Certain quant strategies have had a torrid time since coronavirus tore through markets, but proponents of systematic asset allocation for risk-rated portfolios say removing the emotion from investing has benefitted them in the recent market turmoil.
Last week, the Financial Times flagged up quant investing pioneer Winton Capital’s flagship strategy has endured one of its worst runs on record due to its decision to buck a trend-following approach which has stood other quant strategies in better stead.
According to FE Fundinfo data, the Winton Global Equity fund has lost about 18% between 1 January and 31 March. One of the firm’s other strategies, Winton Diversified has delivered -5.7% over the same period. By contrast, Winton Trend, a Ucits fund, has delivered 4.8% in its sterling share class over the same period, while the dollar share class has returned 12.1%.
Recent crisis akin to the ‘quant quake’ in 2007
Poor performance has led some, including Goldman Sachs, to label the current Covid-19 crisis and its impact on markets as worse than the so-called ‘quant quake’ in August 2007 when investment algorithms created by Goldman Sachs’ Quantitative Investment Strategies division went haywire, resulting in rapid losses and fund closures.
According to Havelock London chief executive Matthew Beddall, who previously spent 17 years as chief investment officer at Winton, quants have been a mixed bag through the recent spell of Covid-19 driven volatility.
> See also: Quants buoyed by rise of data science
Beddell says the turmoil has resulted in some unusual market moves, pointing to large simultaneous falls in the gold price and the stock market when traditionally gold should move higher in times of stress.
“What we’re seeing is highly unusual,” he says. “It’s not something that anyone has seen in their living memory, the idea of just the economy being shut down.”
Market crisis is still in act one, scene one
But it’s too early to make any calls on how the epidemic could pan out and its effect on markets, he adds, noting the performance from trend following has been better but “still not out of this world”.
“My feeling is it’s still very much act one, scene one,” he adds. “The global financial crisis lasted a couple of years and so far we’re three months in. I think it’s still pretty early for anyone to be banking scores.”
The value-oriented long-only LF Havelock Global Select fund relies on a combination of human and machine in that it has a quant engine delivering the strategy but human oversight.
“What we’re doing with this combination of computer and person, we’re trying to have a process which allows for a bit more in the way of human context being put into decision making.”
The fund returned -20.9% in Q1, according to FE Fundinfo. Writing in the fund’s latest quarterly update, Beddall acknowledged the tough market conditions. “The share prices of all but one of the companies that the fund holds fell in the quarter,” he wrote.
Don’t argue with the computer
Unbiased Financial Planning founder Ian Hart recently launched a high yield fund of funds under the AI-Funds brand which relies on price momentum signals from quant provider QAS to switch in and out of high yield and into cash accordingly when markets get choppy.
The fund has delivered a return of 5.6% over Q1, buoyed by the recent rally in high yield, but Hart says it could have avoided some losses had he not buckled to under pressure and intervened with the quant process.
> See also: Ian Hart – I couldn’t find the perfect fund, so I created my own
Hart explains the fund got a daily signal to sell into cash on 24 February when the number of Covid-19 cases started to rise in Italy and the fund subsequently sold 40% into cash. But on 4 March there was second Black Swan event when the Saudis announced an oil price war with Russia.
“Now, what would a human do when the FTSE ended up 7% down on the day? You’d hope for bounce,” adds Hart. “Little did we realise that Trump would then shut US borders and Christine Lagarde botched the ECB monetary policy announcement. So it was too late, as there was a second leg down.”
Hart says he should have sold everything into cash after the initial signal but “as a human I thought, ‘No, I can’t do that, I’m crystallising a loss’”.
“This is why you need to take the human out of the process,” he adds.
Computer benefits risk-rated model portfolios
Those running risk-rated model portfolios also see the benefit of a systematic approach to asset allocation when markets get volatile.
Evalue uses a quant process to model different asset classes over the long term for the next 20, 30 and 40 years and generates a range of forecasts which it uses to create an asset allocation.
Evalue managing director index solutions Akshay Kapoor says: “It’s a completely model driven asset allocation, we don’t have any human intervention, apart from ensuring the governance works. We don’t put in subjective human inputs at all.”
The firm put out a press release last week saying during Q1 its multi-asset portfolio weightings shielded investors from the worst of the corrections of up to -30% in the FTSE 100 by going underweight the worst performing asset classes and overweight the best performing. For example, in its mid-risk portfolio the allocation to Japan was 25% compared with Japan’s 8% weight in MSCI World. The Topix fell by 12% during Q1 compared with the FTSE 100’s -23.8% and the All Share’s -25.1%.
Between 1 January and 24 March, the balanced portfolio, Evalue Standard 4, delivered an -11.8% return compared with -27% for the FTSE 100. Its highest risk Evalue Standard 6 portfolio has delivered -20.1% over the period, according to its own figures.
A systematic process is crucial in market dislocations like the current Covid-19 induced shock, Kapoor argues. “Going into the crisis we had zero allocation to US equities for about three quarters. Questions were asked on that, but we stuck to it.”
Human touch is necessary to fine tune asset allocation
Dynamic Planner meanwhile uses a mixture of human and quant for its risk-rated portfolios. It has a quant process that devises the asset allocation before it is scrutinised by the board.
“The only thing that we [humans] do is make the portfolios investable,” says Dynamic Planner head of asset and risk modelling Abhimanyu Chatterjee. “With traditional optimisation you might get 21.32% to invest somewhere, but you can’t invest 21.32%, so we’ll make that 20% because that’s how life works.”
> See also: Dynamic Planner details how decumulation funds have fared ahead of launch
Chatterjee also notes optimisation can throw out odd results which need human oversight. For example, a lowest risk, highest return portfolio might result in about 60% property, “which is ridiculous”.
“So the constraints are set by us because we are all experienced professionals who have been investing in markets for many years.”
Chatterjee notes in 2008 the balanced Dynamic Planner 5 portfolio saw a 30% drawdown and took 13 months to recover, compared with a 40% drawdown and a 41-month recovery period for the MSCI UK Equity Index.
