• wblogo
  • wblogo
  • wblogo

Where Investment Alpha Is Lost And Found

Chris Woodcock, 2 July 2020

articleimage

This article identifies factors most commonly associated with alpha generation and destruction by equity fund managers. It gets to the heart of the debate about how or whether active fund management can add value, and if it is worth the money.

The end of a decade-long bull market in equities, followed by this year’s coronavirus-induced declines, is an auspicious time to review a long-standing quest: where do above-market returns – “alpha” – come from? How can managers find consistent ways to achieve superior returns? A firm that takes a look at this is Essentia Analytics and we have a guest op-ed article from Chris Woodcock, who leads the research and product teams. This news service also recently spoke to Woodcock’s colleague, Clare Flynn Levy, the chief executive, who explained why Woodcock’s insights are so important: “Every fund manager has behavioural patterns related to whether they generate alpha or not. If you don’t look at these, you are at a disadvantage. There are patterns and it is possible to identify them.”

WealthBriefing is pleased to share these insights and invite responses. The usual editorial disclaimers apply, and readers are welcome to jump into debate. Email tom.burroughes@wealthbriefing.com and jackie.bennion@clearviewpublishing.com

Managers who generate enough alpha to beat the returns of low-fee index funds by more than their own fees demonstrate that active fund management can - and does, in these circumstances - represent a better value than passive index investing.

But like the holy grail, alpha is elusive. It is difficult to generate in the first place, and even more challenging to reliably replicate. In fact, while it’s easy to measure a portfolio’s overall alpha, no one has systematically identified where in the manager’s decision-making process that alpha is actually originating.

The source of alpha - the single most important determinant of successful active portfolio management - has been shrouded in mystery.

Until now.

The Essentia research team has concluded a three-month examination of the origins of portfolio alpha - where in the myriad decisions made and actions taken by portfolio managers it tends to be generated or destroyed.

Our results were twofold. First, we found very few common sources of alpha across portfolios. With a handful of exceptions - which we’ll discuss below - most portfolios have their own unique fingerprint of what leads to the alpha they create (or lose).

Second, and most importantly: in 100 per cent of the portfolios we studied, we identified at least one factor - or 'categorizer', in our terminology - that significantly impacted alpha, for better or for worse. This has profound implications for all active portfolio managers. Our work shows that it is possible for managers to identify where in their decision-making process they tend to create or destroy alpha - and once that is known, the door is open for them to expand or enhance what adds alpha, and correct what diminishes it.

Our study analysed 60 portfolios over 14 years and identified distinct areas of manager decision-making directly tied to the alpha generated (or lost) within each portfolio. We tracked 24 categorizers - ranging from equity sector to holding period to decision day of the week - across six broad investment decision categories, or skills: stock picking, size adjusting, entry timing, exit timing, scaling in and scaling out.

Sources of alpha were found in varying degrees across six key investment skills among the portfolios we analysed. This chart shows the percentage of the portfolios in our study that contained at least one significant factor within a given skill.

 

In all 60 portfolios we analysed, we found at least one factor that was significantly associated with the generation of alpha. Very few of those were common across the portfolios, but we did find that for a majority of managers (63 per cent), alpha was associated with at least one factor within stock picking, most notably:

-- Conviction level: A significant number of managers demonstrated skill in picking stocks where they had strong conviction: namely, those in their top quintile of positions by maximum money invested. By contrast, a significant number of managers showed negative skill in the middle quintile of positions on this metric. This reinforces what we see in managers every day: positions that are neither proactively large nor small often lead to alpha destruction - it’s a common “alpha leak.” In our analysis, these lower-conviction stock picks tended to destroy alpha at an average rate of 2.1 per cent per investment.

-- Exit price momentum: For 37 per cent of managers in our sample, exit price momentum was strongly correlated with stock picking alpha. On average, we found that exiting stocks into negative momentum (a falling price, for a long position, or a rising price, for a short position) destroyed 6.2 per cent of alpha at the portfolio level, per investment. In contrast, stocks exited into positive momentum added 5.1 per cent of alpha at the portfolio level before the manager closed the position. So these managers had a marked tendency to finish a profitable episode on a high note, and an unprofitable one on a low note.

-- Open vs. closed positions: Portfolios often showed a stark difference between alpha generated by stock picking decisions in open positions vs closed ones: on average, we found that open positions were contributing 13 per cent of alpha at the portfolio level at the time of the analysis - while closed positions had generated -4.7 per cent of alpha. While more investigation is required to interpret this result, we find this particularly interesting in light of our work into the lifecycle of alpha: in some managers it may be evidence of the “round tripper” effect (in which managers hold their stocks too long and sell only after all alpha has been exhausted) we identified in that research. (Alternatively, in others, it could be a successful inversion of the disposition effect.)

-- Holding period: For a subset of portfolios (20 per cent), holding period was an important variable in alpha generation. On average, these fund managers pick stocks that rise by 7.5 per cent in their longest-held quintile of positions. This far outperforms other holding period quintiles, which all had negative average returns. Except, that is, for the very shortest-held 20 per cent of positions, where picks generated 1.7 per cent of alpha! Again, there are several possible interpretations of these results. That middle 60 per cent of positions, for instance, may simply represent the positions that didn’t work out; an acceptable cost of business. Our experience, however, tells us that this middle ground is a source of lost alpha, and that more often than not, managers had the opportunity to take remedial action sooner.

Our hypothesis going into this project was that we were going to find a short list of “alpha leaks” that active managers in general tend to suffer. But that wasn’t the case. We found, instead, a wide array of factors that affect alpha both positively and negatively, and very little consistency across the portfolios.

But we were able to identify sources of alpha generation and/or destruction in all portfolios - within a set of factors that are consistent, identifiable and measurable in every case. We think these are incredibly encouraging findings for active managers; once identified in their own portfolio, these can be corrected and optimised, with potentially significant benefits to their returns.


About the author
Chris Woodcock leads the research and product teams for Essentia Analytics. Prior to joining Essentia, Chris was a technology analyst at GAM Investment Management and a hedge fund analyst at GAM Multi Manager in London. Before his career in financial services, Chris was a professional footballer with Newcastle United.

Latest Comment and Analysis

Latest News

****************************INSERT INTO `wealthbriefing_cms`.`cms_tx_2020` ( `tx_date`, `tx_user`, `tx_pub`, `tx_page`, `tx_article_id`,`ip`,`type`) VALUES ( NOW(), '', 'WB', '/html/article.php', '187788', '34.200.218.187', '')************|||||||||||||||