CCTrack Solutions White Paper A New Age for Quantitative Trading

Transcription

1 CCTrack Solutions White Paper A New Age for Quantitative Trading US ending QE policy will change the environment for Systematic vs. Discretionary traders as cash and volatility become more important Size of Quant Funds >$100mn but <$1bn appears to be optimal for returns Risk management stands out a key factor for determining success Market environment and leverage are the main drivers for failure Despite a long history of admirable performance, the alternative investment world has shunned quantitative trading since There were logical reasons for this shift poor returns, investor sentiment favoring physical assets as a hedge against another possible financial meltdown as soon as central bank intervention was lifted, weaker links to traditional macro signals and high asset correlations driving down diversification. In fact, factor analysis trading has failed in the last 5 years precisely because of the intervention of central bankers to boost markets and economies. The FOMC Quantitative Easing programs led to a significant shift in risk appetites as the money the Fed put into the market pushed out duration and credit risks further and faster than traditional business cycles would deliver. As is the goal of any policy easing, the QE policies of the FOMC, BOE, BOJ and now the ECB have shifted demand forward. When economies return to normal or collective fear abates sufficiently, many of the relationships previously understood to work are expected to return. As the US exits this strategy, many see the more traditional links between bonds, stocks, FX and commodities reigniting. Further, fresh investments in big data analysis, faster technology, more e- trading platforms and higher volumes have brought new strategies to old trading styles. The chart to the right highlights the troubles of Systematic CTA and Macro CTA returns over the last 5 years (with the data from Hedge Fund Research and its indices as a comparison). What 1

2 this chart doesn t show is the dramatic turnaround in both Systematic CTA and Macro CTA returns in 3Q as it shows 2014 Year-to-date rather than monthly returns for September Macro CTA HFRX index was up 1.54%. 1.21% in August and 0.52% in July while the Systematic CTA Index was 0.24%, 1.55% and -0.35% respectively. The bottoming out of returns in those styles was notable given the sharp increase in volatility during the same period. In comparison, the Global HF index was down 0.77% in September, up 1.09% in August and down 0.88% in July while the RV Value Volatility Index was up 0.69% in September, up 0.63% in August and up 0.25% in July. Outflows of capital have hit quant funds hard, as returns haven t been stellar in recent years. According to evestment data, there is about $122 billion invested in quant funds around the world, down from a peak of $198 billion in If quant funds are coming back into favor as a result of these recent positive trends, it will take some time for funds to flow back into them. The role of high frequency trading, and where it fits into quant funds strategies, adds to regulatory confusion for systematic trading as well. Even more to the point, quantitative funds prominence as a part of the total alternative universe has shrunk, and will take time to regain pre-2008 levels. According to BarclayHedge, total assets under management for the hedge fund industry was $2,353bn as of 2Q There was $320bn in the managed futures (CTA) community, with discretionary traders making up about 66% of that total, matching the evestment figures. Some of the rise in CTA assets against the overall pool in was explained by credit as CTA trading is about managed futures rather than traditional fund investments. The role of bank credit allowing leverage as a key factor in systematic trading will be something to consider as one driver for turning the market environment towards quantitative strategies. 2

3 There are notable differences between discretionary and systematic funds. Traditional arguments for quantitative funds come from their ability to trade without emotion, to trade consistently across time zones watching markets for every opportunity, diversification of styles and the ability to look at large data sets. But models do no always perform as expected, particularly when policy regimes shift. Correlations move rapidly and many trading models don t work when they shift. Rule-based trading systems don t understand the context of policy and the subtle moves made by central bankers and politicians as they manipulate markets. Many quant funds are static and have trouble evolving with new markets and rules. Evolutionary learning can be difficult for models. Opportunistic trading styles are difficult to model and fit into a broad and consistent strategy. On balance, the style difference between systematic and discretionary trading appears even out and returns appear dependent on the market state of the time. However, managers that consistently prove they have an edge quickly redesign their models to accommodate government policy shifts. This approach and taking a long term investment stance should result in performance evening out over time, albeit with different volatility and drawdowns. 3

4 The Barclays Hedge Fund Discretionary Index compared to the Systematic Trader Index from 1987 to 2011 highlights that the two trading styles get to roughly the same return over longer periods, albeit with periods of divergence and re convergence over shorter ones. The academic literature and the press about hedge funds has turned decidedly negative as a whole and the quantitative subset has been chastised further for its association with high frequency trading. But there are some key points to make about the long term return differences between systematic and discretionary funds even within the CTA community. According to a 4

5 CME-funded 2011 study from Imperial College by Julia Arnold, Robert Kosowski and Paolo Zaffaroni, the average life span of a CTA systematic fund is 12 years while that of a discretionary one is 9 years. Their survival rate for both styles has a failure rate of 11.1% annually and an attrition rate of 17.3% near that of all hedge funds but if you exclude smaller funds (under $10mn) you begin to see some notable differences between systematic and discretionary with systematic failure rates of only 4.1% and attrition rates of 7.8% compared to discretionary rates of 5.9% and 10.8%, respectively. Excluding smaller funds also highlights that the larger systematic funds have the highest probability of survival. What makes a quantitative fund successful? Are their logical components needed to build out a systematic trading fund to ensure survival and positive returns? These questions follow the academic debate about trading styles. There are 4 key parts to systematic trading fund success: the portfolio construction process, the risk management, the position sizing and the research 5

6 behind the trading. These components matter regardless of the trading strategies. There are a number of diverse trading strategies typically used by systematic funds, including: trend following, mean reversion, carry, pattern recognition, correlation, volatility, factor analysis, relative value and market-making. It s important to highlight that high frequency trading may employ many of these parts but that its main focus has been on the execution and market making function, while other larger quant funds lean on slower models like trend and carry for their returns. The portfolio construction process in quant trading begins with the target volatility of the fund. Mixing in a series of models requires some method of measuring their diversification to the entire portfolio along with its effect on the information ratio and Sharpe ratio of the returns. Many funds find that their biggest issue is in measuring the capacity of the model to the real market. The ability for portfolios to remain robust requires a deep understanding of the liquidity and the trading costs factors that show up in the Position Sizing components. Cost of trade analysis, the duration of the trade and the risk/reward methodology used for every model matter in the 6

7 input for making a model successful in a bigger portfolio. As a systematic fund becomes more successful in raising capital, the AUM becomes a drag on the ability to find returns. A Preqin report in May 2014 found that fund size made a significant difference on returns in 2013 funds below $100mn averaged 11.45%, big funds with $1-5bn average 12.08% and the best returns were for $100-$500mn at 13.79%. Furthermore, funds between $500-$1bn had many of the same characteristics as larger groups but with less negative outcomes and their returns tended to be larger, with a 13.71% average. This study highlighted the willingness of new investors into new funds of the right size making it a Goldilocks market for AUM. The nimble start-up is just part of the story, however, as many funds over $5bn are closed to investors. The right size for a fund may be a far more important discussion between investors and manager. The risk management function of systematic funds is a key driver for success, as it enables them to manage across models consistently while discretionary funds require a larger and more varied approach. Both strive for the same goals limiting volatility and drawdowns while fine-tuning leverage. The most difficult part of systematic trading strategies is in the stop-loss function something that discretionary traders have to use while many quant funds refuse to consider it. Stop losses are instead replaced by shifting models or groups of strategies. The end result is that many quant funds hit a wall of bad returns before shifting out of models while discretionary traders have a more forced exit. Both styles suffer the more difficult task of getting back into the market. V-shaped recoveries in markets are notable tests for model funds and discretionary traders that use stop loss functions. Those that want to avoid this are forced into research for regime shifting, seeking new data and new ways of thinking through this problem. In the last 5 7

8 years, many have moved into social media or broader economic data sets for answers things like the Billion Price Project or surprise indices for economics across the globe. The robustness of the solution for getting a better model has one simple pitfall data fitting. The more complicated the model, the more work that needs to be done to find out-of-sample vs. in-sample data. Getting the noise/signal search right remains one of the goals of research, but it starts with both the quality and frequency of data being collected. Given that many of the strategies in quantitative funds are similar to one another, there has to be a larger reason for why systematic funds fell out of favor and are now beginning to turn back. Mood swings are about behavioral finance more than inventories or rate policies. Yet, this cycle for investors may reflect the macro market components of policy at the zero-bound. The cycle may also be about dispersion of policy responses to economics as seen by governments and central bankers. The environment in which you trade isn t something that one model can predict. Judging which asset classes are in favor and out of favor requires complicated understanding of how a global marketplace works. If you study the asset class returns over the last 5 years, you will see one common theme cash isn t paying and remains near the bottom of the return lists. 8

9 This analysis from the Novel Investor gets across a key point about risk management styles and CTA systematic returns cash as a tool in stepping out when markets don t function as models predict has no return and, as such, drives down those that have riskier assets to chase with different models. Further, the chart highlights the tail risk events that 2000 and 2008 generated in portfolios, but with one caveat the QE policy reaction prevented a larger and more extended period of de-risking in markets. Perhaps this is the bubble that pops in the post-qe months and years ahead. The return of volatility in 2014 may be the most significant indicator that something substantial has occurred across markets, and this could bring about the needed mix of the right trading environments for systematic funds to prosper. There are, however, a number of micro market structures that have shifted since the last time we saw significant cross-asset class volatility 9

10 namely the size of the markets, the speed of the markets, the globalization and regulation of them and the amount of leverage allowed to trade them. These factors are all part of the puzzle in predicting the future success of quantitative strategies. The chart from Russell Investments describing the recent typical range (up to 3Q 2014) against the historical one and where we are now in terms of returns is another way of looking at the present market environment and the rise in volatility what seems clear is that we are far from an abnormally high state of risk. This will force a rethinking of trading styles where cash still isn t a viable alternative but where defensive rotational plays stand out. 10

11 Conclusions: Quantitative funds have much like cash suffered disfavor but the recent rise in volatility caused by the larger fundamental divergence of policy paths from global central bankers should result in a rich environment for typical systematic traders with trend, carry and volatility strategies all returning to lead over more discretionary styles of defensive rotational trading. The risk for quantitative traders rests with the battle over regulation and how the future microstructure for markets develops, with the role of credit disintermediation balanced against that of less bank support and leverage. The largest average fund returns in 2013 were observed in the $100mn to $1bn AUM window, and many systematic funds will find that they need to get to the critical mass of AUM fast in order to survive in the new Quant Fund 2.0 environment. The role of investors and their advisors in picking out the best new funds will be critical in keeping the new styles from being lost to an old cycle. End. 11

12 Disclaimer This report has been provided exclusively for informational purposes. The information contained herein does not constitute an offer to sell securities or a solicitation of any offer to buy an interest in any security or investment managed by CCTrack Solutions, LLC (hereinafter CCTrack ) or its affiliates or any other product or service to any person in any jurisdiction where such offer, solicitation, purchase or sale would be unlawful under the laws of such jurisdiction. Accordingly, CCTrack and its respective advisors, agents, affiliates, partners, members or employees will not be liable for any direct or indirect or consequential loss suffered by any person as a result of relying on any statement in or omission from the information contained or alleged to be contained in this report. Access to information about specific products are limited to investors who, among other requirements, qualify as "accredited investors" within the meaning of the Securities Act of 1933 as amended, qualify as qualified eligible persons under CFTC Regulation 4.7, or who meet any other eligibility and investment requirements and generally are sophisticated in financial matters, such that they are capable of evaluating the merits and risks of prospective investments. The information contained in this report is for the confidential use of only those persons to whom it is transmitted by CCTrack and its affiliates and may not be reproduced, distributed to others, or used for any other purpose without the prior written consent of CCTrack. By accepting delivery of this report, you agree to maintain the confidentiality of the information contained herein and agree not to reproduce or distribute such information to any other person or use such content for any purpose other than informational purposes. In doing so, you also acknowledge, represent, warrant and agree that you are an authorized recipient of this document who is permitted to receive this document under applicable laws and regulations and that you have a pre-existing relationship with CCTrack. Statements contained in this report that are not historical facts are based on the current expectations, estimates, projections, opinions and beliefs of CCTrack. Such statements involve known and unknown risks, uncertainties and other factors, and undue reliance should not be placed thereon. Additionally, this report may contain forward-looking statements. Actual events, results, or actual performance may differ materially from those reflected or contemplated in such forward-looking statements. An investment of this nature would be highly speculative and involve a high degree of risk, including risks related to lack of liquidity, changes in economic conditions, institutional risks, and lack of operating history. The use of leverage is capable of amplifying the effect of any such risk. This report does not contain a complete list of the risks and other important disclosures involved in investment, and is subject to the more complete and specific disclosures contained in relevant offering documents. Profit is not guaranteed, and there is always the potential for loss an investor may lose all or a substantial portion of its investment. This report does not take into account the investment objectives, financial situation or particular needs of any recipient hereof and should not be construed as legal, tax or investment advice. Before making any investment, an investor should thoroughly review relevant offering documents with the investor s financial, legal and tax advisor to determine whether an investment in the hedge fund is suitable for the investor in light of the investor s investment objectives, financial circumstances and tax situation. 12