AC: Let’s begin by talking about the investment approach Quant Hedge takes for currencies.
VL: We have started traditional systematic CTA short-term algos on foreign exchange. It’s an absolute return investment strategy using short-term and medium-term investments. What we do, we simply find microstructure and discrepancy in the market, and trade these short term investment opportunities with a range of algos. Part of it is very computational, trying to understand the best microstructure in the market, on an intraday basis or on an intraweek basis. The other part of the trading is mainly being sure that the algos are working well, checking operational risks are not deviating from the expected performance and catching enough return to make sure that the cost of trading is compensated and that we benefit our investors. On the treasury side, we need to be invested because we are managing short term positions so we have to enter the treasury market while algos are not trading to keep cash invested in risk-free assets. So that’s typically what we do as a traditional CTA short-term fund.
AC: In terms of risk, one of the themes we hear about is the return of volatility. Is that something you’re seeing and how do you deal with that?
VL: On the risk side, I think there were tremendous issues on the foreign exchange market in the last five years. If you look traditionally, it seems to be like the crisis we had in the 90s and the Asian currency crash, where there was a very long international exposure to some currencies for some people and short exposure in others. The large moves in the exchange rate from the national banks and the national treasury services really shook everything in the foreign exchange market. Most junior people that came into capital markets after 2002 were not used to that, and for many young fund managers, used to managing long term investment portfolios, the currency volatility impacted the risk of the models. For, short term fund, as we are, what happened was that our algos were not subject to this kind of long movement. We benefit from volatility moves.
Typically what happens is that we have to build some risk framework based on global macro indicators and extreme situations, what we call fat tail, to make sure we’re not too exposed to an extreme situation using this portfolio of algos. Unlike equities, we aren’t using the kind of global equity long term portfolio model, based on “relative valuation” strategies but we are building on an absolute return basis. Every time we go on the market, we know we have to leave it intraday or after one or two days, or at maximum one week. So the risk has to be measured in an absolute loss, directly. We tend to be very cautious in terms of investment because we can’t stay two or three months in the market and then close the position. Our strategy is to focus on tight management of the short-term investment as much as possible. So we are very reactive to the market.
AC: It sounds like you’re always ready to take a loss if you need to?
VL: That’s typically what systematic absolute return funds do, whether in futures or foreign exchange. It’s more similar to active management, to long/short equity funds. But we are really doing that based on this statistical short term investment opportunities. The benefit is really to be reactive in the market, not to be in some situation where you’re locked in your portfolio positions. Many companies were locked in on the Swiss franc situation, which was really terrible for many brokers and fund managers, knowing long term FX hedges are very expensive. We didn’t have this kind of exposure simply because there was no volatility in the Swiss franc so we said: We can’t trade that, there is no volatility, we’re not going to make money with it. Then there was a spike of volatility but so huge – about 20-25% against the euro at the beginning – which is crazy. Quant Hedge algos were not trading the Swiss franc so we were happy to rely on our global macro factor analysis to anticipate extreme market conditions. In fact, we have decided to spin-off a start-up from these global macro analysers, offering ‘reality mining’ indicators analysing large sets of data, and producing real time structured information.
This is also the kind of situation we understood, that there could be some intervention from national banks on the foreign exchange and not only on the exchange rate, from the deposit rate of the country. So we’re really careful about macroeconomics. This is another type of risk information we try to gather, which is global macro risks based on the currency deposit and fixed income and exchange rates. We look at the impact from that to make sure central banks are not going to suddenly change their policy over currency and treasury management.
AC: Are you trading Swissie now?
VL: No, simply because with all this movement of the Swiss franc the spread on the currency is still quite high. It is too early to deal with that right now. We’re just waiting the market to re-adjust before trading CHF on a systematic way. It has never really been an opportunity because the Swiss franc is more trend following and not very volatile. Every currency reflects its own country in terms of philosophy, behaviour or people – CHF is very slow in terms of trading. When you have a trend, the Swiss franc is really following the train for a long time. Whereas if you look at an opposite situation, the British pound, you have a lot of small spikes during the day and very strong moves also. It’s the behaviour of the currency. The exchange rate is really different from one country to another. The Swiss franc is more a long-term directional engagement for us and we don’t see really short-term opportunity. Usually CHF strengthens when there is a crisis and weakens when it is growth time, so we don’t expect to see it lower as the EM crisis is in progress.
AC: How do the widening spreads affect execution?
VL: For the first strategy, intraday and intra-week, it has no real importance. We may pay a bit more but it’s not impacting our performance. If the spread is going up, it’s simply because there are fewer people trading and less liquidity providers. Also, most of the banks have opened a kind of private pool between themselves. This means typically they take off liquidity from the global market and ECNs, where we used to see that liquidity. It only has a low impact on us, just making trading a bit more expensive, but not so much because we have a spread management algo.
We have seen some changes with a new strategy for high frequency arbitrage – sub-second trading – where we are doing very short term market microstructure analysis, extracting cash directly. What happens is that we see the spread is really important for these HFT strategies. You may sometimes get 0.5 to 0.6 pips in benefit per million. And if you do that all day long you may end with good return net cash.
But with the widening of spreads globally it becomes more difficult because opportunities disappear. Where you were used to extracting euro/dollar let’s say $40 three or four years ago, now typically you will end up with $15 per million, so it isn’t really worth it. Globally and you have to find another way because it won’t cover your cost of trading.
I’m not sure it’s going to stay like this because it’s still a very fast market. There are lots of Chinese and Turkish people coming in, new brokers coming in the market from the Middle East and so on. So I think it’s just a reorganisation, a phase.
AGGREGATED ALPHA
 |
AC: What makes your firm different?
VL: We are doing something very, very different, completely the opposite of what is called systematic equity with traditional risk/return adjusted portfolio management. We are doing alpha aggregation, meaning we want to catch very short-term alpha, short-term alpha and medium term alpha. Rather we are an ‘alpha maximiser’ I would say.
The fact is, with these short term strategies, you’re exposed to risk but as we are an absolute return fund we only invest the amount of money we agree to lose. Typically the maximum loss we have on a daily basis is 2% (VAR based). So if we lose 2%, we stop trading. Then we are waiting for the next best market condition to try again. We know exactly how much money we’ve put into that and also in terms of how it’s allocated with each strategy. That’s the way we differentiate from others who usually do trading based on asset diversification. We are diversifying in terms of market timing and strategies. That way we think is better because the engagement in cash directly in the market is actually more diversified and we are not locked in specific asset closure processes. You know you’re going to invest some amount of money in the very short term but you will have it back soon or you will lose it. If you make a return you can re-invest in short and medium term, and so on. That’s the philosophy of the fund.
AC: You use algos to trade this way. What’s the process in terms of developing them?
VL: We have some message routing tools and APIs in FIX Protocol to connect to the market. But on the algo side it’s really data mining-based and microstructure analysis, I would say. We catch all the data we can get from the market, also high frequency data. We try to get data with a frequency with less than one second, even if it was two or three years ago, to get the complete set of high frequency data every day. For instance we had a series of tests on FX liquidity providers and on BATS HFT data to understand the order book dynamic for some stocks. We take also into account micro indicators and macro risks to see what impacts the markets and the behaviour of the currency. So there is a bit of work in data management around that and also data standardisation before we are able to introduce a new algo or do some research into that. After, the main thing is trying to understand what we call market timing. If a pair is moving in one direction, when is the best time to trade it and which strength and which length should we anticipate? In terms of trading opportunities it could be from five to six minutes to two hours and some opportunities are here for several days. We’ve built a framework to understand what the microstructure of the market is on a daily basis and a weekly basis, so we know more or less how the market will react. One more thing is that we don’t believe in the fully automated systems, the ones where you let robots trade and you go on holiday. We always keep a functional validation and a human behind the robot because market conditions are always new and robots still haven’t learned economics.
FRONT TO BACK
AC: How do you get to this stage in terms of developing the strategy?
VL: That is the beginning of the observation. It’s not purely algo based. Then, we have the construction of the strategy itself, and we back test the models. We’ve tried to introduce a new way of doing back testing. It’s back testing not from a historical point but from the last point of today. We test based on an observation from today and go backwards. Instead, what most of the funds do when back testing is they test a historical set and say: Does it work now? So we are completely re-engineering the process to make it work the other way. It’s been four years now that we’ve been doing back testing, so we know an algo is more reactive where the data is freshest. So you’d rather test on the short term period, in your real trading environment.
AC: In starting with the present and working backwards, it’s kind of an archaeological approach as opposed to starting with the past and working forward.
VL: Yes, people are doing that indeed (starting from the past) because most of the time they’re trying to identify invariance in the market or market fundamental behaviour also known as Factors in the Fama-French models. I remember last year we had for about six months the euro not moving around a lot of systematic funds said there was no volatility and didn’t make money. So that’s typically the kind of thing we try to avoid, a low volatility environment. Trying another way of doing back tests really gave us good insight about how the market is reactive or not regarding the past. So we are not trying to find fundamental value in the past. Definitely there is fundamental value, but that doesn’t mean the fundamental values are true and invariant all the time in the past.
AC: How do you go about reducing market impact? Do you use algos that help in that respect?
VL: We are not really sensitive to market impact regarding the size of orders. I would say that it’s around 5-7 million for the medium term strategy, short term and very short term are usually less than one million in terms of execution. I think the biggest one we had is 15 million in the market. We are using dedicated, specific tailor-made algos for execution. We’re not using traditional VWAP, instead we have a more high-frequency approach because the VWAP is really equity-based originally and recomputed every 30 minutes usually, while we execute in several seconds or minutes. You have to reconfigure it with some specific data for market. You also have to configure and back test it to make sure it’s going to execute and slice the order the right way. For each pair, we have a specific configuration of the slicer and spread optimizer, but usually we don’t use it excessively.
AC: How do you think the structure of the FX market differs in respect to other markets?
VL: Unlike equity, you can’t say really there will be winners and losers. Also, the movement of the market is quite different from equities because the spreads make the market very cheap. That’s a good opportunity for everyone to do electronic trading and use algos. Another strong point is the number of exchange pairs. Let’s say you have 150 you can trade globally; in the equity market you will have maybe 10,000 worldwide. It’s become huge and you have sectors, openings and closings on the multiple equity market venues and time zone, which you don’t really have in the foreign exchange market unless it’s over the weekend. Also, when equities are very difficult, you can be locked in an open position end of day. With foreign exchange we have the opportunity to go in and out if we’re not confident with what’s happening. Even if it’s widely OTC and quickly impacted by specific policies, we tend to think it’s a good market. It’s easier to trade than equities, where you’re forced to diversify. In the FX sector, you really don’t have to do that. You can go Asian currency or emerging currency, but you don’t have so many factors so you can read the market easier. It is a bit similar to the futures market in that it is very liquid. However futures are a transparent market, while FX is OTC and subject to manipulation as we have seen with the regulatory scrutiny and sanctions over some banks these last few years.
AC: Are there any new developments for the firm that you can talk about?
VL: Today we’re working on arbitrage strategies. We have several pieces of research going on. One focuses on portfolio management models for our electronic trading algos, re-engineering the equity portfolio model to develop a global risk integrated investment management framework. For the Greek exit we have built a real time ‘De-Europeanizer’ to estimate the real value of one euro in each euro-country, where you see one Greek euro value is 0.71% of a euro, while a German euro is worth 1.1 euro. So if you find a euro banknote starting with a ‘Y’ code, this means it is Greek allocated so spend or change it right away. One starting with an ‘X’ is German allocated so keep it in your safe. We also started analysing digital currency and blockchain to see whether there are trading opportunities there. In this we are following one of the French blockchain pioneers, Frédéric Peters, founder of Eqinity.
As I mentioned before, we expect to spin off the global macro indicators with a firm called Daimeri. We also have a research partnership with ENSRF, a private research scheme in Paris led by Pr. Duc Pham-Hi, and the ECE Paris where I teach e-trading and capital markets. We’re doing some research on cultural asset trading models such as art, books, classic cars, and some other research on high-frequency data analysis, EM indexes strategies and FX-AI models. It’s very stimulating.
