Using Analytics To Improve Investment Process

By Cameron Hight and Justin Harris
‍

Data is taking a precedence in modern life as a necessity for those who want to improve. With specific relevance to asset management, finding insightful feedback in data is where managers can use unbiased facts to flesh out the inefficiencies in their investment process. As more and more managers rely on data to inform their investment process, the cost of not doing so increases. We, at Alpha Theory, have spent considerable time over the past year working with clients to help them hone their process, based on insights from data. Here are a few examples:
‍

PROBABILITY AND FORECAST ANALYSIS

Alpha Theory uses analyst forecasts to optimize portfolio position sizing. Managers must therefore know what confidence they can put on analysts’ research, because optimizing on bad research could be deleterious to returns.  Said another way, managers need to know who to trust.
‍

As a starting point, the average analyst in our system forecasts that they’ll make money on their investments 75% of the time compared to their actual success rate of 53%. Additionally, they forecast upside returns of 43% compared to their realized returns of 26%. Clearly there is a pervasive, systematic overconfidence. With that perspective, you can evaluate your own team.
‍

Probability Analysis. One of the first areas to evaluate is the forecasted probability of making money compared to the actual percentage of investments that were profitable. In the graph below, we show a representative comparison of forecast versus actual success rates. As a manager, this is excellent information because in one glance, you see which analyst is most accurate and who has the most bias. The actionable information would be requesting that Jimmy and Matt cap their forecasted success percentage at 50% and reviewing specific names with Jim to see why he has such dramatic bias.
‍

Additional analyses could include long/short and sector breakout or alpha-based analysis. These additional insights may help clarify the analyst’s differences. The point is that you ask questions based on empirical data that help your analysts improve while, at the same time, giving the portfolio manager the real-data to back up hunches of analyst bias.

Performance Analysis. In the graph below, we show how price forecasts (bars) compare to actual outcomes (underlay). We see that, while Jim Braddock had a disappointingly low success rate (graph above), his price forecasts were excellent. In fact, because the asymmetry was positive, his ideas were net profitable even though only 25% of them made money. From a manager’s perspective, encouraging Jim to make more realistic probability forecasts, gives you data that could dramatically improve fund performance.

Simple Graphs, Profound Insights. These two simple graphs provide many insights, as illustrated above. A few others to show the power of the picture:

1. Matt Doherty’s names should be given lower confidence given that he only makes money in 50% of his names and he loses more on his losers than he wins on his winners.

2. Ahtray Dahurt’s forecasts should receive higher confidence as his reward and risk price targets and probabilities are in line with actual results.

3. Jim Braddock’s forecasts are net profitable, but his reward probabilities are over inflated. There is an opportunity to profit from his ability to call names with extreme upside returns, but, in the near term, scaling back confidence while working with Jim to improve his batting or at least decrease his forecast probabilities would improve the forecast process.

These are a few of the many insights that Alpha Theory provides to clients to help them become better investment managers. Data provides insights that lead to action that result in an improved process. This chain of improvement is the benefit of capturing and analyzing data. Until we “see the data”, the answer may not be intuitive. This is why it is so important to create a data driven approach, focused on process improvements, that allow you to keep pace with the rapidly evolving data-driven world.