Alpha & Information Management

Alpha and Information Management

Active investment management involves managing a lot of information both on the Alpha side and the risk/portfolio construction side.

Many investors mix returns and skill. While it helps to be skilled in order to deliver enhanced return, a skilled person can underperform by pure bad luck. And a non-skilled manager may deliver good value added by chance. I explain these concepts in detail in a paper on Manager's Selection.

We all dream of positively skewed value added. If we look at surviving managers, by construction they have positive value added even if/when no skill is present:

When we hear the marketing speech of many active managers, they all seem to have superior forecasting skill. The following chart illustrates the superior forecasting skill that many of them seem to have:

The reality is quite different. Typical first quartile managers have an average forecasting skill that looks like the following chart in smaller markets. For larger / more efficient markets, the correlation between forecast and subsequent returns is even lower, yet slightly positive for first quartile managers.

We should not despair though. Small skill can be leveraged over large volatility and if we apply our skill numerous times (breadth), we can achieve superior returns per unit of risk from skill and not luck.

But remember that even for a manager with a superior IR of 0.5, it takes some 16 years to be relatively sure (95% confident) that this performance was not achieved by luck! And for good managers with an IR of .25, it takes 64 years!

While it takes a lot of time to know that a manager’s IR is not from luck, it takes much less time to know that a manager is truly skilled if we have access to all his historical forecast which we use to measure his forecasting/investment skill. It is surprising to observe that many managers do not calculate such investment skill measure. Perhaps they should be reminded that once they start measuring it, they will think of ways to improve it.

For now, we observe many managers with great confidence in their unmeasured skill because they had some past luck. They think they have a superior forecasting skill:

“People exaggerate their own skills. They are overoptimistic about their prospects and overconfident about their guesses, including which [investment] managers to pick.” Professor Richard Thaler, University of Chicago

I observe the same behavioral error in all asset classes: Equities, Fixed Income, Real Estate, Hedge Funds, and even more in Private Equities.

I was speaking to a leading Private Equity manager from a Canadian mega-fund a few years ago. His knowledge of the Global Private Equity world was impressive. His network of high profile contacts was equally impressive. In fact, he reminded me of this guy:

As is usually observed in such cases, his overconfidence lead to negative surprises. Yet, I noted no proper risk measurement, no proper performance attribution and no proper risk budgeting. A year after my meeting, that manager posted a superb … underperformance by over 20%! He then left the fund to retire. It is not possible to underperform by that much without having taken an unacceptable level of active risk – yet active risk is easy to monitor, even with Private Equities.

Improving the information content of our forecasts

Once we understand that even for first quartile managers, our forecasting skill is limited, we can focus on improving it. As with any thing we do, it is hard to improve something we do not measure. But investment skill is measurable and return is only very weekly correlated to investment skill.

Armed with an incredible tool, a measure of our investment skill, we can work at improving it. Various tools and tricks to improve our investment skill include:

• Better valuation models
• Better industry classification or other stock groupings
• Better financial data including better pre-treatment of the data
• Better forecasting techniques
• Better signal and forecast combination techniques
• Better macroeconomic forecast
• Better behavioral science understanding and modeling

This leads to alpha forecast with higher information content than if we did not use this tool to improve our forecast.

Risk modeling and Optimal Portfolio Construction

When properly used, these improved forecasts will lead to more value added in the long run. Information Management does not stop here. An optimal use of these forecasts requires an optimal portfolio construction approach, which requires better risk measurement.

Optimal portfolio construction is a risk budgeting exercise. Every investor wants as much value added as possible, with as little risk. As these are conflicting goals, we need to balance our forecast and the positions we would like to take in our highest alpha assets, with the risk we need to take in seeking those alpha. Again, measurement is critical. We need a good measure of the risk – so a better risk model is critical. Intuitively, we know that such risk model needs to be closely link to how our forecasts are built. The rest is math: optimally balancing alpha and risk.

Finally, Information Management also involves better explaining what happened – i.e. better Performance Attribution. Many active managers still use the old story telling approach which generally hides the truth regarding their over or under performance. Proper performance attribution should closely attribute value added to the valuation models or other factors used in deriving alpha forecasts.

For more information on Information Management and how to build a better risk-diversified portfolio, please contact us.

Dominic Clermont, ASA, MBA, CFA