Looking Under the Hood, Part 1

Smart betas, factor-based indexing, and risk premia

THE OPALESQUE UK ROUNDTABLE

The academic literature of the last 30 years is filled with papers and research that support the existence of certain premia in various asset classes. Risk premia products are no longer a novelty; most bulge-bracket banks have successfully rolled out their suites of “risk premia” and “smart beta” indices, which has been of great interest to Nordic and US pension funds.

Risk premia such as value, momentum, carry, etc., provide very efficient portfolio building blocks, simply because the correlations among risk premia are generally low and relatively stable. Their use in portfolio construction can help achieve more efficient returns than those using the traditional geography or market cap building blocks that are commonly used in many portfolios.

Today, these new strategies are also called smart beta, smart systematic beta, factor-based indexing or systematic risk premia. But when the quintessential mechanical engineers and astrophysics PhDs start kicking the tires and looking under the hood, deeper questions open up…

Is it just a back test?

Obviously, risk parity seems to have worked in the past, so how different are the new products from, say, selling a back test? Moreover, when it comes to equity index beta, the beta that we have from, say, the S&P comes with a clear and uniform definition, and it’s absolutely free. But, looking at other betas like, say, FX carry, what’s the beta for that?

“What if you’ve got one and I have got one, and mine is different?” asks Ewan Kirk. “It’s not really a beta then. Maybe my beta is more beta than your beta. Maybe your beta has got more alpha in it than my beta.” On top of that, there are costs associated with dynamic positions so, in a sense, it can’t be beta. So maybe a dynamic trading strategy can’t be beta, maybe it’s just a dynamic trading strategy. However, it is undisputed that risk premia can provide portfolio managers with a greater number of low-correlation portfolio building blocks than the more typical geography and market cap building blocks.

It’s not about your model – it’s about your execution.

Today, it’s not a challenge any more to write down a nice, simple trend model, which in gross terms looks great and appears statistically indistinguishable from the best CTAs in the world. However, that will be in gross terms. Owning a cap-weighted equity index is free because you don’t have to trade. On the other hand, if the rebalancing cost of a portfolio is only 1 basis point a day, that will be 2.5% per annum, dwarfing the cost of management fees. Therefore, huge efforts go into saving tiny fractions of a basis point and executing at “lag 0”. Still, some offered risk parity strategies are quite stripped down in their substance and therefore appear to be inexpensive.

Matthias Knab
knab@Opalesque.com

****

The Opalesque 2014 UK Roundtable was sponsored by Salus Alpha and Eurex and took place on 3 September 2014 at the London offices of Eurex with:

  1. Oliver Prock, CEO, Salus Alpha Group and CIO of Salus Alpha Capital;
  2. Ewan Kirk, CIO and co-founder, Cantab Capital Partners;
  3. Antoine Haddad, founder and CIO, Bainbridge Partners;
  4. Stuart MacDonald, managing director, Aquila Capital;
  5. Nacho Morais, CFA, FRM, CAIA, Pragma Wealth Management;
  6. Akshay Krishnan, head of macro strategies, Stenham Advisors;
  7. Renaud Huck, senior vice president, Eurex Group.

****

Matthias Knab: Some of you have already mentioned a number of new products that you have launched successfully. Let’s look at those and other new launches in greater detail. Please share with us what products you are working on at the moment.

Oliver Prock: At Salus Alpha we are currently working on some new arbitrage models to add to our commodity arbitrage strategy. Besides that we constantly try to improve execution technology for all our models. We achieved fully automated execution with straight-through processing capabilities that is able to add positive alpha and eliminate slippage entirely. I think the execution engine we developed is unique and gives us a significant competitive advantage.

Renaud Huck: For us as an exchange 2014 was a very interesting year. As you will be well aware the past few years have been fairly busy in terms of regulation. The Dodd-Frank Act was introduced in the States, while European regulators decided to launch the EMIR regulation, and as a European clearing house we had to go through a process of being reauthorized by ESMA in order to become a qualified CCP.

When we did our preparations ahead of EMIR, we realized that a lot of the buy-side and sell-side players in the industry who were extremely active in the OTC space would as a consequence of regulation see the scope of activities drastically diminished or constrained. Some would even have to potentially exit the OTC space and move their business model towards a more exchange-listed and exchange-cleared model.

Consequently, we thought long and hard about what new products and services would be needed. Obviously our clearing house (Eurex Clearing) has been offering OTC IRS clearing for more than two years and, in terms of products, we recognized that in the fixed income space there would be a need for a swap futures and a repo futures product. We are delighted to be able to offer a complete suite of fixed incomes futures, covering the whole range of short and long-term money and bond markets with listed and OTC products.

As of 1 September 2014 Eurex launched swap futures and in mid-November we will launch repo futures. Earlier in July we launched FX futures, which for us was a whole new asset class that had not been targeted or offered before. Subsequently, we are currently very busy promoting these products, but also engaging with the buy side and sell side in a broader approach; looking at how the new regulatory landscape will influence business in the future, the new rules of engagement for market participants and, taking all this into account, we offer them dual avenues to suit their needs – whether it be traditional exchange-listed products or OTC cleared possibilities and services.

Ewan Kirk: I have a question. I haveseen currency futures on the CME for many years, and they have not taken off. They are still illiquid, and the volume remains in the OTC market. How do you think yours will be different and why do you think they have not yet taken off?

Renaud Huck: Within the incoming regulation we anticipate that the regulators will expect FX to become a more regulated market – after looking at fixed income and equity markets. I would not rule out a potential clearing mandate for OTC FX contracts, for instance, forwards or options. I am in a good position to answer your initial question on the difference between ours and the FX futures at CME, as I was the head of hedge funds and sovereign wealth funds at CME for almost nine years where I dealt with those products globally and answered those questions in front of buy-side and sell-side entities. It is true that the CME had a difficult position of being the first one to enter the space and perhaps was then not able to structure the products in the way that we at Eurex structured them.

The CME decided to structure them not from an FX spot protocol approach but from a US market approach, whereas our sterling-dollar product is sterling-dollar, as is the FX spot convention. Our products are within the protocol of the spot FX market, which significantly means that you are not in a situation where you have to face reversed quotes firstly, but also for those who are more used to spot FX it is not such a big adaptation since you don’t have to reverse the quote to see the equivalent with your FX spot price. We also decided to go for nominal values which are 100,000 of the base currency unit to avoid having amounts of 125,000, for example, which is far more practical.

Entering the market after the CME allowed us to refine the product offering. We observed that launching options on futures was not necessarily the best approach so our options are on spot.

Those who choose to use our options will have a physically delivered currency at expiry – and not in the futures contract – (our currency futures are physically delivered); the major risk in currency is not necessarily the counterparty risk but rather the settlement risk, and all our currency futures contracts are CLS-settled. CLS is the standard currency delivery system and processed within the FX space and that met our requirements. We wanted the market participants to have confidence in our product in the way they were structured but also in the delivery process if they wanted to go through delivery of those instruments.
So in a nutshell, our FX offering mimics the existing and dominant market structure as much as possible, with the idea to offer a complementary transparent and efficient exchange-listed and centrally cleared FX segment.

Antoine Haddad: Here is quick question for you regarding costs – do you have any particular features for long-term holders of currencies so they can minimize their quarterly roll costs?

Renaud Huck: It’s true and it is a very interesting question about the futures contract. If you are a portfolio manager or a small or large hedge fund, you are going to have to roll your position on a quarterly basis; it’s true that it does add on the cost. We are conscious of that and that is why the pricing of our futures is in line with the fixed income pricing that we have for other futures. So it’s now really a question of whether you are or not a member of the exchange and are paying a multiplier of what a member would pay if you are a non-member.

So in this case it’s $0.30. This information is in the public domain – on our website – so $0.30 per contract and you have the possibility to benefit from volume incentive if you are very active. Still, the question that you raised is something that we are thinking about breaking potentially by looking, for example, from that quarterly cycle of the rollover of the position to maybe going forward having yearly contracts. So we are conscious about it, but I think that this is something which is very much for the next generation of futures instruments.

Antoine Haddad: Coming back to your question about new product launches, allow me to frame my answer within the two separate activities that we have – “alpha” generation on one side, and “beta” portfolio construction on the other. Before I speak about the new launches within our alpha portfolios, I will say a few words about the environment for alpha generation. It is no secret that the extremely low-volatility environment of the recent past has not been extremely supportive for the “trading skill” alpha that we tend to look for. Looking forward, however, I cannot imagine volatility in equity markets or foreign exchange markets, for example, remaining near their 15-year lows at 12% and 5.5% respectively. The coming expansion in volatility will help two things: first it will help improve the return profile of short-term directional traders, and second, it will increase the appetite of long-biased investors for strategies that are more defensive and neutral. Both points will be supportive for the strategic development of our “alpha” arm.

The new development on that front is that we are internalizing some of the alpha strategies that we used to access via allocations to third-party funds. Instead of allocating the assets of our pure-alpha “Aperio Master AlphaStrategy” product to external managers, we are hiring portfolio managers to run those strategies in-house. We are doing this because institutional investors, in the current risk compliance climate, are requiring full transparency from us, down to the position level, something we weren’t able to offer as a traditional fund of hedge funds. This has naturally led us to build our own internal systematic trading strategies and prop-trading teams, replacing some of the external managers with internal managers. That is one significant change in the structure of our alpha department. It is this constant effort to innovate on the multi-manager side of our business that has helped us post positive returns during the 2008 financial crisis.

On the beta front, we are launching two UCITS funds – the “Sequent GTAA – Cross Asset Risk Premia” and the “Sequent TAA – Equity Risk Premia” trading programmes. Both programmes will be long-biased, but will rely on risk premia building blocks to diversify their exposure to the various asset classes they are exposed to. Both launches will be done in partnership with CBP-Quilvest, an innovative Luxembourg-based private bank that has clearly moved away from the simple product distribution model, and that focuses on finding optimal investment and management solutions for its clients.

These launches follow a strong interest from both practitioners and academics in the improved portfolio construction that can be achieved using risk premia (sometimes referred to as factor-based indices, or smart betas). Our focus at Bainbridge has been twofold: in a first step we have embarked to create systematically the various known risk premia (traditional, value, carry, momentum, etc.) for each asset class (equity, fixed income, foreign exchange, commodities), and in a second step, we have overlaid a tactical component allowing us to change the portfolio weight for each risk premium.

Risk premia products are no longer a novelty; most bulge-bracket banks have successfully rolled out their suites of “risk premia” and “smart beta” indices, with great interest from Nordic and US pension funds. Risk premia such as value, momentum, carry (and many other premia that are well documented in academic research) provide very efficient portfolio building blocks, simply because the correlations among risk premia are generally low and relatively stable. Their use in portfolio construction can help achievemore efficient returns than those achievable using the traditional geography or market cap building blocks that are commonly used in many portfolios today.

Our two new products – the Cross Asset portfolio and the Equity Only portfolio, will use risk premia as their diversification engine. The differentiating aspect these products have is their reliance on a tactical module that uses various macroeconomic variables, allowing their exposure to the various risk premia to vary over time. Both will be managed using average and maximum VaR targets.

Ewan Kirk: There is a mathematical definition of beta and it is the undiversifiable market risk. When it comes to equity index beta, the beta that we have, say, from the S&P, we need to bear in mind that there is only one S&P and only one way of market cap weighting equities, and it’s absolutely free. But in terms of foreign exchange carry, what’s your beta for that? What if you’ve got one and I have got one, and mine is different? It’s not really a beta then. Maybe my beta is more beta than your beta. Maybe your beta has got more alpha in it than my beta. Additionally, there are costs associated with dynamic positions so, in a sense, it can’t be beta.

If you have a dynamic trading strategy, it’s not beta, it’s just a dynamic trading strategy. Or to look at it another way, it’s just a systematic trading strategy that happens to have worked in the past. Risk parity happens to have worked in the past. Another way of looking at this is that it is quite close to selling a back test and I feel that some smart beta products are skirting that a little bit too closely. We have a cost-effective trend and value product in the macro space. I am not saying that is beta, but just a cost-effective systematic trading strategy. That’s all it is, but I think many people would call it a beta strategy.

Antoine Haddad: Correct, a lot of people name all these new strategies smart betas, factor-based indexing or risk premia. I agree with Ewan that, ultimately, these are simply systematic trading strategies. We like to categorize or subcategorize these systematic trading strategies under various labels. Some systematic trading strategies can be as simple as adding a risk-parity module to a long-only portfolio, other systematic strategies can rely on filtering the instruments traded with a market capitalization filter, or a geography filter. In our programme, we have chosen to use a well-documented category of systematic strategies – equity and cross-asset risk premia.

Our ultimate goal is to improve the efficiency of our portfolio with a basket of systematic strategies that fulfil the following requirements: they have to have a strong and well-documented economic rationale, be very liquid, and can be easily modified to achieve a “long bias”. Many cross-asset risk premia fit those requirements. The academic literature of the last 30 years is filled with papers and research that support the existence of certain premia in various asset classes. The research we investigate is only a first step in the validation that we are looking for; it provides us with a rational fundamental premise for a strategy before inclusion in a portfolio. In a risk premium (such as carry, or value, or momentum, for example), investors are “paid” a premium for taking a risk away from other participants not willing to take that risk. The true advantage of this approach derives from the fact that risk premia can provide portfolio managers with a greater number of low-correlation portfolio building blocks than the more typical geography and market cap building blocks.

This is the direction we have taken at our firm; we are creating, for example, products that will be replacing some of the “equity beta” clients are searching for with baskets of equity risk premia. We are also creating portfolios of “fixed income beta”, with a tactical basket of fixed income risk premia. If clients are looking for commodity exposure, we will give them the option to participate more efficiently in the upside of commodities via well-known commodities risk premia. The tactical aspect applied to risk premia is critical to the success of our approach. Most of these risk premia may suffer from down periods, should they become “overcrowded”. We look to use our tactical approach to navigate the short-term negative environments that these systematic risk premia will encounter, and take advantage of their well-documented long-term sustainability.

Ewan Kirk: I may be wrong, but I do believe there could be a small problem embedded in this approach. Let’s take the size bias as a specific example. This is a well-known factor premium. If you can manage the risk and buy small companies rather than buying large companies, over time you outperform the index. The trouble is we can’t all do that, or else eventually every company will have the same market cap. For example, a tiny Norwegian fish canning company would have the same market cap as Apple because if we all invest in small stocks they become big stocks and the premium goes away. So although it has been true in the past, if everyone does it, it goes away by definition.

Oliver Prock: Sure, but not everybody is doing it. I gave up on trying to respond to potential investors “Why is not everybody doing that?” The short answer to this is, “because they don’t”. Of course there is a long answer as well…

Ewan Kirk: Right, but still, not everyone is doing it. So there is a first-mover advantage here and you need to get into small caps before everyone else. We should realize that there is no getting away from market cap weightings for the market; everyone has to be market capital-weighted on average. However, maybe some people are smarter and buy the small cap stocks, value stocks or low-beta stocks, or they buy the high-yielding currencies and sell the low-yielding currencies, but this is not a solution for the industry, it’s only a solution for first movers.

Stuart MacDonald: Smart beta has become a very popular concept, particularly amongst some leading institutional consultants. And smart beta, like any other umbrella term, covers a whole variety of different things. There is smart beta going on in credit, for example. Others position risk parity as smart beta or as we do at Aquila, as smart systematic beta. When I first came across the concepts of smart beta and risk parity, in common with many other people, I felt that you could apply it to almost any mix of anything, but actually you can’t, particularly if liquidity plays a role in what you’re doing.

Ewan Kirk: The aspect that is often overlooked in risk parity strategies or indeed any systematic strategy is the role of execution costs. To go back to cap-weighted equity indexes, this is free because you don’t have to trade. However, if it just costs you one basis point a day to rebalance your position, which is not an awful lot of money, but that one basis point a day is 2.5% per annum. That is a lot of money and you are already paying that into the market just to get to your risk parity position. The cost of trading dwarfs the cost of management fees – it’s the cost of trading that you really need to worry about. That is why implementation of such strategies matters. We all know certain strategies work, but in the end it’s all about how do you do it.

Stuart Macdonald: You need to look at the execution or general implementation costs, but the hard reality is that the manager has to make sure that they are kept down to a level sufficiently low so as not to make their charges uncompetitive. There are risk parity strategies that are quite stripped down in their substance and therefore appear to be inexpensive.As with traditional hedge funds, the fee levels ought to reflect how much is actually being done over and beyond the basic, but nevertheless it’s in everyone’s interest that the costs are kept down. Something that we may go on to later in the conversation is the role not just of execution and implementation but also matters operational as a whole, which is increasingly being seen almost as a source of alpha.

Akshay Krishnan: We have talked a lot about betas, carry, risk parity and systematic trading in this discussion. However, at Stenham today we have taken a slightly different approach. Within our macro fund of funds today we are purely focused on discretionary trading strategies. This is more a temporary rather than permanent change. Around 12 or 18 months ago, we started to feel a bit uncomfortable regarding the potential impact of the end of QE on CTAs, especially considering the level of leveraged long fixed income positioning. We also noticed systematic strategies had entered new markets to identify sources of carry, such as emerging market interest rates, credit derivatives; markets which we think are more prone to liquidity gap risks. I’m interested in your perspective on how you think systematic strategies will handle this potential inflection point as we perhaps embark on a tightening cycle led by the Fed as well as extending systematic strategies to newer markets?

And finally another question I have as an allocator for everyone here is that, as Stuart pointed out, a lot of things are being wrapped in risk parity or in smart betas, so sometimes I do worry the end investor isn’t fully aware of what they are getting. For example, we often hear about investors using certain systematic strategies and risk parity approaches as a diversifier to their pro-risk allocations elsewhere in the portfolio, but then also people can be unaware that these strategies can have a very long equity positioning right now, or short vol, or long carry in some ways. I am curious to get your thoughts on those issues.

Stuart MacDonald: I think the majority of risk parity managers out there are not jumping around wildly between different asset classes or whatever they may be exposed to. We certainly have clearly defined bands for our exposures, and therefore I am not quite sure where you are coming from on that. And I really wonder why people question the concept itself. We are directly aware of nearly 20 different risk parity managers, some of whom are running more than one variation on their theme. Some are not correlated to the others at all, because they all have their different takes on what instruments to use, what mix of asset classes they use, what liquidity they want, what measures they are using in terms of backward-looking risk to make adjustments to their exposures, if they make them at all.

I find most critiques of the risk parity concept slightly fatuous if one accepts the basic premise that effective diversification is the cornerstone of successful investment and if one believes, as many do, that this can be achieved only by selection at the level of broad asset classes rather than at the level of sectors or geographies, let alone individual securities. If one accepts that as an industry, we face certain limitations in our ability to make sustainably effective predictions and if one compares the performance of such strategies against most others, then I don’t really see how the critique holds. In any case, no one would sensibly suggest that everything is put into one strategic basket, however broad it may be.

Ewan Kirk: What we, or in fact all risk parity managers are saying, is that ex ante on a risk-adjusted basis, the Sharpe ratio of every asset class or every asset is identical. That is, of course, coming from the fact that it’s unknowable. Ex ante, I don’t really know what the risk-adjusted returns of bonds are going to be next month. So, if you don’t know, then yourbest guess is they are all equal, on a risk-adjusted basis. That means the Sharpe ratio of every single one of my positions is the same. If you are saying that, then that is risk parity. In fact, that is all we are saying and it seems like a reasonable thing to say. There is a discussion about how you do it, but I think it’s a great thing to do.

On systematic managers being long equities and long bonds, what would you like us to have done last year? Systematic macro is an uncorrelated asset class. If you are going to be uncorrelated to equities, sometimes you will be long and sometimes you will be short. We are not a hedge, so by definition there are going to be long periods of times where CTA managers will be long equities. In fact, you would probably expect CTAs to be long more than you would expect them to be short, because on average it appears as if equities have gone up, and so have bonds. Therefore you would expect CTAs to be long equities more often.

Akshay Krishnan: Right, I totally get that about systematic managers, but the question for maybe Stuart and Ewan: from a forward-looking perspective, how do you think about this potential rate rising scenario and what central banks might do next year? That is question one. And secondly, maybe this doesn’t apply to Cantab, but I have seen a lot of other CTA/systematic managers get into things like trading swaps in South Africa or swaps in Brazil or credit indices or power, and I do worry just about the liquidity impact of going into these markets, and will investors, who may think of these guys as a hedge, be surprised when we have an end of QE scenario?

Stuart MacDonald: Regarding QE, I cannot tell you how many investors and consultants over the last couple of years have been riding the consensus that emerged about imminent rate rises. The question has become, for how long does the consensus have to be wrong before you accept that it may be wrong? And one of the premises on which this is based is the difficulty of making forecasts with pinpoint accuracy.

Ewan Kirk: As Niels Bohr, the Danish physicist said: “prediction is hard, particularly about the future”. The received wisdom for the past four years has been that rates just cannot go any lower, but they have. And some CTAs have made money when it’s turned around. We in particular had a very difficult year last year when it turned around, but then it has come back again. We are not pure trend followers. But if we talk about trend following here…

Oliver Prock: When I hear “predictions about the future” I need to add here that actually we do exactly that in our Directional Market Strategy. The model forecasts each market one day out, and I agree, it’s hard! Ewan, excuse me but can I ask what exactly you are into, if you’re not a trend follower?

Ewan Kirk: We do momentum, value, risk premia, some short-term trading, some cash equities, so we are systematic multi-strategy. Now, if QE ends, then presumably your world view is that rates are going to go up and bonds are going to go down. If you really have that view, you don’t need a manager, because you can just go short bonds. If you have a very clear view that rates are going up and bonds are going down, to express that view through complex, heterogeneous CTAs doesn’t seem to be the most efficient way to do this.

Now, of course the reality is, if your view is true, then presumably rates will start to rise and bonds will start to go down, and presumably that will continue and that’s called a trend. And then you would imagine that trend followers would become short bonds. So yes, we have been long bonds and it worked out well. But that’s not the same as being a long bond fund. It just means that the trend in bonds has been up. CTAs will get short bonds, but they are not going to get short bonds tomorrow. It’s going to have tocause some pain, but eventually, if it really is the case that rates are going back to say 6% next year, which is unlikely, but presumably the models will pick it up.

Akshay Krishnan: And on the second question about starting to invest in less liquid markets?

Stuart MacDonald: I don’t see that happening.

Oliver Prock: Right, I don’t see that happening either since CTAs do not just add a market for the purpose of adding a market. All these exotic markets are actually extremely illiquid and limited in terms of history.

Ewan Kirk: There are a few things to be said here. It sounds better to some investors to say we trade 250 contracts rather than we trade 70, but it’s not really changing your profile very much. If you are a big CTA, most of your risk is in the 10-year bond, the Bund, the Euro STOXX, the e-mini S&P and the euro. If you are smaller, you can afford to go into smaller things, and that’s ok. But if you are a large CTA, trading the rolled oats contract doesn’t move the needle. I don’t see a reason why CTAs shouldn’t trade less liquid things as long as liquidity-adjusted returns are not affected.

Oliver Prock: Well, actually I prefer being called a quantitative manager than a CTA.

Stuart MacDonald: I can give you a good alternative euphemism for the term, CTA…

Oliver Prock: Okay, I am curious! So, when CTAs get big, the commodity exposures will be lower, because that’s typically the illiquid stuff, as Ewan pointed out. I have a question for Akshay. In your multi-manager macro fund, are you just investing in discretionary traders?

Akshay Krishnan: As a macro fund of funds we look at both systematic and discretionary strategies but we don’t feel compelled to have fixed sub-strategy allocations between the two. We think of all our managers having the ability to add diversification to our portfolio. The decision to reduce our allocations to systematic trend-following strategies is a view we have had for the last 18 months, and it’s not something permanent. Further, it’s important to clarify that in our case it was also a bit more specific to one or two managers where we were uncomfortable with their levered positioning, and this obviously varies by manager. We have been investing in macro strategies since the late ‘80s as a firm, so we are very familiar with the space, but we don’t claim to foretell what the markets hold.

I think what I do worry about is when I talk to other allocators, a lot of people are using systematic trend followers as a hedge within their book, and I am just worried about how they are going to react when you have a replay or an inflection point like we had last May and June, because it’s something that’s going to affect the entire industry, including macro in general. So it was just to throw out the question from that perspective.

Oliver Prock: I think I can answer this from my perspective. We are a quantitative manager, and trading very sophisticated strategies. I have gained experience in the CTA space since the early ‘90s, and I have done all the mistakes you can do. The current environment is a conundrum. It is very hard to understand for discretionary traders. Typically, I receive a couple of questions from allocators like “if bonds would start to fall, would you adapt your trading model?”, and in our case the answer to them is: the model has run for 11 years untouched, so we think it is robust.

But in any case, discretionary traders either need to adapt as well, or if they are not able to then they need to drop out. So why would it be bad to adapt a model? But as I said, in our case we believe we have a robust model for all market environments. So in fact we did not need to adapt the model, but what we had to change was the way we approach execution and slippage. We needed to make sure to not lose any alpha through execution.

Akshay, in your macro fund you invest in discretionary traders. I am having a hard time to find the ones that were really good, particularly during the last three or four years. With QE and all the central bank activity, the markets are in rather unchartered waters. When bonds go up, allocators want that we have bonds, right? So there is nothing wrong with making money in bonds when the price goes up, and I think there is nothing wrong to have a model that is robust and able to figure out when it’s the right time to short bonds. Having said that, our model forecasts prices and measures the forecast quality. It will short bonds definitely at the right point of time. However, I think it might take another year or so; who knows what the future really holds?

As a side note, I thought it was interesting hearing Ewan talking about Sharpe ratio. I thought physicists always say that the Sharpe ratio is a very stupid number.

Stuart MacDonald: People have a requirement to categorize things, and Sharpe is part of it.

Ewan Kirk: It’s not a bad measure of risk-adjusted returns.

Stuart MacDonald: It doesn’t have to work totally to be usable; there simply needs to be a common understanding of what it covers effectively and what it doesn’t.

Oliver Prock: We use a different measure of risk. It is called Conditional Drawdown at Risk and we optimize for that in one of our strategies. I actually think it is a good measure of risk, because it’s just focusing on drawdowns. We decided to focus on drawdowns since it is a number that actually matters for investors since nobody likes to lose money. If Sharpe is 0.7 or 1.1, this is secondary.

Coming back to the point whether models can adapt better to the environment or discretionary traders can. I have issues finding discretionary traders doing that. I think the best will always adapt, they will always be able to adapt, but the best are maybe 10% or less, not 90% of the discretionary trader bucket. As I mentioned, in our case we did not need to adapt the model, but we needed to get smarter in execution. If you have ever done a back-test, then you would have found that the price you input in your model is gone already due to lag 1 execution, i.e., on the next bar. But if you are able to get to lag 0 execution, i.e., on the same bar, your back-test or your trading will always be better than with a lag 1 execution, irrespective of what the model is. This is what we have basically done in our recent work. We have come as close to lag 0 as it is possible. And that really makes a difference in the current environment. We believe that execution is the part where some alpha is lost, which didn’t matter pre-crisis, but it definitely matters post-crisis, especially when the risk-free rate is at 0%.

Nacho Morais: I wanted to elaborate on something which we touched upon before, the fact that factor-based strategies are basically the sale of a back-test. In this context, I wanted to focus on how over-reliance on some factors can be adverse, if you do not take into account other variables. For instance, certain parts of the value school are partially based on the fact that during the last 30 years, we have been in a decreasing rates environment. In this sense, many people have made the strategy of investing in bond-alike equities into the cornerstone of their career. The secular down-trend in rates has provided a huge tailwind for these strategies. The fact is that, at this point, this is not replicable any more, simply because we cannot have another 15% reduction in bond yields. So if we extrapolated raw performance data for that strategy, we would not be right.

Also dealing with the data, another dimension to be taken into account is the market participants. If you are looking at something that worked for a prolonged period of time and try to draw a trading strategy based purely on the data, you have to take into account that the market is different now, and that the participants do not operate in the 2010s as they did in the ‘90s. Back then in the ‘80s or ‘90s, systematic quantitative investing was about second-guessing how the fundamental investors were deploying money, but now you have a large portion of the market that is trying to apply those kinds of techniques. So, there is a phenomenon similar to multi-colinearity, as there is a larger proportion of participants trying to second-guess market behaviour in a market which is already heavily participated by second-guessers, and that creates some kind of noise.

From a commercial standpoint, I think that one of the key issues here is replication, defined as whether an investor can replicate the activity of a manager in a cheaper way. On many occasions, we have seen some long/short equity managers behaving like the index or like a very simple mutation of the index. You can even see that phenomenon at the most simplistic end of the systematic strategies, like a trend-following manager that follows just a raw moving average strategy. You probably do not need to pay somebody 2 and 20 or higher fees, for something that you can replicate with an index or with very simple code.

Ewan Kirk: That is true, but it’s not all about the models – it’s about the execution. We can all write down a nice, simple trend model, which in gross terms looks great and appears statistically indistinguishable from the best CTAs in the world, but that’s in gross terms. At Cantab, we execute 12 times a day with a lag of 0. The signal comes out and less than a second later we are starting to do execution at that price, and we are managing it, thinking about it and optimizing it to fractions of a basis point. A huge amount of effort goes into saving tiny fractions of a basis point. Again, it is all about the implementation.
The analogy that I use is that we all know how to make a table: it’s one flat piece of wood with four legs, and that’s all there is to a table. But neither you nor I can really make a table – there is a certain craftsmanship involved. Just because a simple trend-following model does well, it doesn’t mean that trend following isn’t worth money. It’s worth quite a lot of money. If I go back to my statistic earlier, one basis point a day translates into 2.5% per annum. If we can reduce our trading costs by one basis point a day, that pays for the entire management fee, and you get an extra 50 basis points back.

Nacho Morais: What I mean is that replication sets a boundary on the minimum net returns that a manager needs to provide. If I can take a simple trend-following model and a trading algorithm, and put it into an ETF, with very low costs, that would be the minimum benchmark that a manager that was to launch a trend-following fund has to beat.

Ewan Kirk: You probably assume that this ETF, with a simple trading algorithm, will cost you less than one basis point a day to trade. But more likely, it is not going to be one basis point a day. Simple trading algorithms don’t cost you one basis point a day, not across 100 different assets, with volatility weighting, execution at different times of the day, relative value execution – all of that is really hard to do. This means that this ETF is probably going to lose money, because it’s likely going to be paying around five basis points a day, which is 12.5%.

Nacho Morais: I agree with you, but you are focusing on the reality of your own specific implementation of the strategy, with your own expertise, size and trading frequency. Talking from the buy side, I can tell you that I come across much simpler models with poor execution being marketed to us. For instance, and I am not talking about something happening in 2005, but this very week, I got a call from someone who is setting up a fund that implements a very simple – directly observable – strategy, rebalances it every week, and puts it in a fund format.

I guess the point I am trying to make is that either from a fundamental or systematic standpoint, managers who are applying very plain-vanilla strategies, even if they make some money, can be beaten by something passive, either an index or a basic investment rule.

Ewan Kirk: Yes, of course. Probably everyone around this table has been beaten by something passive – by the S&P 500 last year. That was a nice, simple strategy. If you can get into a nice, passive strategy at the right time, then that’s a free strategy that can make you 30% a year, with no management fee, no performance fee. But you have to be able to pick that.