The Gateway to Algorithmic and Automated Trading

The Automated Trader Interview

Published in Automated Trader Magazine Issue 19 Q4 2010

For this month’s interview, we brought together former Brevan Howard COO of Trading, Mike Khorrami, now a Partner at Kiski Group*, and Marc H Malek, Managing Partner, Conquest Capital Group, to discuss Conquest’s approach to systematic trading, trend-following, codified position management and supervised automation.

Mike Khorrami

Mike Khorrami

Mike Khorrami: You've made over 16% this year already. Something's working well. Tell us about it.

Marc H Malek: Our models have effectively captured the brief rises in volatility that occurred during the Flash Crash as well as other market sell-offs while avoiding much of the downside in volatility. Our trend models have also been able to capture the short-term trends in currencies and interest rates.

Mike Khorrami: Sounds like a case of having the right models in place at the right time. How are you positioning yourself for the coming year?

Marc H Malek: We do not position ourselves per se. However, we continue to analyse our strategy and the drivers of market movement. We have confidence in our strategy and believe that the route to continued success is to integrate our accumulated experience and knowledge rather than make wholesale changes.

Mike Khorrami: Tell us about the markets you trade and more about your trading styles.

Marc H Malek: We run a variety of short-term strategies that trade the most liquid markets. Our risk is concentrated in financials, such as foreign exchange, equity index futures, and fixed-income futures. The other 10% of the risk is allocated to liquid, macro-dependent physical commodities such as crude oil and gold. We have found these instruments to be an excellent source of incremental returns and an effective diversifier within the portfolio.

The four short-term strategies capture different idiosyncratic market behaviors: volatility of volatility and rises and volatility; short-term market trends; reversion and reversals within market trends; and selective risk capture.

The fourth strategy is our newest strategy and was implemented to improve our performance during more bullish market environments. It receives the smallest risk allocation by far.

Mike Khorrami: You talk about capturing different idiosyncratic market behaviours. Does the market have to be idiosyncratic for you to succeed, and is there a wrong kind of idiosyncracy?

Marc H Malek: We have found one particular pattern to be difficult: the "V" reversal wherein risk sells off one day only to see an aggressive rally the next. This whipsaw risk is inherent in a short-term momentum strategy. We have taken measures to mitigate this downside exposure without changing the properties of our strategy and expect our ability to minimize downside during these periods to improve.

Mike Khorrami: The investable universe has changed over the past twelve years. New markets, BRICs, changing relationships between currencies. Have such changes just been new opportunities, or have they prompted a more fundamental shift in your approach?

In my experience, the largest constant has been the consistent change and evolution of the markets. Our strategy has navigated a variety of risk regimes and economic cycles at this juncture. The changes we make to our strategy are to improve it as an overall strategy and make it more resilient to the continued evolution of the markets.

Marc H Malek

Marc H Malek

Mike Khorrami: So, a significant proportion of your trading activity is systematic - what led you to become a systematic manager?

Marc H Malek: The decision to build a systematic strategy stemmed from my experience managing proprietary traders. I found that trading mistakes typically stem from human errors and breakdowns in discipline: relaxing stops on open positions (riding losers); exiting positions early despite predefined rules (cutting winners); pressure to trade leads to lower-quality positions.

I also found that traders tend to have a style dictated by how they enter and exit positions and manage risk, so codifying position entry and risk management also reduces the variability.

Mike Khorrami: But you still use what you describe as a 'discretionary overlay' and express strong macro views. Does this ever tempt to intervene when all of your 'macro instincts' are telling you that a model is trading the wrong way?

Marc H Malek: The strategy we trade reflects my theses on the behaviour of markets, so on some level it does reflect my macro views. I do think that it is important that a systematic trader has opinions because it reflects a consistent level of engagement in the markets. However, the temptation to intervene based on my current take is not part of the calculus.

When I have an extremely high conviction view that is consistent with the mandate of our strategy, I can allocate a small percentage of our capital to such a trade. I did it in 2007 with excellent success and repeated the trade this year. In each case, I bought low delta one-year S&P Puts with the view that volatility would expand and the markets would decline.

Mike Khorrami: At what point would you intervene?

Marc H Malek: We intervene on a defensive basis, specifically around exogenous events.

Models are built to work 98% of the time. However, there are certain times when there is a high likelihood of a major intervention such as the ECB meeting this year or the government bailout of Fannie and Freddie last year. As these events happen on a one-off basis, our models cannot necessarily predict them and are often very successful at identifying an erosion of economic conditions prior to these interventions.

In this case, I have found it to be the responsible course of action to reduce risk prior to such events because the outcome going into these events is essentially random. Our models give us a probabilistic advantage over time, so no one day dictates our performance.

Mike Khorrami: How many individual models do you run?

Marc H Malek: We run more than 30 models spread across our four strategies.

Mike Khorrami: Is that a fairly static number of models? Could you talk about the process of adding and subtracting models from the active set.

conquest reception

Marc H Malek: The number of models has grown slowly over time as we continue to perform research and expand the range of techniques we use to trade.

We use a rigorous process of quality control for evaluating models currently in the portfolio such as beta analysis and out-of-sample analysis. Models are removed when they underperform on an absolute or relative basis, or when they begin to deviate from their expected properties. We find that we have had to remove only a few models over the past few years because our model development process rejects all but the most robust models.

This process is one of the hallmarks of our strategy construction. Without going into excruciating detail, I can tell you that the testing process encompasses a broad array of safeguards and rejection criteria. Before a model is even eligible for inclusion in the portfolio it has to go through stress tests, beta analysis, code review, and an extended period of paper trading. Once a model is eligible for inclusion, it is evaluated as to whether it is accretive to the larger portfolio.

Mike Khorrami: Describe the typical trade lifecycle. Time horizon, etc.

Marc H Malek: Our average trade length is six days, although individual models have trade horizons ranging from one day out to a few weeks. Models enter positions after they identify an appropriate entry condition followed by market activity that confirms the thesis.

We evaluate each open position each day. Downside exposure is managed through algorithmically-determined stops, while upside is managed via a variety of techniques.

Mike Khorrami: To what extent is your systematic activity automated?

Marc H Malek: Our systematic activity is what we call "automated with supervision." Although our position evaluation and order generation calculations are automated, every order is inspected and approved by a senior quantitative analyst and trader before going to market.

Mike Khorrami: Does that mean you're giving yourself freedom to disagree with the system? Who typically wins that argument?

Marc H Malek: Absolutely not. The inspection process is strictly for quality control. Because we automated the calculation procedures, we find it essential that we ensure that the orders reflect correct input data. This process also confirms that the portfolio has the positions that it should have and that the orders reflect these positions.

Mike Khorrami: How many orders do you generate in a typical day?

Marc H Malek: We average approximately 300 orders per day.

Marc H Malek

Mike Khorrami: What's your order:fill ratio?

Marc H Malek: We are filled on approximately 15% of our orders.

Mike Khorrami: Moving on to technology, describe your technical infrastructure. Hardware, software, etc.

Marc H Malek: We have developed a proprietary infrastructure for testing models, evaluating positions and booking trades. We found that a proprietary implementation gave us the flexibility and transparency necessary to integrate research, trading, and risk management.

Mike Khorrami: As your AUM has grown, to what extent have capacity and market impact become issues for you?

Marc H Malek: Our strategy's expected capacity is well over $2 Billion, so we are well below that level. In terms of market impact, we have seen our increased size has created little impact in terms of slippage or liquidity. Simple liquidity management measures have been more than sufficient to mitigate any effects.

Mike Khorrami: You run an equity fund within your portfolio of products. Given your trading style, to what extent has market fragmentation been an issue for you in recent years? Do you feel we are about to see similar fragmentation across other markets/asset classes?

Marc H Malek: We don't expect to see much of an effect in any of the asset classes we trade. We only trade the most liquid foreign exchange markets, so prices tend to be largely consistent across counterparties. The broad distribution of trading platforms has also created more convergence.

Mike Khorrami: You trade some OTC products. There are growing calls from regulators and the clearing community to move some of these instruments onto exchanges. What's your view on this?

Marc H Malek: It is not anticipated that such a move would materially affect us from a trading perspective. It would actually result in the portfolio consisting entirely of Level 1 assets and therefore make valuation even easier both from an internal and external perspective.

Mike Khorrami: But if those OTC products moved to exchanges, would there be anything about them that you would seek to replace?

Marc H Malek: One of the largest advantages of the OTC foreign exchange markets is that they offer excellent 24-hour liquidity, particularly for the major currency pairs. We believe that this liquidity would translate to a significant extent.

Mike Khorrami: Conquest are now in their twelfth year. During that time, more and more firms have entered the systematic space - particularly in the last two or three years. Has the wider acceptance of a systematic approach made it easier for you as a firm, or has the increased competition in this space made it harder for you to find and keep an edge?

Marc H Malek: We definitely think the wider acceptance has been nothing but beneficial to Conquest. We feel that we have a unique program based on extensive research and our lack of correlation to other short term managers sets us apart from firms in this space.

Mike Khorrami: Do you have imitators? More seriously, how do you stay unique and uncorrelated over twelve years, if you're visibly successful?

Marc H Malek: Our dual mandate and the diversity of research that is incorporated into the strategy have differentiated us from other managers thus far.

Mike Khorrami: What's your view on high frequency trading? Is the growth in HFT activity welcome in terms of additional liquidity and tighter spreads or has it simply resulted in crowded trades?

Marc H Malek: High-frequency trading is simply another strategy for trading the markets. I think it is certainly less prevalent in futures, wherein exposure is concentrated in a few markets, than it is in equities. However, we do see its effects. Given that we have a significant momentum component to our strategy, high-frequency trading has improved liquidity. On balance, it has probably been favourable for us.

Mike Khorrami: Regulators and the mainstream media were quick to point the finger of blame for the so called "Flash Crash" of May 6th this year at the HFT community. What's your opinion on this subject, and if you were to engage in some finger pointing yourself who would it be at? Market participants in general, the HFT community, the exchanges, or perhaps even the regulators themselves?

Marc H Malek: The Flash Crash accurately identified the largest danger with high-frequency trading: that the market has accepted it as part of the landscape. Traders and investors expect the liquidity that HFT algorithms tend to provide and have changed their behaviour accordingly. When the liquidity disappeared, orderly market function ceased.

Marc H Malek

Mike Khorrami: Looking back at your own performance during this period, it's interesting to note that your best month so far this year +11.92% coincided with May's "Flash Crash". Is there a connection?

Marc H Malek: We did earn more than 3% on the Flash Crash day alone. However, the Flash Crash was symptomatic of the larger market environment on that day. Volatility and investor risk aversion had already begun to rise at the beginning of the month and our models had assumed short positions well ahead of the Flash Crash. In fact, during the Flash Crash, we were getting pushed through our profit targets as the market collapsed.

Mike Khorrami: From what you say, you weren't taken by surprise by the flash crash. Were you surprised that it seemed to take so many others by surprise?

Marc H Malek: Extreme market events are always surprising, even when you are positioned correctly and have an inkling that conditions are ripe for such a shock. The key to handling these events is to have procedures in place to address them and the ability to diagnose and react quickly.

Mike Khorrami: Regulators are struggling to work out an appropriate framework for controlling HFT activity. What should their role be, and what advice would you give them?

Marc H Malek: To regulate and improve the environment where deficiencies are identified without stunting its growth. There is a need to recognize that it is an important and beneficial component to the financial sector.

Mike Khorrami: Describe your creative process. How does a new model come into existence and what are the creative triggers?

Marc H Malek: Every single model that we develop begins with an easily-articulated thesis that we can test empirically. We find these ideas through our own analysis of the markets and our continued efforts to quantify seminal market factors. In this regard, the constant interaction between research and trading proves invaluable for us, as anecdotal observations can be either validated and monetized or disproven.

Mike Khorrami: You talk about analysis and research. Do you ever have 'lightbulb moments'?

Marc H Malek: I certainly do. However, I am convinced that they happen because our persistent research efforts give us an incrementally better understanding of and more information about the markets and our strategy over time. The "light bulb moment" comes when these small improvements coalesce into a new revelation and deeper understanding.

Mike Khorrami: Explain your modelling and testing process that follows the identification of what seems like a good idea.

Marc H Malek: After we quantify an idea and determine that it has predictive value, we evaluate the sensitivities of the strategy to a number of factors, including volatility and risk appetite. We also add own position management rules. In doing so, we are able to refine a raw concept into a streamlined trading strategy. Once the strategy has been developed, it is stress tested and sent for code review and evaluation to the Portfolio Manager and two senior analysts. After the model is approved, it is observed in staging for a minimum of one quarter before it is even eligible for inclusion in the portfolio.

Mike Khorrami: Which software package(s) do you use for modelling purposes?

Marc H Malek: We use a proprietary testing infrastructure for development. However, we also verify our work using standard commercial products.

Mike Khorrami: When backtesting and optimising how do you avoid the traps of curve fitting?

Marc H Malek: Our stress testing, strategy review, and observation processes all help mitigate the effect of curve-fitting.

Mike Khorrami: What boxes does a new model have to tick to make it into production?

Marc H Malek: First and foremost, a new model should be robust in its construction. If the thesis is shaky or the behaviour is ill-defined, the model has no chance of being included in the portfolio. A model must also pass our stress tests and exhibit a coherent pattern of behaviour relative to the markets. Beyond that, a model must behave as expected on an out-of-sample basis and be accretive to the portfolio.

Mike Khorrami: Do you ever over-ride a model's output based on a fundamental view?

recruit

Marc H Malek: No. We monitor the portfolio for risk concentration rather than injecting our fundamental views on a trade-by-trade level.

Mike Khorrami: How far does a live model's performance have to fall outside the window of expectation for it to be retired from the production set?

Marc H Malek: It depends on the model and the conditions under which performance deviates. If a model is losing money but performing well relative to its beta, it receives more leeway. We find that the majority of models are retired not for poor performance but because we develop superior but correlated models that outperform the older models in live trading.

Mike Khorrami: Picking you up on that, to what extent is model development a process of improvement and refinement of existing models?

Marc H Malek: Model development takes many routes. While most of our research is devoted to finding new models, we do find opportunities where we can take the investment thesis of an older model and expand upon it.

Mike Khorrami: Moving on to performance, at first glance there appears to be quite a lot of seasonality in your returns with November and December being particularly good months for you - did you know that since inception you've only had one year when you lost money in December? What do you put this down to?

Marc H Malek: December has a peculiar set of year-end factors that can cause dislocations in the markets. They range from investor allocations and tax considerations to portfolio rebalancing and risk adjustment. These effects combined with risk aversion and vacation schedules can create strong short-term moves and spikes in volatility.

Mike Khorrami: Do you look at every month in a similar way, or is December unique? In the sense, does each month have a distinctive character in terms of their impact on the markets?

Marc H Malek: We are conscious of certain seasonal characteristics but do not believe that our sample space is sufficiently large to make material changes to the strategy on a month-by-month basis. The one factor that would draw our attention would be the disappearance or diminution of market liquidity.

Mike Khorrami: Some investors must also have noticed this trend, and as you have no lock-up, do you see an inflow of funds in November followed by redemptions in January?

Marc H Malek: Not particularly. We find that our investors choose Conquest with a longer investment duration in mind and our flow of funds support this mindset.

Mike Khorrami: The last two or three years have seen what you might call significant regime change take place in financial markets, and 2009 was a tough year for many managers, yourselves included. What's your opinion on why so many managers struggled to make money in 2009?

Marc H Malek: 2009 was very different inasmuch as the correlations were extremely high even as the markets exhibited an unprecedented level of risk appetite. Typically that level of correlation was reserved for market shocks and risk-averse environments. Our strategy encountered difficulties as our models took positions across all markets as markets fluctuated violently within a range.

Mike Khorrami: With the benefit of hindsight, what do you think Conquest and the industry in general might have done differently in 2009?

Marc H Malek: I think 2009 illustrated that high asset correlations create a need for a broader mix of strategies within a portfolio. [See "Correlation Conundrums" on page 72]

Mike Khorrami: Have you adjusted the portfolio with that in mind?

Marc H Malek: We did implement a strategy enhancement this year. However, it was not driven by 2009 performance. Rather, 2009 confirmed a number of portfolio management ideas that we had been discussing internally for some time.

Mike Khorrami: 2010 is much better and you're now well above your 2009 high watermark. To what extent did 2009 force a change in your approach to trading, your existing production model set, and the design of new models?

Marc H Malek: 2009 did not force a change in our trading but it did validate one of the investment theses that we had been evaluating for over a year. In November 2008, we introduced a framework for replicating the risk premium capture component inherent in hedge fund returns. Although we found the idea conceptually attractive, we were very conservative about integrating it into our portfolio and only did so after a year of observation. This addition combined with a selective leverage reduction during risk-seeking periods are the substance of our 2010 strategy improvement.

Mike Khorrami: What are you evaluating for 2011 and beyond?

Marc H Malek: We continue to implement our three-pronged research approach of system design, quality control, and quantitative macro analysis. This multifaceted approach allows our knowledge base to grow in an organic manner instead of forcing it to be a linear process. I can tell you that one initiative that we have undertaken is expanding the range of quantifiable factors we use for identifying trading opportunities.

Mike Khorrami: Over the last decade or more you've had some phenomenal months in terms of performance - for example 20.63% in March 2002, 11.55% in July 2007 and nearly 12% in May of this year. Putting it crudely, was it the same thing that went right every time? What did you do right on each of those occasions?

Marc H Malek: In each case, we were able to capture a rapid turn and move in the markets. This ability to respond quickly to changes in market sentiment is a hallmark of our short-term trade horizon.

Mike Khorrami: Given the events of the last three years, to what extent have you seen a change in the risk appetite of the investment community?

Marc H Malek: We find that investors are more risk-averse on balance. Investors have become much more wary of risks stemming from changes in market sentiment and market shocks. Although leverage has declined to some extent, we see that the zero cash yields are forcing individuals to become more and more invested. The difference is that they are primarily directing their capital toward debt securities, which are perceived to be safer, rather than equities and real estate.

Mike Khorrami: For a firm with a relatively large amount of assets under management, and an excellent track record, you have only twelve staff. Other asset managers of a similar size have many times this number. What's the secret to this efficiency?

Marc H Malek: We feel that Conquest has assembled a great team with a superb industry expertise.

Mike Khorrami: Please describe the daily routine, in terms of who does what. How does a team of twelve work together to achieve such success?

Marc H Malek: Each person has a specific set of responsibilities that are owned by that person. That person performs these tasks and keeps another member of the firm abreast of the process to create redundancies.

Marc H Malek: Where we leverage our talent is in the interaction between the various parts of the firm and between the senior and junior staff. Everyone in the firm works directly with each other depending on the needs of a specific project. This interdependence of the departments streamlines workflows by highlighting each individual's talents. It also gives each team member a more global view of the strategy and the business, and is the key for the rapid development of our junior staff.

Mike Khorrami: How do you recruit?

Marc H Malek: We add team members for either defensive or offensive reasons. Defensive hires are made when the workload in one area becomes excessive. Offensive hires are done on an opportunistic basis when we find a person whose background will expand our capabilities and whose talent gives us further optionality down the road.

Mike Khorrami: In our experience, many successful money management firms have a mantra that runs through the entire firm. Often, staff either buy into that mantra or they leave. Do you have one, and if so, tell us about it?

Marc H Malek: There is no official firm mantra. As mentioned before, we have assembled a great team, one that works well together. We've had no departures, only hires for over 3 years, which is a testament to the positive attitude and environment within our walls.

Mike Khorrami: Where are you currently focusing your development efforts?

Marc H Malek: We continue to build models that expand the core strategies portfolio but have also been looking to quantify additional factors that can be used as filters in conjunction with our position management techniques.

Mike Khorrami: Please say some more about that.

Marc H Malek: We are leveraging the research process that led to the development of the Conquest Risk Aversion Index, Conquest MFS, and our long risk proxy. Each market has its own set of factor sensitivities, but there are common traits across assets. Identifying these common traits creates a new set of criteria for systematically identifying trade ideas.

Mike Khorrami: Putting your macro hat on for a moment, which asset classes and trading strategies do you feel will offer the best opportunities over the next couple of years, and why?

Marc H Malek: Although I do expect to see renewed risk aversion in the relatively near future, the timing of the correction is much more difficult to predict. I can say that I expect the level of correlation to continue and that outsized moves in markets will continue to be the norm. With those factors in place, I expect directional strategies to outperform relative value. The empirical analysis also suggests that it will be favourable for strategies that capture changes in volatility and volatility-of-volatility.

Mike Khorrami: Where do you go from here, and how do you see markets changing and the investment management industry evolving over the next couple of years?

Marc H Malek: The primary evolution in the markets over the past several years has been the rise in asset correlations, both between individual equities and across asset classes. Prices move according to a set of binary factors, be they risk-on/risk-off, long-dollar/short-dollar rather than micro-level details. We expect that this factor will continue and managers will have to adapt their strategies accordingly.

*Kiski provides strategic advice and services to institutional investors, corporations, and asset management firms. Kiski's customized solutions enhance investment returns by integrating investment, business and risk management best practices to address the clients' needs. www.kiskigroup.com