< Back to Research

Anyone for Alpha?

Raphael Fiorentino
28th October 2020 - 4 min read

Despite unprecedented (and still rising) levels of financial asset price manipulation by central banks since our note on mindless passive investing two and half years ago (and its follow-up parable “of elephants and mice”), we feel comforted in our view that going after alpha remains the superior (ex-post and ex-ante) alternative, especially when Butterwire is added to one’s equity research workbench.

The UK is a case in point: the FTSE 100 index generated a total return (assuming reinvested dividends) of -17% over the past 3 years (1-Oct-17 to 23-Oct-20), which does not even compare favourably relative to the +10% return that could have been expected from randomly picking stocks amongst the index constituents (see the green line in the graph below).


Conversely, running a truly active portfolio with Butterwire’s assistance delivered considerably better results. For instance, simply buying BW’s basket of “interesting and recession-resilient” large cap (UK) stocks (in roughly equal amounts) as of 30-Sep-2017 would have returned +27% (dotted purple line in the graph above). Rebalancing such a basket at the end of each quarter (i.e. selling the holdings that were no longer flagged as “interesting” and buying those that had become so) returned +46% (purple line in the graph above). This was attainable without even taking advantage of the other available features such as stock alerts (e.g. to spot exit candidates), global macro signals (e.g. to inform asset allocation and portfolio positioning), or portfolio risk-targeting (e.g. to mitigate unwanted factor exposures).

The basket resulting from applying such criteria on the Butterwire platform (e.g. candidate status, recession-resilience score and size) typically yields close to “truly active” portfolios (small number of holdings, high idiosyncratic risk, low factor timing risk), as illustrated with the latest list of UK holdings (see table below), but it is very much recommended to run the selection through Butterwire’s portfolio assess feature and mitigate any overly salient or unwanted factor exposure (from industry to market beta to valuation to interest-rate sensitivity to etc.).


Over the past 3 years, 136 UK equities have been “in play” (with 125 of them still publicly traded) and a total of 72 have been held in Butterwire’s UK basket at some point.

The basket’s average number of holdings at any given time has been averaging 18 and holding periods have been averaging 6.4 months.

The largest performance contributors over the period have been JD Sports, Polymetal and IG Group whereas British Airways, ITV and Meggitt detracted most (see graph below of relative returns over the holding periods).


The percentage of holdings outperforming in a given month (i.e. the % win rate, which correlates highly with excess return and explains 66% of its variance, see graph below) has been averaging 60%, with levels as low as 21% and as high as 89%.


While in this case the 60% average monthly win rate and 1.6% average monthly excess return over 3 years track clearly above what Butterwire set out to extract (respectively ca. 55% and 0.5% over 5-10 years), it is in the ballpark.

More importantly, it reflects how Butterwire’s approach is positively biased toward beating a passive strategy in the long (highly uncertain) run, thanks to its dynamic focus on relative fundamentals and controversies vs. size and hype.

To illustrate, we simulated 1,000 portfolios ran over 10 years. Monthly market returns were generated to yield an expected value of +0.3% (+3.5% pa) with high volatility and a fat left-tail distribution (see left graph below).


Excess monthly portfolio returns (alphas, see right graph above, expected value of 0.35% per month) were formulaically derived from monthly win rates drawn from the distribution below (expected value of 55% with reduced likelihood of sub-40% occurrences thanks to Butterwire’s algo, see graph below), as well as epsilons (highly volatile, fat left tailed distribution variable) to reflect the unaccounted variance between win rates and the alphas.


As the summary graph below illustrates, in the long run a truly active strategy, even on conservative alpha and left tail distribution assumptions, beats a passive one (dotted line) about 75% of the time. Or put another way, it delivers a large gain 50% of the time (index above the 50th percentile value of 119.6) and 50% of the time offers an equal chance of ending up on the winning side (index between 86.7 and 119.6) and the losing side (index between 56.4 and 86.7).


“Out with falsely passive. In with truly active”. This has been our core message from the start and one that we have been heeding with an average monthly excess return since 2017 of +0.6% across all regions. So, anyone for alpha?