How many winners do you need? Richard Skelt from Fidelity explains how investment performance is actually generated and asks whether the concept of ‘batting average’ is a useful one for investors.
Suppose I am a holder of an investment fund which over the last 36 months has outperformed its benchmark or performance hurdle by 2 per cent per year.
This performance could have been generated by either a broad spread of positions generating on average a 2 per cent return, or alternatively the return could have come from a small number of positions that have paid off “big time”.
Imagine that this fund has 30 positions. At one extreme, each of the 30 positions might have generated an excess return of exactly 2 per cent, giving me my 2 per cent outperformance overall.
At the other extreme I might have one position that has generated a return of 60 per cent and each position in the rest of the portfolio may have generated 0 per cent, or perhaps my winning position paid 90 per cent and each of the other positions generated a return of -1 per cent.
Whichever way, my portfolio return is 2 per cent, but which is better?
Often people describe the number of winning trades as a proportion of the total number of trades placed as the “batting average”.
In the first example the batting average would be 30/30 = 100 per cent – in the latter example it would be 1/30, or about 3.5 per cent.
Most investors would probably rather see a spread of contributions, but some investment processes have surprisingly low batting averages.
Many technical strategies, or other trading disciplines, rely on stops to get you out of losing-trades fast and allow you to run with your winners.
There are many examples of successful strategies where the batting average may be say 1/3, but where the losing-trades are stopped before too much capital is lost and the performance of the winners more than compensates for the small losses incurred on the losing side.
This approach is not applicable across all disciplines: effective stops are a necessary part of the process, and these might be unworkable if for instance trades are blocked up or liquidity disappears. Stops may simply be out of scope: many value investors have a philosophical difficulty with stops.
If I buy a position at $2.50 because I believe it offers outstanding value, and the stock falls – if nothing appears to have changed other than the price – surely it is even more compelling as a value proposition now? And, rather than be stopped out, I may want to buy more.
If we would rather see a broader spread of successful investments in a portfolio, what is the right number for the batting average?
Is 51.3 per cent a good enough number? It is the house batting average for a roulette wheel – and operating a roulette wheel is normally considered to be pretty good business.
Remember that the wheel spins many times a night, many nights a year: provided the house can survive the occasional big loss to keep playing, the numbers work in your favour.
Quantitative fund managers know this well, and a feature of many quant funds is a large number of positions with a batting average which is not much higher than 50 per cent.
Risk-controlled portfolio construction and a diversified and wide spread of investments are key to making this perform.
As an investor in such a fund, I’d expect to see a pattern of relatively small positive and negative payoffs, but where the positives prevail and where over time the compounding works in my favour.
Rightly or wrongly, investors would generally like to see a higher batting average in concentrated funds.
There is a feeling that in a concentrated portfolio, the positions have originated through a process that marries high conviction with high levels of due diligence; the presence of too many failures would suggest that the process is not working as advertised.
For instance, investors may be prepared to tolerate a low batting average in venture capital, where the upside from the winners can be very high, but in the quoted markets where the businesses invested in are more mature, all other things being equal, a lower batting average and a concentrated portfolio can be a less comfortable combination.
Can portfolio managers control their batting average? Or more generally, can they deliberately target a less skewed spread of contributions in their funds?
It is difficult to do this directly because there is usually uncertainty in the outcomes of the positions taken – but there are some things that can help.
Most people believe that conviction is related to probability of success. In that case we can manoeuvre the portfolio so that there is a relationship between the amount of risk we are taking in the position (whether measured by contribution to risk, the percentage of “Value at Risk” or “Net Asset Value” lost in the event of a bad outcome) and the conviction.
Low conviction positions are more likely to decrease our batting average, and low conviction/high risk contribution positions are especially damaging if things do not work out as expected.
Thinking in this fashion should bring about a more uniform pattern of contributions across the portfolio over time.
Unfortunately, it is hard to see any substitute for skilled analysis, strong due diligence and good timing in order to increase the batting average.
Remember that a batting average is an average – and investors should be asking different questions of a portfolio manager who returns a fairly constant 55 per cent batting average year-in, year-out, as opposed to a manager who presides over a process that that swings between 85 per cent success and 25 per cent success, but on average yields 55 per cent.
Having a “super-normal” contribution from one or more positions is not necessarily a bad thing.
If a well-structured discipline is able to take advantage of one-off opportunities, perhaps brought about by unprecedented changes in market structure, regulation and suchlike within the investment opportunity set, then investors have no cause for complaint.
The key point is that in terms of articulating investment processes and managing expectations for investors, the strategy should be clear, and where appropriate the manager has to recognise a one-off for what it is, and place this in the context of more normal expectations for returns over the long-run.
For a potential purchaser, the discipline the batting average and spread of contributions are key metrics to consider.
Investors who see the majority of added value coming from a small number of positions in a wide portfolio will want to ask how sustainable or repeatable this pattern is going forward.
Napoleon famously wanted his generals to be lucky. We can take luck were we find it, but skill is more dependable over the long run. While there are many measures to consider, the batting average can give some good insights.
Richard Skelt is an investment strategist at Fidelity Worldwide Investment.
