Universe / Calendar Managers • Strategy Manager • Integrated Development Environment based on Microsoft Visual Studio • Visual Alpha tool for fast prototyping • Strategy Runner for backtesting • Database administration tools QARMIN June 2012 - Confidential EFTA00556685 Technical Platform Features 2/2 ❑ QARMIN interoperability layer • Flexible architecture allowing us to plug-in within various environments • Real-time data connectors • Pricing, corporate actions and fundamental data connectors • Proprietary engine for index and scoring indicators generation ❑ QARMIN Trading Framework • Extensive financial and mathematical library La • Integrated with well established Econometry, Statistical, Financial and Solver libraries. C • QARMIN own library with custom indicators a) Signal Instrument facility, providing directly usable meta instruments for signal processing (chain a • management, corporate action adjustment, pairs, baskets) a • State-of-the-art Meta Models facility for static and dynamic discovery/aggregation/allocation of strategies • Real-time Risk Engines, enabling proactive decisions • Order Processors, with custom execution algorithms and operational costs management ❑ Monitoring / Reporting tools • Integrated within Hyperic HQ monitoring suite / alert center • Real-time Trading Console • Reporting engine with realtime charts • Audit trails / logs QARMIN June 2012 - Confidential EFTA00556686 Technical Platform Database Features • a• . o .1 • if . •• an. !tan • OOOOO MOO 0.111S1 0n. kora Wm. ••••11•1 no •••••1110.1 %PA.CS Oa S ~~~~¢~~4~44#~~i~#~~~4~~49~~ SU. 41.0MOM In •••••••••• ann. •• am .....• 40 C... MN) SI•C MMUS Q. s.e..•••••,.•fie SC ba a MW ••••••abn, atnall.PD arl. lienem AMA ateil•t tab 0.01•01.10) 1••••••af ntall kti. Ill yea ••••• WO 0.01.1a MI •14.1.•••• 001.21.... M... 'VOW M...... In .11111an la.01.4a USSR ha Ma. ALS.I. S W Swat UM a a Ya a MS al449.14•1•••••• Jana .."1 W. ta 0