Winton’s global equity strategy
The West London offices of Winton Capital Management, best known for the diversified managed futures programme that has helped it grow into one of Europe's biggest hedge funds, feel more like a university campus than an HQ of an asset management firm.
There are no trading desks, for example. Instead, a small cadre of physics, mathematics and other hard-science PhDs sit at whiteboards covered in long, scribbled equations and diagrams. In a larger room a latte-sipping staffer throws his research open to discussion by a tableful of note-taking colleagues. Co-founder David Harding, himself a Cambridge natural sciences graduate, funds the Winton Professorship for the Public Understanding of Risk at his alma mater as well as a series of annual Royal Society lectures by prominent scientists.
It all makes you wonder how Mark Precious - who joined in 2005 to develop cash equities components in existing programs as well as an all-new long-only global equities strategy - wound up there. "My background isn't typical Winton," he concedes. "Soft sciences rather than hard - a lowly PhD in economics from Oxford!"
His thesis addressed integrating supply-side economics into investment theory, so he was steeped in Keynesian theory, ‘animal spirits', the impact of the exogenous on long-run pricing of assets, and other aspects of behavioural finance that were not so fashionable in the mid-1980s. At that time, the number crunchers were in the ascendancy, which is why Precious soon got ‘fed up with economics' as an academic discipline and decided to get his hands dirty with the real thing: after working for the Foreign Office in Argentina in the 1980s he moved to the IMF, and then in 1994 to SG Warburg as head of emerging markets and later global equity strategy.
"The mathematical models economists were developing, while powerful, do not go a long way to explaining a lot of things in the world," he says. "Once you introduce uncertainty, the mathematics has to become so much more abstruse and focus down on ever-smaller spaces."
But what about all those equation-strewn whiteboards? Isn't Winton a prime example of those ‘rocket scientists', whose myopic attempts to model the economic world with the laws of Brownian Motion led (as the popular story goes) to the financial crisis? Quite the opposite, in fact. Rather than building models that articulate how the world ought to behave, Winton extracts its mathematics from what it observes in market data. The philosophy comes from the empirical, hard sciences - not the idealising soft ones. Indeed, medium-term trend following, for which Winton is best-known, could be seen as the returns-seeking expression of many of the insights of behavioural finance.
"A really important step in my education at Winton was to realise that it is not a model-dependant firm," says Precious. "It's an ‘Is it there in the data?' firm. Winton is managed futures, a CTA - but what it really excels at is gathering and cleaning large amounts of data, and using it to trade financial assets." There is an affinity between this culture and Precious's career at the IMF, which was about making sure data was robust. SG Warburg was among the first firms to attempt pan-European equity analysis, which required major efforts to clean and standardise balance sheet data. "The frustration in investment banking is that you just don't have enough time or resources to test things like that properly," says Precious. That is almost Winton's raison d'etre. It also means that those core data-handling capabilities and processes can be standardised across the entire firm.
"While the global equity programme has been developing for five years, all of the infrastructure, processes and intellectual capital built up over 14 years is applicable to what we do," says Precious. "The three full-time data people who work with the equities team sit with the equities team but are not part of it - they are formally part of the Winton data team. The same goes for the two full-timers working on the simulations kit."
So what sort of fund emerges when one of the world's top managed futures firms rolls out long-only global equities? In a way, one could characterise it as an attempt to preserve the overwhelming advantage of the market cap-weighted portfolio (low turnover and therefore low cost) while doing something about its considerable disadvantages (it has no risk management model at all; and its expected return model is market cap-based, when size is no predictor of returns).
Winton's expected return models are a mix of fundamental and technical indicators in ‘Sub-Systems' underneath its ‘Base System'. The Base System weights the 1,000 stocks from the MSCI World index according to forecasted volatility (more in the higher-vol stocks), so that each stock contributes approximately the same risk to the portfolio, and re-balances monthly.
This is surprisingly simple and, while it definitively breaks the link between portfolio weighting and price (as many new ‘alternative beta products' do), it also goes with the flow of the cap-weighted portfolio in important ways: it allows each stock's respective volatility to generate weightings rather than trying to bend them all to a target volatility for the portfolio, for example, as a minimum-variance product would.
"Designing the strategy involved a trade-off between Sharpe ratio, information ratio and turnover," Precious explains. "Targeting lower volatility can deliver a higher Sharpe ratio, but you lose some absolute performance in the process - and actually that's not what investors want. What they want is slightly lower volatility than the index but similarly strong performance."
The return expectations models in the Sub-Systems, which add a positive or negative weighting on top of the weighting for each stock from the Base System, again with monthly re-ranking, continue this theme. One is fundamentals-based and uses traditional quantitative factors like P/B and P/CF ratios alongside proprietary signals to identify under-valued stocks. "That's where my experience comes in," says Precious. "Using financial balance sheet data was new and took about a year to get all that data onboard."
That clearly pulls against the momentum bias inherent in the cap-weighted benchmark, but again, the extent of that pulling (and extra turnover) is controlled. The strategy does not deploy mean reversion-based technical indicators, partly because those same signals are picked up by the fundamental value indicators (by the same token, analysts' estimates were dumped from the fundamental indicators because technical momentum signals picked out the same patterns just as well). But it is also because the shorter time horizons over which mean-reversion signals are effective would demand too much trading.
The sub-system technical return expectation models are therefore exclusively momentum-based. That is important, because they represent another part of the system that goes with the flow of the cap-weighted benchmark, which is inherently biased to momentum - it just does so in a much smarter way (drawing on Winton's knowledge of medium-term trend following).
This is probably why, despite the equal risk-weighting of the Base System (which accounts for 80% of each stock's weighting) and the negative correlation between the two sides of the Sub-Systems, the strategy does not diversify away market risk. It lets the market's overall volatility express itself, and it balances anti-market fundamental indicators with pro-market technical indicators.
The result: since taking on live capital in June 2008, its 24% volatility looks similar to the MSCI World Index's 27%; while its gain of 3% stacks up very nicely against the benchmark's loss of 15%. It had enough diversification to outperform the benchmark by 13.3% in 2008, but enough momentum to participate in 2009, when it outperformed by 4.6%. In that instance, the broad-based upside volatility of the market was captured by all three parts of the strategy - it is probably the only scenario in which all three become strongly positively correlated. The Base System definitively fights against the momentum element in the Sub-System (and with the fundamental element) only when the market is led with excessive strength by a narrow group of stocks - the classic bubble scenario. This is why sector weights can drift far from the benchmarks, while regional weights track it closely: currently, the strategy lags in financials; during the tech bubble of 2000, it would have severely underweighted IT. During these markets, the strategy will exceed its expected tracking error of 5-8%, and underperform: it would have lagged MSCI World by 11.2% in 1998 and 9.5% in 1999, before bouncing back when the crash came in 2000, outperforming by 22.8%.
Winton sets out to deliver a good Sharpe ratio by going with the flow of the market rather than against. And fees are compelling, too: institutions can get 0-and-20 for life until March 27, 2010, 0.2-and-20 thereafter, or even a flat 0.5% management fee if you want that. What sort of fund emerges when a CTA rolls out a global equity strategy? The answer is a little unexpected.