A classic discussion in the markets revolves around the question of whether technical or fundamental approaches work better for discretionary investment strategies. However, this is not the subject of the Winton article mentioned[1]. An attempt is made to make an analogous assessment on a systematic level. In absolute terms, this means that clearly defined, price-based concepts such as trend-following strategies are compared with models that are also systematic but developed using fundamental data.
Spongy fundamental data
The crucial problem: Fundamental data, unlike actually traded prices and turnover, are associated with a certain degree of uncertainty. Thus, individual data points can be inaccurate and may have to be significantly corrected later. This can already have a significant impact in the strategy development phase, for example, on the reliability of a backtest that is not based on high-quality point-in-time data.
Winton now argues that Covid-19 has considerably worsened the validity of many economic data models. The near shutdown, a simultaneous supply and demand shock, and extreme data points in statistics on employment and other economic indicators are unique events that can upset relationships previously considered solid. Added to this is the enormous influence of monetary and fiscal policy. According to the article, it was only in the 1940s that a similar situation existed as a result of the Second World War.
The corona crisis brought unprecedented swings in some key economic data.
Sources: Winton Group (2020), Price-Based Strategies at Times of Fundamental Uncertainty; Federal Reserve Bank of St. Louis; Bloomberg here (optional)
According to Winton, the rapid and extreme changes in the global economy make it even more difficult than before to gain a systematic advantage based on fundamental data. For example, the development of corporate profits and thus share prices are highly dependent on whether and to what extent central banks and politicians (continue to) intervene. The corresponding expectations play a major role, especially for share prices. These can be assessed much more quickly and directly on the basis of the actual price trend than is possible with potentially inaccurate, lagging fundamental data.
The disadvantages mentioned above speak against fundamental strategies - but especially in the current market environment. The range of some much-noticed estimates for the current year alone is beyond the scope of this report. This is a new challenge for all asset managers, but especially for those who use models based on interactions in fundamental data. Stable correlations previously considered fundamental could be softened or even become irrelevant. This would mean that the corresponding models would be fundamentally questioned.
Estimates for the year 2020 show extremely high ranges for some economic data.
Source: Winton Group (2020), Price-Based Strategies at Times of Fundamental Uncertainty; Bloomberg; Wall Street Journal
The advantages of price-based strategies
There were also some extremes on the price side: for example the historically fast, deep "waterfall crash" on the stock market, the fall in the yield of 10-year US government bonds to below one percent and, for the first time, negative prices for WTI oil futures.
However, data based only on prices and turnover have a decisive advantage: uncertainty is eliminated. These are reliable inputs on the basis of which all traded assets can be continuously valued. This can be optimally implemented in trend following models and other systematic trading strategies. This is because, unlike fundamental strategies that often implicitly make forecasts and then position themselves accordingly, price-based approaches can do without analysts' estimates and other inaccurate data.
In terms of performance, trend following strategies in particular often performed particularly well when fundamental estimates were very poor. Winton illustrates this in the article with a few selected examples. For this reason, trend-following strategies allow good diversification compared to approaches based on fundamental inputs. The positioning of the two approaches may well contradict each other. This is also related to the fact that trends often emerge when a consensus opinion slowly changes. In systematic trend-following strategies, the continuous, systematic adjustment of exposure to each new data point means that there is no problem in determining the "optimal" exit, which is often attempted (in vain) in discretionary technical strategies. Position sizes are usually determined simply by the strength of the trends and their volatility.
Conclusion
Among systematic strategies, course-based approaches should prove more reliable than fundamental models. The latter are subject to increased uncertainty about the accuracy of the data basis and the (implicit) forecasts derived from it.
