An Introduction to Alternative Beta and CFM Isdiversified

Discussion Note

March 2016 AN INTRODUCTION TO ALTERNATIVE BETA AND CFM

ISDIVERSIFIED

Contact details

Call us +33 1 49 49 59 49 Email us [email protected]

www.cfm.fr CFM An introduction to Alternative Beta and CFM ISDiversified

CFM has 25 years of experience in researching financial Introduction markets and developing quantitative strategies. The purpose of this note is to introduce our approach to Alternative investments, in aggregate, have disappointed a Alternative Beta investment, addressing the significant number of investors because of a failure to following themes: provide diversification when it is most needed. Hedge funds have typically overcharged and underperformed.  We define Alternative Beta strategies as a systematic Undeniably, many have provided a boost to risk-adjusted approach to capturing persistent behavioural biases performance, by being decorrelated over the long term to and alternative risk premia, which often make up the the equity beta that dominates most portfolios. Yet building blocks of hedge fund strategies. crucially they have not been decoupled with negative  We outline key selection criteria for the inclusion of equity performance when it matters the most; i.e. when individual strategies in the program, noting that equity markets experience deep drawdowns due to their persistence and statistical significance are more inherent negative skewness. In aggregate, hedge funds important than recent (presumably favourable) appear to deploy capital into strategies that are negatively performance. A careful selection process is a critical skewed, such as event driven, fixed income arbitrage, credit, part of the mandate of an Alternative Beta manager. fundamental equity long-short, and equity momentum. The six initial strategies of CFM ISDiversified are briefly presented in this context. What makes this such a common experience? Investors, in  We discuss implementation practices that we believe their hunger to diversify, have pushed excess capital into to be crucial in effectively exploiting strategies with seeking alpha, and managers have responded by relatively modest Sharpe ratios. As Alternative Beta allocating this excess capital into risk premium strategies. strategies are often well-known, we see implementation Alpha exists, but it is rare and expensive to access, and its as a key differentiator between managers. capacity is very limited.  We close by examining some of the benefits of a Investors are seeking two kinds of returns, which they have diversified Alternative Beta portfolio. Specifically, we called beta (i.e., returns from more traditional markets) show that diversifying across strategies improves both and alpha (which they define as anything that is different the risk-return and the skew-return profiles of the from beta). Unfortunately, what has been called alpha is program, and that as a result of low correlation, frequently only beta-in-drag. So-called alpha seems to fall allocating to a diversified Alternative Beta program into three rather messy categories: (1) true alpha, which we significantly enhances the experience relative to a will propose to define as a non-negative skew strategy traditional balanced, or 60-40 benchmark, portfolio. with zero correlation to traditional markets and a Sharpe We constrain ourselves to two Greek letters – β (beta) and α ratio exceeding one; (2) better beta, which delivers a (alpha) – and provide a short glossary of lesser known Sharpe ratio of (say) between 0.5 and one, with low terms at the end of the note. correlation to traditional markets, (though in some – but not all – cases with a negative skew); and (3) worse beta, meaning that, in general, you’d be better off without this stuff. Alternative Beta

We think the better beta component is worth talking about. It diversifies; it is scalable. At two-and-twenty, we A few words on a think it is overpriced. Assuming you can get it at the right price – well-implemented, risk controlled (remember that (relatively) new concept potential for negative skew!), and with a lot more transparency than it often comes with – it makes a An introduction to Beta and Alpha lot of sense.

It makes sufficient sense that we are proposing to treat this as a separate class of investments. We call it In defining Alternative Beta we will begin with a short Alternative Beta, and we think that with the right discussion of our first Greek letter, beta. When combining implementation, appropriate infrastructure, and a rigorous two investments in a portfolio, a serious investor estimates research process it can be delivered to investors at more the correlation between them and accounts for it when appropriate fees. making a decision to buy or sell. An investment in a fully correlated product gives no additional diversification in financial risk and hence does not improve risk-adjusted

www.cfm.fr 02 CFM An introduction to Alternative Beta and CFM ISDiversified

returns. Beta can be considered a type of correlation and the all-important (in the context of this note) measure between a strategy and a traditional benchmark Alternative Beta. investment in, say, equities or bonds. Beta is most While a strict definition of Alternative Beta strategies commonly associated with equity indices, meaning a remains elusive, important themes tend to emerge: stock’s beta is a reflection of its correlation with the index. If the beta of a given stock is zero, it is decorrelated from  Pricing inefficiencies that persist over long periods of the index, while if it is positive or negative, the correlation time – decades or (when possible to measure) with the market is correspondingly positive or negative. centuries. Levels of statistical significance need to be Beta is therefore crucial in judging how much extra sufficiently high to establish strong conviction that the diversification can be obtained with the inclusion of a systems are not over-fitted but well anchored in the stock to a given equity portfolio. structure of the markets.

We now turn our attention to the second Greek letter  Strategies that are slower moving, rather than short common to financial parlance, alpha. As stated above, term inefficiencies that are generally arbitraged away beta is a measure of correlation with a benchmark. Alpha through time. As a result of a lower sensitivity to on the other hand is a measure of excess return generated trading costs, these strategies tend to be scalable to above and beyond the traditional investment benchmark, higher capacity. i.e. irrespective of it. Positive alpha represents the added  Strategies that are explainable, understandable and value of an investment, the extra juice independent of plausible. We think these strategies subdivide further into everything else, added to a given portfolio. Alpha is seen as two types – risk rewarding, or so called risk premia [1]; and the far more desirable and elusive of the two Greek letters. pure market anomalies, often of behavioural origin.

To bring further clarity to the discussion let us now think in In short, a diversified Alternative Beta program is a mix of terms of exposure to any particular benchmark. conceptually straightforward, plausible strategies, seeking Alpha should be considered as the exposure to to deliver persistent excess returns with scalable capacity, non-benchmark-related risk, while beta is the exposure to while exhibiting low correlation to traditional equity and the benchmark, the sum of the two terms encompassing fixed income benchmarks. There are numerous such all the risk carried by the target investment. Having a high strategies described in both academic and financial beta relative to alpha means we have no protection literature. The next section addresses the selection against market moves whereas high alpha and low beta question, and introduces six initial strategies chosen for means little exposure to market moves and a return the CFM ISDiversified program. stream independent from the market. Selecting relevant Alternative Beta: our definition strategies

There are some strategies that were previously only Statistical significance is crucial… employed by hedge funds but which have now made their way into more mainstream investment circles. One example is trend following, for which techniques are Of primary importance when selecting a strategy for an commonly known. These investment styles continue to be Alternative Beta program is its persistence over time. known as Alternative for historical reasons and because Typical strategies employed in the program tend to trade the strategies are often decorrelated from equity and slowly (so as to reduce costs) and, at least on a stand-alone bond benchmarks. Alternative Beta is then an investment basis, exhibit modest Sharpe ratios. As such, the challenge style which is correlated to these off-the-shelf, well-known of choosing strategies becomes an issue of statistical strategies but delivers no alpha, i.e. no excess return above significance. With lower Sharpe ratios and fewer trades and beyond that generated by the alternative benchmark. comes the need to test over multiple decades or, where The definition of the alternative benchmark is not as clear possible and relevant, centuries. In this way we build cut as it is for equity or bond indices, but we will assume it confidence that these strategies are exploiting effects that to be a standard implementation of a standard alternative have persisted over long periods of time, and we maximise strategy. The more mathematically minded reader is the likelihood that these effects will continue to exist and directed to the back of the note for a discussion of how persist in the future. this is more formally broken down between the various contributions of pure alpha, traditional benchmark beta

www.cfm.fr 03 CFM An introduction to Alternative Beta and CFM ISDiversified

…and is the main criterion for strategy inclusion1 companies based on the analysis of fundamental data like earnings, accruals and cash flow measures, independent of market price. We began Any measurement of the Sharpe ratio of an investment trading EMN quality strategies in 2006. contains measurement noise or statistical error which, as  A Long Term Trend Following Strategy seeks to one might expect, decreases with the length of the back- participate, via the futures markets, in both up and test. Of more importance than the Sharpe ratio itself in down trends, on an equal risk basis across stock establishing persistence is to ask how significantly the indices, fixed income, currencies and commodities. observed Sharpe ratio has deviated from a Sharpe ratio of We have back-tested this strategy over more than 200 zero (what statisticians call the null hypothesis). The years, and have deployed trend strategies since 1991. t-statistic (or t-stat) is a variable that measures this. The  Two Risk Premium Strategies t-stat is the product of the Sharpe ratio and the square  FX Carry Trade profits from the interest rate root of the back-test length in years, and it represents the differential between high and low-yielding number of standard deviations that the Sharpe ratio has currencies at the expense of an exchange rate risk. deviated from zero: the higher the t-stat, the more The program currently trades eighteen currencies; significant the result. A t-stat of three is considered the final portfolio consists of (192-19)/2=171 currency relatively significant with an approximately one in 370 pairs. We first traded the FX carry strategy in 2005. chance of choosing an attractive strategy through luck  Short Volatility captures the so-called volatility rather than skill. The precise details of the t-stat are less premium that market participants are consistently important here than to note that a value over three is, for willing to pay when they seek protection by buying our purposes, statistically significant. This is the primary options. The strategy maintains a constant and statistical tool applied to the decision to include a diversified vega risk by selling delta-hedged particular strategy in the CFM ISDiversified program. options across four asset classes: stock indices, fixed income, currencies and commodities. We Our current strategy selection have been trading volatility since 2005 and first deployed a short volatility strategy in 2009.

From a universe of many candidates, we selected an initial set of six strategies to launch the CFM ISDiversified Implementation program. Each strategy has a t-stat in excess of four, and has been employed in some form in our alpha programs across the 25 year history of the firm. Over time, relying on The devil in the detail the same selection criteria, we expect the count to increase. Each of the initial six is described briefly below. The implementation details of these strategies are all-important: transitioning from a paper traded back-test  Three Equity Market Neutral (EMN) Strategies applied to a real portfolio involves several key ingredients. Because to a trading universe of more than 3,500 single stocks Alternative Beta strategies tend to be both well-known from global markets. and frequently common across different managers’  Momentum uses a price history of stocks, and portfolios, we believe that the added value is in the quality relies on the insight that past outperformers tend of a manager’s implementation. to continue outperforming and past In this section we illustrate with a few examples how our underperformers tend to continue to underperform. implementation of these conceptually straightforward We began trading EMN momentum in 2001. strategies differentiates CFM ISDiversified in the  Value exploits the well documented tendency of Alternative Beta space. undervalued (overvalued) stocks, based on a range of price-to-fundamentals ratios, to outperform (underperform). We first used fundamental data to deploy EMN value systems in 2005.  Quality assesses the likelihood of future outperformance or underperformance of

1 For the sake of brevity here, we do not cover the idea of plausibility – how well we can understand and explain a particular strategy, an important qualitative, though secondary, consideration for strategy selection.

www.cfm.fr 04 CFM An introduction to Alternative Beta and CFM ISDiversified

Portfolio construction liquidity of instruments, consistency between traded volumes and expectations, size and speed of orders, margin usage, and a host of other metrics. Portfolio construction, alongside return generation and execution, is a vital aspect of building a robust investment program. Our equity market neutral portfolio provides an Industrialized, large scale data processing interesting, if somewhat complex, example. We have developed statistical methods to identify the primary Our data team of approximately 20 IT engineers is market factors (often made up of combinations of sectors) responsible for collecting, cleaning, manipulating and that need to be hedged out, in order to extract returns managing terabytes of incoming data every day. We thrive from stock forecasts while avoiding exposure to large, on new data forms as input to systematic trading ideas. collective market or sector moves. Correlations between We also rely on data to simulate the whole production stocks are noisy, meaning that the real correlation is chain from end to end, to ensure that positions and masked by measurement error arising from a lack of data performance are in line with back-tests. Data sets need to with which to measure it. It follows that relying on simple, be as broad as possible to inspire new ideas, but must also naive measures can lead to a poor estimate of portfolio be reliable and readily accessible to the research team. risk. Our methodology (described in greater detail in a technical note [4]) enables us to better distinguish real The data we collect ranges from market information (e.g., correlations from erroneous ones, tainted by prices, implied volatilities) to fundamentals (e.g., corporate measurement noise. These techniques have been financial statements) and non-financial information for generalised and are used across many portfolios at CFM. trading ideas that exploit idiosyncratic, market-specific inefficiencies. Given the lower turnover of Alternative Beta strategies, achieving statistical significance is facilitated by Systematic risk management, built on a the longest back-tests possible. Time-series data for bespoke global IT infrastructure futures and equities extend back minimally to the 1960s and 1970s. Where possible, we maintain data series as far The IT infrastructure at CFM, built over more than two back as 1800 (e.g., monthly data for many indices, decades, has been specifically designed for quasi real time commodities, bonds and various interest rates). We risk evaluation. The systems estimate and monitor risk at currently monitor approximately one million instruments regular intervals based on current positions and market in real time. conditions, enabling us to deliver a constant, targeted We limit third-party risk and improve the reliability of our volatility at both the strategy and the program level. The data processes by engaging multiple data providers. Our portfolios within CFM ISDiversified tend to turn over slowly; systems operate via seven redundant data centres located however, regular updates of positions and market across Asia, Europe, and the Americas. condition indicators are important in building a reliable risk control infrastructure. This robust, internally built, global technology network runs around the clock, across The importance of minimizing costs the global trading week.

Controlling trading costs in the implementation of any Robust operational risk control strategy is crucial, and CFM ISDiversified benefits from 25 years of experience in this domain. Our dedicated execution research team manages our execution Just as financial risk is managed by our global infrastructure and seeks to model, measure and reduce infrastructure in quasi real time, the monitoring and costs. We have an extensive database of the execution of management of operational risk at all levels in the our own trades – data which is not commercially available, production chain is of critical importance. While and which provides valuable insight into market operational risk monitoring tools are embedded in the dynamics. The execution team has also published trading systems and maintained by the front office teams, numerous papers, in particular on the theme of we also maintain a dedicated independent risk team, market impact. reporting directly to the firm’s directors, whose mandate is to independently validate financial risk estimates and Certain portfolios incur other costs. Financing costs, impose operational risk limits. The latter includes limits on important for equity market neutral strategies, represent position concentration, position sizing relative to the another area of expertise developed since 2001 in the

www.cfm.fr 05 CFM An introduction to Alternative Beta and CFM ISDiversified

context of our successful equity market neutral program. to build more reliable back-tests. These techniques are Minimizing the cost of hedging in our short volatility broadly applied to all of our strategies. strategy is also of significant importance: we have amassed significant experience here since 2005, through Value proposition successful volatility trading programs. These costs must be carefully managed by efficient portfolio construction algorithms. The strength of Cost reduction is also important at the level of strategy diversification design. In the study presented in [2] we show that the Sharpe ratio of the trend following strategy only weakly depends on the timescale used. As a result, we chose to Improving the program Sharpe ratio through run the program at the slower end of the spectrum, giving diversification … us a turnaround time of six to eight months on average for the system. Further adaptations were made in order to limit the number of trades executed – a common cost CFM ISDiversified currently combines six strategies in theme in the efficient implementation of systematic three portfolio groups. Individually, each strategy has a trading systems. modest Sharpe ratio (albeit with strong statistical significance). In combination, this is sharply increased Generally, in implementing any systematic strategy, we relative to each of the underlying strategies. employ various internally developed techniques to reduce trading frequency and, hence, cost. Methods to slow our Naturally we would like to get a sense of the expected trading inevitably reduce expected Sharpe ratios slightly, Sharpe ratio for the program. This, however, is a difficult but (when appropriately applied) the corresponding gain endeavour due to the inherent problems already noted of from the reduction in costs far exceeds the loss in data mining and in-sample over-fitting. Notwithstanding theoretical Sharpe ratio. An understanding of these the challenges, in the spirit of transparency we outline the elements of portfolio tuning are, we believe, crucial to basis of our estimate of expected performance. Based on running a successful Alternative Beta program. a paper trading back-test – appropriately discounted for trading and financing costs, including the effect of market impact – and employing techniques already noted to Allocating to get the best, most robust, account for in-sample bias, our best estimate of the out-of-sample performance forward-looking Sharpe ratio for each sub-strategy, is of the order of 0.6. Decorrelation amongst the strategies helps to improve the performance of the combination, In-sample over-fitting is one of the biggest pitfalls in resulting in an estimated Sharpe ratio of approximately 1.2, portfolio management. Our underlying assumption is that net of advisory fees. To account for unknown unknowns, we are not able to pick the best performing strategy in we apply a significant discount to arrive at a long horizon advance, a conviction coming from more than two Sharpe ratio estimate of 0.8, a value we aim to exceed.2 decades of observing the in-sample effect: recent performance provides limited information on the future out-of-sample performance of a strategy. Hence, discipline … and mitigating asymmetric downside risk is of utmost importance in strategy allocation; we choose an allocation as close to equally-weighted as possible. We maintain a conviction that all our systems are of CFM ISDiversified invests in strategies that exploit (statistically) equal value, even in the presence of recent persistent, long term pricing inefficiencies which, we poor performance for any single approach. If the strategy believe, can be categorized into two groups: risk premia is sufficiently significant over a long term back-test, we and behavioural anomalies. believe that the likelihood of achieving the best future We define a risk premium strategy [1] as a return stream performance is improved by equal allocation. that provides compensation for an exposure to a so-called Our expertise in this field also extends to techniques for asymmetric risk profile. In other words, a risk premium estimating this in sample bias and correcting for it in order provides a positive average return, but occasionally experiences large negative drawdowns. In statistical terms,

2 We have nonetheless achieved an out-of-sample real Sharpe ratio of 1.7 since the inception of the program more than two years ago.

www.cfm.fr 06 CFM An introduction to Alternative Beta and CFM ISDiversified

these strategies exhibit positive drift and negative skew. 235% 60/40 balanced portfolio Currently, two strategies within CFM ISDiversified clearly 215% fall into the category of risk premia, namely FX Carry and 20% CFM ISD, 80% balanced 60/40 Short Volatility, while others appear as pure market 195% anomalies, which we presume to be of behavioural origin. 175%

Of the latter, trend following on futures markets in 155% particular exhibits both positive drift and a significant 135% positive skew (see [2] for a discussion of this point). 115% We believe that part of the mandate of an Alternative Beta 95% manager is to diversify away the negative skew of risk premium strategies, blending them with relatively 75% uncorrelated strategies characterized by a weak, or even positive, skew. The inclusion, in particular, of the trend following and value components of CFM ISDiversified Figure 1 contribute to what we believe is a satisfactory level of both Our Alternative Beta program (specifically, CFM Sharpe ratio and skew. ISDiversified) has a low correlation – approximately 5% – to the benchmark. It is evident that the addition of a modest CFM ISDiversified in combination with a investment in CFM ISDiversified to the benchmark benchmark portfolio of bonds and equities portfolio ought to improve the overall Sharpe ratio. In Figure 1 above, we have plotted the returns of a portfolio that combines 20% CFM ISDiversified with 80% in the We turn now to the added value of an Alternative Beta benchmark strategy, a combination which both improves investment in an existing balanced portfolio of equities the Sharpe ratio by 29%, and also reduces the depth of and bonds. In looking to improve the Sharpe ratio of any the drawdowns by nearly one third: the worst drawdown portfolio, an astute investor is instinctively drawn to adding in the back-test is reduced from -25% to -17%. decorrelated strategies in the search for diversification. The Sharpe ratio of the diversifying investment is important – In summary, CFM ISDiversified exhibits low correlation adding a decorrelated but negative Sharpe ratio program with a portfolio of equities and bonds, and thus improves reduces the Sharpe ratio of the combination. It is the Sharpe ratio of the combined portfolio. There is no mathematically simple enough (though beyond the scope guarantee, of course, that the equity and bond index of this note) to show that any combination of two benchmark portfolio will continue to perform with a decorrelated investments with the same Sharpe ratio will positive Sharpe ratio; in the same vein, neither can we always offer an improvement over either investment taken guarantee that CFM ISDiversified will continue to exhibit on a stand-alone basis. positive performance in the future. However, on the basis of many years of experience, we can reasonably state that Consider the following balanced portfolio as the correlation between the two will remain stable and a benchmark: low and, by applying the techniques described in this note, we have delivered a product that offers the best  60% allocated to the MSCI World equity index statistical prospect of future performance.  40% allocated to the Barclays Capital Global Aggregate Bond Index3 Conclusion

This note both describes the challenges in constructing and managing Alternative Beta strategies, and illustrates the valuable contribution such a program can provide to a portfolio. Previously offered at premium fees as alpha by many hedge fund providers, Alternative Beta strategies are now available at lower fees and with more transparency. Many of these strategies require implementation techniques and infrastructure available only to firms with

3 LEGATRUH: Barclays Capital Global Aggregate Bond Index hedged (Total Return).

www.cfm.fr 07 CFM An introduction to Alternative Beta and CFM ISDiversified

the requisite experience and expertise. Market tested cost of trading, buying pushing the price up, and selling portfolio construction and risk control techniques, pushing it down. combined with an active research program in execution, Market Factors – Statistical factors (vectors of portfolio are elements essential to the success of an Alternative weights) representing the main drivers of market volatility. Beta program. With 25 years of relevant experience, CFM Certain of these factors tend to be (loosely, but has built the tools, the team and the expertise to deliver. conceptually) related to market and sector movements. Alternative Beta for the Persistence – The tendency of a strategy to consistently produce a risk adjusted return through time. mathematically minded Risk Premium – The expectation of a financial reward in reader exchange for accepting a financial risk. For instance, corporate bond holders are paid a higher coupon rate for The P&L of a portfolio can be expressed as follows: accepting the risk of default. The premium, in theory at least, exceeds the expected actual loss. 푃&퐿푝표푟푡푓표푙푖표 = 훼푎푙푡푒푟푛푎푡푖푣푒 + 훽푡푟푎푑푖푡푖표푛푎푙 ∗ 훿푃푡푟푎푑푖푡푖표푛푎푙 Skew (of returns) – Negative returns being, on average, where 훼푎푙푡푒푟푛푎푡푖푣푒 reflects the contribution of alternative either larger (negative skew) or smaller (positive skew) than positive returns. Infrequent large negative returns investments, 훽푡푟푎푑푖푡푖표푛푎푙 expresses an exposure to traditional markets, such as equities or bonds and with frequent small positive returns gives a negative skew; infrequent large positive returns with frequent small 훿푃푡푟푎푑푖푡푖표푛푎푙 the price return of the traditional benchmark. negative returns gives a positive skew. It has become recent market practice to further decompose αalternative into two components, the first t-Statistic – A statistical measure of persistence or representing real, uncorrelated alpha (proprietary, actively robustness. For a strategy or return stream, it is measured managed and accessed at higher fees), and the second a as the product of the Sharpe ratio and the square root of mix of well-known strategies, now broadly identified, the measurement period in years. A t-statistic greater than which have become, in some measure, benchmarks for three is generally accepted as significant. alternative investments. This new classification may be expressed as follows: Vega – A measure of the sensitivity of the price of a derivative instrument to changes in the Implied Volatility 훼푎푙푡푒푟푛푎푡푖푣푒 = 훼푝푢푟푒 + 훽푎푙푡푒푟푛푎푡푖푣푒 ∗ 훿푃푎푙푡푒푟푛푎푡푖푣푒 of an underlying asset. where 훿푃푎푙푡푒푟푛푎푡푖푣푒 is now the return of the alternative benchmark. References

[1] Y. Lempérière, C. Deremble, T.T. Nguyen, P. Seager, M. Glossary Potters and J.P. Bouchaud, working paper and short version, Risk Premia: Asymmetric Tail Risks and Excess Behavioural Biases – Patterns observable in the markets, Returns, and references contained therein (2014). which are thought to be driven by intrinsic human behaviour. For instance, trends might be seen as related to [2] Y. Lempérière, C. Deremble, P. Seager, M. Potters and the human tendency to follow the crowd. J.P. Bouchaud, Journal of Investment Strategies 3(3), 41–61, Two centuries of trend following, and references Correlation – A statistical measure of how closely related contained therein (2014). two entities (example, price movements of two securities) appear to be. A value of zero implies no discernible [3] CFM note (2014), Behavioral alternative beta strategies. relationship; one implies lockstep; minus one implies [4] CFM technical note (2014), Equity market neutral perfectly opposed movement. portfolios: hedging out market factors, and references Implied Volatility – Expected volatility of a market or contained therein. instrument, implied by the price of options on that market or March 2015 Original release instrument. On average, it is higher than the actual or realized September 2015 Slightly revised and updated volatility, resulting in an implied-vs.-realized premium. December 2015 Slightly revised and updated Market Impact – The movement of prices in response to April 2016 Slightly revised and updated buying or selling. It is often the largest component of the

www.cfm.fr 08 CFM An introduction to Alternative Beta and CFM ISDiversified

statements are or will prove to be accurate or complete in Important disclosures any way.

This document is being provided at the request of the HYPOTHETICAL PERFORMANCE RESULTS HAVE MANY recipient for information and discussion purposes only, INHERENT LIMITATIONS, SOME OF WHICH ARE and should not be redistributed. This document does not DESCRIBED BELOW. NO REPRESENTATION IS BEING convey an offer of any type and is not intended to be, nor MADE THAT ANY ACCOUNT WILL OR IS LIKELY TO should it be construed or used as investment, tax or legal ACHIEVE PROFITS OR LOSSES SIMILAR TO THOSE advice or an offer to sell, or a solicitation of an offer to buy, SHOWN. IN FACT, THERE ARE FREQUENTLY SHARP an interest in any trading strategy or investment vehicle DIFFERENCES BETWEEN HYPOTHETICAL PERFORMANCE managed by Capital Fund Management SA (“CFM”). RESULTS AND THE ACTUAL RESULTS SUBSEQUENTLY ACHIEVED BY ANY PARTICULAR TRADING PROGRAM. Any description or information involving investment strategies, process or allocations is provided for ONE OF THE LIMITATIONS OF HYPOTHETICAL illustrations purposes only, may not be fully indicative of PERFORMANCE RESULTS IS THAT THEY ARE GENERALLY any present or future investments and is not intended to PREPARED WITH THE BENEFIT OF HINDSIGHT. IN reflect anticipated performance. ADDITION, HYPOTHETICAL TRADING DOES NOT INVOLVE FINANCIAL RISK, AND NO HYPOTHETICAL TRADING Targets or objectives, including with respect to returns, RECORD CAN COMPLETELY ACCOUNT FOR THE IMPACT volatility or leverage, if any, are used for measurement or OF FINANCIAL RISK IN ACTUAL TRADING. FOR EXAMPLE, comparison purposes only. Such targets or objectives THE ABILITY TO WITHSTAND LOSSES OR TO ADHERE TO reflect subjective determinations based on a variety of A PARTICULAR TRADING PROGRAM IN SPITE OF factors, including, among others, investment strategy, prior TRADING LOSSES ARE MATERIAL POINTS WHICH CAN performance, volatility and leverage measures and ALSO ADVERSELY AFFECT ACTUAL TRADING RESULTS. expectations and market conditions. There can be no assurance that such targets or objectives will be met or THERE ARE NUMEROUS OTHER FACTORS RELATED TO met over any particular time horizon. Performance may THE MARKETS IN GENERAL OR TO THE fluctuate as can volatility and leverage. Targeted IMPLEMENTATIONS OF ANY SPECIFIC TRADING returns are not intended to be actual performance and PROGRAM WHICH CANNOT BE FULLY ACCOUNTED FOR should not be relied upon as an indication of actual or IN THE PREPARATION OF HYPOTHETICAL future performance. PERFORMANCE RESULTS AND ALL OF WHICH CAN ADVERSELEY AFFECT ACTUAL TRADING RESULTS. All of the figures presented in this document are for illustrations purposes only, are unaudited and may not reflect the full cost structure of any investable fund or product. No representation is made that CFM's or a fund's risk management, investment process, trading performance, investment objectives or the control of operational risks, credit risks and other risks involved in any trading strategy will or are likely to be achieved or successful or that any fund or underlying investment strategy managed by CFM will make any profit or will not sustain losses. Any measure of risk is inherently incomplete and does not account for all risks or even all material risks such as risks due to unforeseen catastrophic events.

Any statements regarding market events, future events or other similar statements constitute only subjective views, are based upon expectations or beliefs, involve inherent risks and uncertainties and should therefore not be relied on. Future evidence and actual results could differ materially from those set forth, contemplated by or underlying these statements. In light of these risks and uncertainties, there can be no assurance that these

www.cfm.fr 09 CFM An introduction to Alternative Beta and CFM ISDiversified

CFM has pioneered and applied an academic and scientific approach to financial markets, creating award winning strategies and a market leading investment management firm.

Capital Fund Management S.A. CFM Asia KK 23, rue de l’Université 9F Marunouchi Building, 2-4-1, Marunouchi, Chiyoda-Ku, 75007 Paris, France 100-6309 Tokyo, Japan T +33 1 49 49 59 49 T +81 3 5219 6180 E [email protected] E [email protected]

CFM International Inc. Capital Fund Management LLP The Chrysler Building, 405 Lexington Avenue - 55th Fl., 64 St James’s Street, London New York, NY, 10174, U.S.A SW1A 1NF, UK T +1 646 957 8018 T +44 207 659 9750 E [email protected] E [email protected]