Hamlin Lovell: What distinguishes Cantab from some other systematic investment managers that GAM might have partnered with?
Ewan Kirk: We now have nearly a ten-year track record and have performed well against our peers. The Quantitative programme (running c $2.7bn) has annualised at 8.1% since March 2007, and the Core Macro programme (running c $1.7bn) has annualised at 7.9% since January 2013. We had exceptionally good years in 2008 and 2014. We found 2013 difficult and underperformed by a considerable amount, but learned things and changed things as a result.
Core Macro is in the vanguard of the low-cost products. Some people jumped on the low cost bandwagon with less sophisticated products, whereas Core Macro runs on the same infrastructure as the Quantitative programme, using sophisticated models and risk management tools, and has performed exceptionally well over the past 3.5 years.
We have always said that one of the advantages of our quantitative programme is that it is not a strict trend-following CTA – it is something different. The firm has stuck to its scientific-based approach to investment, technology, and risk management, and has built a platform that enables people to do new and innovative things.
THE DEAL STRUCTURE
HL: Before agreeing to the acquisition by GAM, did Cantab talk to other potential acquirers and explore other types of deal structure?
EK: Part of running a firm is to think about strategic direction, and our CEO Adam Glinsman [formerly COO and Partner of Lansdowne Partners] does that very well. We thought about structures to enhance the stability, growth and longer term future of the firm in a number of different ways. We received a number of offers from acquirers with different structures.
HL: Which professional advisers assisted you?
EK: Barclays on the investment banking side and Schulte Roth & Zabel International LLP as lawyers. [The London office of Schulte Roth & Zabel International LLP (SRZ) represented Cantab Capital Partners LLP. The SRZ team was led by Jim McNally, investment management and M&A partner, and Christopher Hilditch, co-head ofSRZ’s London office and investment management partner. The team also included tax partner Nick Fagge, regulatory & compliance partner Anna Maleva-Otto and associate Anthony Lombardi.]
HL: Was GAM the highest bidder, or the best fit on a range of criteria, or both?
EK: It was not a question of the highest bidder, and anyway, not all offers were completely contemporaneous. The primary driver for the transaction was not the economics of the deal per se – GAM was the best fit. We were convinced of a complementary fit between GAM and us in terms of brand, global reach, and sales force. We don’t have 80 salespeople and offices around the world as they do. They didn’t have a scalable systematic business so it was a very good fit for them too. It was also a good match culturally.
HL: What aspects of GAM’s culture chime well with Cantab’s?
EK: We always serve our clients well and do investor relations well, as this is basic blocking and tackling [Dr Genia Diamond, who heads up Cantab’s Business Development, was selected for THFJ and EY’s ‘Leading 50 Women in Hedge Funds’ survey in 2013]. Client service has to be done really well. GAM is also very supportive of our style of research and trading, and very understanding that we are experts in doing this, and so should continue to have autonomy over systems, software, hiring, exploring research avenues, and make ultimate decisions on products.
HL: Even so, the price appears pretty competitive. The up-front valuation of $217mm works out at 4.4% of current assets of $4.4bn and the potential price of $292mm, including the $75m earn out, works out at 6% of current assets. Even ignoring the fact that Cantab partners retain 40% of performance fees, the existing and potential consideration seems to compare favourably with other deals where amounts were disclosed. Would you agree?
EK: We did not look at comparables. This was not a cash-out deal to maximise money for the firm, it was a strategic decision for the long term and successful future of the firm.
HL: Why is deferred consideration based only on management fees and not on performance fees?
EK: The existing partners retain a share of performance fees and all staff are highly levered to performance fee income so performance fees continue to be a significant incentive and ensure alignment as they always have. The deal is about building long-term, stable revenues with growing long term management fees. It would be irrational to base long-term valuations on more volatile performance fees. One great year in three does not make us worth more – it is just part of a very broad probability distribution.
HL: Was it vital for Cantab partners to retain 40% of performance fee income?
EK: It is certainly important to be aligned and incentivised – that’s why the performance fee model works.
HL: How many Cantab staff are ‘locked in’ with long term contacts offering revenue shares? Are client-facing, operational, administrative and clerical staff included?
EK: Everyone, from me to assistants.
HL: Cantab partners are recycling most of the sale proceeds into their own funds. Have they always done this?
EK: Yes we have always recycled profits into the funds.
HL: Is any current or future remuneration based on GAM shares, which are listed in Switzerland?
EK: It’s a topic we have discussed and something that may be put in place in the future. It’s certainly an idea to consider, including the need to meet any applicable regulatory obligations.
HL: Do you have any view on why shares in GAM Holding halved over the past year from 20 to 9 Swiss Francs as of August 2016? We realise that valuations of publicly-listed alternative managers have generally been under pressure, but GAM seems to be lagging the peer group.
EK: As a quantitative manager, I have no views on specific share prices. But the industry is under some pressure. GAM is in the middle of a very public turnaround so investors are waiting to see the results. But I must repeat, I am not a stock picker.
HL: GAM’s market capitalisation of CHF1.5bn appears to be only about 1.2% of its own assets of CHF 119b, in part because long-only fees are lower. On an EV (enterprise value) basis, roughly 20% of GAM’s EV of CHF 1bn is being spent on the deal. GAM clearly has big ambitions for the acquisition – what are the long term targets for asset growth and profitability?
EK: GAM does not disclose that for specific divisions.
GAM SYSTEMATIC AND SYNERGIES
HL: What is the vision for GAM Systematic and how is it structured?
EK: Currently, it is Cantab to a great extent, though there is a range of risk premia products. Broadly, GAM Systematic will be an umbrella for all systematic trading within GAM, and we are hoping for synergies between styles of systematic trading.
HL: What is the balance between Cantab remaining autonomous, and sharing knowledge with other discretionary, or systematic, managers on the GAM platform?
EK: It is not a question of autonomy. We retain autonomy as it is over models, risk management, software, technology and data as part of the deal.
HL: What will each side gain most from the deal?
EK: We have a very strong technology base, trading infrastructure, execution infrastructure, and risk management infrastructure, all of which could be attractive to other parts of GAM. But I particularly look forward to the flow of information going the other way. Discretionary managers have interesting views of the world, and see things that a pure systematic manager would miss. The ability to talk to discretionary fixed income traders and credit traders, and understand more about markets in more depth would be interesting. We trade nearly 200 futures, forwards and interest rate swaps as well as 2,500 stocks. Almost by design we cannot be an expert in the idiosyncracies of each one and, indeed, many of our strategies are about discovering the common features of all markets. So the ability to talk to discretionary peers, who run much more concentrated portfolios, may provide valuable insight.
HL: Could GAM’s discretionary managers provide the genesis for future model development?
EK: Possibly, yes. It is an interesting question to ask: where does a model come from and who generates the idea? We have developed scores of models over the past ten years. Ideas do not come fully formed – you cannot just throw data into a giant machine-learning system. We need to see a long term persistent source of return, and financial hypotheses behind each individual model. This can come from academic papers, or talking to clients. It can come via broker research, shipping or weather data, etc. There is something valuable in talking to people who live and die by a particular market.
HL: What are the plans for product launches?
EK: We only have two products for now (the Quant programme and Core Macro). We are launching Core Macro as a UCITS when the deal closes and then an Equity Market Neutral product. This is not a product proliferation play, so we are not throwing 35 products at the wall to see what sticks. At the same time, one of the attractions and benefits is our ability to scale – our operational leverage is very high. Our systems, infrastructure, models, quantitative techniques, risk management controls are all applicable in more than one place. So over time,it will be possible to build products for specific client types.
HL: Some Cantab strategies are more scalable than others. What level of assets could Cantab run across all current strategies? You mentioned $10-20bn on the conference call discussing the GAM deal, does that include long-only launches?
EK: The flagship CCP Quantitative programme has capacity of $4bn and we have said that from the start. It is relatively fast and trades some exotic, less liquid futures, such as milk and cheese in Minnesota with whey behaving as a spread between cheese and milk. These markets are not very liquid so capacity is constrained and you need sophisticated, well designed models to trade them. We also trade German power, interest rate swaps and options, so are comfortable with capacity of $4bn. Core Macro, in contrast, trades a cluster of trend modes, and a value/carry style cluster, all of which are, by design, more scalable. Its capacity is estimated at $10-15bn. So before adding additional products we see scale opportunities that will allow us to grow comfortably at a measured pace, as we have done in the past.
HL: What would capacity be for any long-only launches?
EK: As yet we have no launch plans for long-only products. But to provide some perspective, high frequency statistical arbitrage is not scalable whereas long-only is extremely scalable, judging by the trillions run by some passive index providers. There is a trade-off between scalability and alpha, and also a business trade-off between scalability and fees. Products running $20-50bn tend to deliver less pure alpha to end investors, and so should attract lower fees. In contrast a super high frequency stat arb programme can command high fees as it is pure alpha of very high quality – but capacity-constrained perhaps as low as $250m. But we realise that there is a completely different set of buyers for a robust, long-only system which outperforms over the cycle.
HL: Would Cantab potentially develop long-only products by simply using the long book of existing strategies, or would the models need to be changed for long-only?
EL: Our current models are a mix of trend, value and carry, with the market-neutral equities cluster another diversifier. We use very sophisticated proprietary strategies, portfolio construction and cost control techniques. This is all integrated into the cash equities cluster to create an equity market neutral product. To just take off the short book does not leave you with a hypothetical long-only product. A car is not a motorbike if you just take off two wheels, as the two have different design constraints. Similarly, going from a highly optimised book to a long-only strategy would be a different exercise.
HL: Clearly Cantab’s systematic expertise is very complementary to GAM’s renowned active discretionary offering, and the performance profile is lowly correlated or uncorrelated with GAM’s other products. Do you expect to see hybrid discretionary/systematic products or will the two be kept distinct?
EK: I don’t understand the concept of a hybrid product. Systematic trading is like being pregnant – you are or you are not. If you add a discretionary overlay to a systematic product that is strictly separable into a systematic and discretionary. Why not run them separately? You are just running a systematic signal alongside a discretionary one. So hybrid discretionary/systematic trading is just discretionary by another name.
HL: Do you think you would already have $10bn of assets had you been part of a larger distribution network?
EK: It is hard to argue a counter-factual. Distribution is important in any business but it is not a panacea. Just having 80 relationship managers serving institutional and intermediary clients, and over 100 staff inproduct development, marketing, client servicing and product specialist roles – as GAM has – does not guarantee raising assets.
HL: GAM is ranked as the world’s top three manager for liquid alternatives – by which firm and on what metrics?
EK: In terms of AUM and number of products. The source is Absolut Research (a very reputable German industry source).
HL: In how many European countries are GAM’s UCITS distributed? In how many countries outside Europe are they sold eg. Hong Kong, Taiwan, Singapore in Asia? Eg. Chile, Peru in Latin America?
EK: GAM UCITS are distributed in pretty much all European countries, plus the UK and Switzerland, and the other countries you mention in Asia and Latin America. For retail investors, as you know, authorisation needs to be given by local regulators even if it is UCITS fund – so public/retail distribution may vary from country to country. But GAM certainly sells the full range to institutional clients in all of the markets aforementioned.
HL: ETF assets are now roughly equivalent to hedge fund assets at around the $3 TR mark, but most ETFs offering low-fee beta are perceived as the polar opposite of hedge funds aiming to deliver high-fee alpha. Do you expect to see actively managed quant ETFs?
EK: An ETF is just a delivery vehicle for something, just as a ‘40 Act or UCITS is just a delivery vehicle for something. A quant fund is just a set of rules as is a gold ETF that tracks the gold price. Most ETFs have rules embedded in them. ETFs are already systematic, although in general they have fairly simplistic rules. In principle I see no reason to use more complex rules in ETFs but I’m at a loss to know why an ETF structure would bring any advantages.
HL: To what, if any, extent will GAM help Cantab with recruitment?
EK: We have fabulous recruitment infrastructure already. We get 2,000 CVs a year, interview 200, and hire five. This is exceptionally important and has been from day one. We have specific requirements, know what we want and have much experience in interviewing and identifying the best of the best. However, being part of a larger organisation while still retaining the small company culture should be positive for recruitment.
HL: Are quant funds competing with technology firms for the best talent?
EK: Yes, but that’s a universal truth. We are always competing and always want the best people. It would be interesting to see how many mathematicians, programmers, and physicists are created worldwide by the education system. It is in the millions. A lot of them are very good indeed. Though there is competition it is not a strictly limited market; quant firms are going after a small number. Some people interviewed would be better working at Google, and we tell them that. With others we say they should have been academics as they would be happier. We want our hires to really enjoy the job.
HL: Softbank’s bid for ARM Holdings has reminded us of the value of expertise residing in Cambridge. What are the benefits of being located in Cambridge?
EK: ARM Holdings is only one of a great many technology firms here. Cambridge is a technology hub for Europe – not just the UK. There are exceptional high technology firms here. We benefit peripherally from the location as we are based in the region with the highest density of scientists, programmers, and start-up companies. It is a great place to be as a scientific firm. The world’s best university is on our doorstep – Cambridge is number 1 or 2 on rankings. And I went there.
HL: And Cambridge has lately benefitted from Cantab’s charitable giving. You are philanthropic both personally through you and your wife’s charitable foundation, and at the corporate level via Cantab. What is the motivation behind Cantab’s GBP 5m funding for the Cantab Capital Institute of Mathematics of Information, at the University of Cambridge? Is it designed to provide research for Cantab or train up scientists to hire?
EK: It is purely philanthropic. There are no strings attached and we are not directing the research. The motivation is that it is much harder to get basic science funded – and even harder to get basic maths funded. (In contrast, biochemistry, engineering and computer science are relatively easy to find funding for). Cantab gets good karma for doing a good thing. Doing basic research into the maths of information is very hard, and funding research may be useful on a returns basis, but who knows in future? We are not outsourcing research – they decide what to do and they revert with a proposal. They appointed a director, hired two lecturers, and started a PhD programme. The institute also has a rolling cohort of six PhD students, so in three years’ time there will be 30-40. It is cross-disciplinary, combining maths and computer science to address big data problems. This is the rock and roll of science and it is fantastically important.
Finance has been the original big data problem since the early 80s, when Reuters sent high frequency data on those old fashioned green screens. At Cantab, we pride ourselves on our high performance data storage and analysis infrastructure, and we are good at dealing with big data, storing and analysing it. We might be able to help out post-doctoral students and PhDs with access to data and might offer some of them internships, but there is no necessity to do so.
DATA AND TECHNOLOGY
HL: Will GAM increase access to market data eg on prices, volumes, sentiment, etc?
EK: Our CTO, Tom Howat, is in that room with GAM’s IT people working it out right now! As a 55-person hedge fund we have been somewhat parsimonious on data as we need to be fairly certain it is going to be generate profits for our investors. I assume data sources in GAM are useful to the people there and even if they are not central to our business, we may integrate it if useful. I do not expect any super-secret alpha, but more data is better than less.
HL: GAM mentioned technology as an attraction of Cantab. What are the key benefits of your technological infrastructure?
EK: It is scalable in a way that people are not. A discretionary trader may only grasp a handful of positions – eg. 20 of them – and does not have enough time to focus on much more than those positions. Good discretionary traders are often extremely concentrated. Technology lets a computer analyse 2,500 stocks and 200 futures all day long without getting tired or bored. This gives us scalability in terms of size. Core Macro runs $2bn, and could easily run its capacity of $15bn. That scalability, and operational leverage, is all down to technology. We do not run on MATLAB, ExcelE and CSV files. We have institutional quality, industrial strength, software and this is critical. We regard technology as a core competency for a systematic hedge fund.
HL: In addition to scalability in market coverage and assets, what other benefits come from your technology infrastructure?
EK: We can easily customise. In principle, we can run long-only products or customised products, and solutions for individual clients, rather than a one-size-fits-all approach. This makes more sense. We understand clients’ needs and each one has a different portfolio. Systematic Macro is in general a great addition to most portfolios, but can be optimised around long-only portfolios to better match underlying assets, or liabilities, of clients. How can you have 100clients with different requirements without fantastic software? You cannot do it in a spreadsheet. Better infrastructure in terms of systems, risk management tools, models and everything else is needed. Technology is the core component and here, GAM is also very important. It is not hard to find somebody with a model – thousands of people have models. The key is to translate models into a system that can trade reliably, robustly and reproducibly, day-in, day-out, every day. That’s where most of the work is and we understood that right from the start.
HL: Cantab co-founder and CTO, and your former colleague from Goldman Sachs, Erich Schlaikjer has retired. Has he left the industry? Can you comment on Dr Tom Howat, who replaced him as CTO?
EK: Erich has left the industry. Tom has taken over the reins and was closely nurtured towards the job for three or four years. He has exceptional talents in programming, team leadership, strategic vision, software architecture, and appreciates the primality of technology in our business. Tom was our second hire and the first or second person we interviewed back in 2006. He has been a huge part of the success of Cantab over the past ten years, and will be a huge part of our future. You should check out quotes from Tom on our site.
HL: You trade 150-200 futures markets and over 2,500 single stocks? Will GAM help you to access more markets eg. through counterparty, exchange or venue relationships?
EK: We already have good relationships with execution brokers and prime brokers, so market access was not a prime driver for the deal. For instance we already trade practically every future we can from the 10 year bond future to more unusual ones such as cheese, rice, German power, freight and coal. There are not that many more markets to trade on the futures side (and Chinese commodity markets have their own challenges). For most markets there is only one execution venue – an exchange. We cannot think of any developed equity markets that we do not trade. As we are not doing large block trades, dark pools are of less interest. We do not currently trade cash bonds or credit so GAM might help with them, however.
HL: How are you expanding your investment universe?
EK: We are broadening the asset set with more esoteric assets, such as cheese, freight and coal that are very different from existing ones. They do not have the same liquidity but are uncorrelated – cheese is uncorrelated to everything apart from milk. Then the quality of the “incremental” alpha is very high. Our research project to add interest rate swaps started early 2016, and already a third of our interest rate risk is now in swaps. The objective here is not to get more capacity. There are plenty of super liquid contracts in developed markets but all of them are effectively doing the same thing, so there is no point in adding exposure to USD, EURO, JPY, GBP.
In contrast, Australia, Canada, New Zealand, Sweden, Mexico, Czech Republic, Hungary and Singapore all have relatively illiquid bond futures markets or no bond futures markets and are very different markets. Cantab is trading interest rate swaps in these markets and will soon be adding listed options on the more liquid equities, interest rates, commodities and currencies futures. We are now trading volatility as an asset class. We could add credit in principle. All of this involves new sources of data as well.
FUTURE RESEARCH AVENUES
HL: Big Data, Artificial Intelligence, and Behavioural Finance are all becoming buzzwords in quantitative finance. Can you give any hints or clues on future research avenues eg. news and text analysis, sentiment analysis?
EK: I am wary of buzzwords, especially in AI, which is hard to pindown. But research into techniques such as machine learning or AI is just a technique. No new technique will turn FX carry from a Sharpe of 0.7 to a Sharpe of ten, and no trend-following system has a Sharpe of three. But many of these statistical techniques are about multivariate problems. Traditional systematic trading models, such as trend following models, were just univariate models with inputs being facts and time series about one market. More sophisticated machine learning techniques let you take a multivariate view of the world, and forecast the joint distribution of a group of assets which might be 10 futures or 1,000 stocks. This joint distribution tells you more interesting things about each market versus the others. Then it is as if you had 100 humans looking at an individual asset rather than having one trader looking at one market without forming any view on the others. Machine learning and multivariate techniques can look at all the assets simultaneously and produce a coherent view.
HL: In what other directions is your research heading?
EK: Alpha capture data has historically focused on single stocks where most work has been done in areas such as broker recommendations in terms of which forecasts are better or worse. It is more unusual to apply alpha capture to macro data. This is a project.
HL: What proportion of your research projects lead to actionable results?
EK: Many research projects don’t come to fruition. We are big fans of null results. Investigating a new source of return or market technique – and coming up with a new profitable, accretive model – is good. But discovering that there is not any return to a project is also a good thing. It is all part of the scientific method. We fully expect some projects to fail, as it shows interesting things. If you are not doing something that is hard you are not going to fail.
HL: Which of your research projects seem most promising?
EK: I do not know yet. If I knew the answers, I would not need to do the research!