by Maz Jadallah
Articles Featured Featured-Homepage
[Updated June 2019]
Moneyball for Hedge Funds
If you’ve followed us for any amount of time, you know that our Clone Score methodology is at the heart of our manager selection process. The best analogy for our process to think about a manager’s score like it was a “batting average” where every stock the manager selects is an “at bat”. We believe the manager’s batting average derived from their actual stock selections that are disclosed in SEC filings is a much more effective way to identify skilled managers than analyzing a manager’s actual reported returns. Why?
A manager’s reported returns can arise from multiple sources besides just stock selection skill. If the manager uses leverage, makes a market timing decision and/or employs a unique trading strategy, those decision will all impact reported returns significantly. By using SEC filings to assess skill, we can eliminate the effects on performance of anything other than the manager’s investment decisions.
Our scoring process is very straight forward:
- We run multiple simulations (clones) that simulate investing in a manager’s disclosed holdings (e.g., largest 1, 3, 5, 10, 20 holdings, etc).
- We construct a profile for each manager which is an average of their clones’ monthly performance
- Clone Score then looks for two things; 1) by how much the manager outperformed the total market (the average batter) and 2) how often they outperformed
- We then repeat the scoring process every six months – once for the February rebalance and again for the August rebalance (i.e. the same managers selected in February will be used in the May rebalance while those selected in August will again be used in the November rebalance).
The rationale of our approach is simple –we’re looking for managers who outperform by a lot and do it often. In a word, persistence. Our process also seeks to avoid a few pitfalls;
- We run multiple simulations for each manager because we don’t want to be fooled by luck (e.g. the manager’s top 3 simulation does well, but his/her top 10 doesn’t).
- We want to avoid managers who did very well in the distant past but no longer (obviously)
- We want to avoid high performing managers with short filing histories – not enough “DNA” to rule out reversion to the mean.
- We want to also level the playing field between managers with track records of varying lengths (i.e. batting 400 is a lot harder for a manager who has a 10-year track record, than one who has a 3-year track record).
Our approach is certainly not the only way to leverage the 13F dataset. Other research firms have developed their own approach at identifying skilled managers. One main difference in our approach is that it does not seek to filter out a manager’s exposure to investment factors or sectors when identifying skill. The reason is simple, managers with good stock selection skill also tend be good at being exposed to the right mix of “factors”. While investors in traditional hedge funds don’t want to pay 2-and-20 for performance that is driven by factor selection, investors in our strategies pay the same relatively much lower fee for both the manager’s alpha and his/her dynamic factor exposures, so why not get the benefit of both?
Once Clone Scores are run, we select managers with the highest scores to construct our strategies. For example, our AlphaClone Select Downside Hedged strategy (available in separate accounts) and our AlphaClone Hedge Fund Masters index (available as an ETF), construct their portfolios from the same 10 managers with the highest Clone Score. Both strategies use the same “gene pool” of managers. Besides the dynamic hedge employed in Select, the only difference is that it is a little more concentrated with 30 holdings while our index has 50 holdings. Both portfolios are equal weighted at each quarterly rebalance.
But how often does our list of 10 managers change, what’s the persistence of managers on the list, and how have the investment styles of selected managers changed over time? To answer these questions and a few others, we analyzed our list of the ten highest scoring managers over the past 6 years. That’s 13 selection periods from Aug 2012 through Feb 2019. We summarize our findings below:
- Number of selection periods = 15 periods
- Total number of managers that made the list (over 7 years) = 68 managers
- Most persistent manager to appear = selected in 10 of 13 periods
- Average manager turnover = 5.5 of 10 managers
- Maximum manager turnover = 8 of 10 managers
- Minimum manager turnover = 3 of 10 managers
For those of you who consider an average turnover of 5 managers every six months high, consider that most manager research concludes that even managers who have demonstrated stock selection skill, will show persistence over relatively short periods of time. In fact, this turnover and the absence of the usual related “manager switching costs” (e.g. lock ups), is an important aspect of our methodology because it allows our strategies to stay aligned with top performing managers. To put a finer point on it – the table below illustrate which styles featured most prominently over the past 7 years and more importantly, which mix of investment styles were selected over time:
Table 1 – High Clone Score Managers: Investment Styles Over Time
Manager selection is difficult during the best of times. That we have been able to return the performance we have and at a time when active managers have generally struggled, is a testament to the process we’ve created. Best of all, investors in our strategies will never have to pick another equity manager again, they will avoid the usual behavioral risks (e.g. performance chasing) and switching costs that can come with selecting active managers, all while still retaining the efficiency and benefits of a passive, rules-based investment approach.