Tuesday, July 28, 2009

Joel Greeblatt's New Back-Tested Results

A few days ago, I (and everyone else, I assume, whose e-mail address was on file at Greenblatt's Magic Formula website) received an e-mail announcing that Greenblatt's Formula Investing venture had released new back-tested results for the Magic Formula over the last ten years. Here are the results. The first number that jumped out at me was the positive double-digit return in 2007, which seemed surprising given the negative performance of my Magic Formula stocks that year. A glance at the first disclosure offered a partial explanation. From the Formula Investing site:

The results of the model portfolio performance:

  • Reflect the strategy of buying an equity portfolio of 24 top-ranked US listed equities that are within the 20% largest companies, as measured by market capitalization. As of June 30, 2009, these companies would have market capitalizations of approximately $890 million or greater.

Well, limiting one's portfolio to names of $890 million or greater would certainly have kept you from micro cap Magic Formula stocks such as these two1:

Why, one might ask, would someone have bought micro cap Magic Formula stocks in the first place? Greenblatt's book2 suggested two reasons:

1) The back-tested results for the all-cap version of the Magic Formula, which included micro-cap stocks, were higher than those of the larger cap version: 30.8% per year, on average, versus 22.9% for the larger cap version.

2) The worst one-year return for the all-cap version was -4% (in 2002) versus -25.3% for the larger cap version (also in 2002).

Recently though, as I noted in a previous post, Greenblatt raised the minimum market cap on the screener on his Magic Formula site from $1 million to $50 million. It's too bad he refused to explain why he did this during his recent Q&A with GuruFocus readers. My guess is that Greenblatt back-tested the numbers for all-cap portfolios since his book was published and found those numbers were uglier than he expected.

1Another simple modification to the Magic Formula would also have eliminated these two names: including a requirement that each company have positive earnings in each of its previous four quarters. Neither of these companies were consistently profitable; each had one windfall quarter in the last four, and in both cases the windfall was from a one-time legal settlement. I knew this at the time, but (perhaps I over-thought this) I assumed that the inclusion of these sorts of speculative stocks was the reason for the higher back-tested returns of the all-cap portfolios. Perhaps it was.

2See pages 56 and 61, respectively, of The Little Book that Beats the Market.


Paul Price said...

Almost all the firms and funds cheat as much as legally allowed.

Many funds compare their total returns to the changes in indices (without dividends or dividend reinvestment) to make their own relative performances look better.

When I worked at some of the biggest brokerage firms they would often take a stock off the recommended list when premarket news came out that would crush them on the opening. Then they would say (legally) they sold at the last trade price thus avoiding the huge gap down on the opening.

Their investors couldn't avoid the losses but the theoretical portfolios did.

Albert said...

Whenever I buy baskets of stock for my MFI account, one of the things I do is to look at their past four quarters to see whether the revenue and earnings had any anomalies. Especially things like legal settlements, etc. I can't imagine that any of those companies really contributed to the good returns of the backtested MFI results.

DaveinHackensack said...


That's true. In the case of Greenblatt's new firm, Formula Investing, it may be that there are liquidity issues which limit the minimum market cap on the stocks. But since Greenblatt hasn't given any explanation for why he raised the minimum market cap on the screener, I assume it's because the numbers on the microcap MF stocks recently reflect poorly on the system.

BTW, I saw your comment about the Value Line #1 Timeliness system on GuruFocus. I had the misfortune of being tasked with wholesaling separately managed accounts based on that strategy several years ago. I remember later reading an interview with the fellow who created the system -- a guy who had been with Value Line since the 1930s or '40s, if memory serves -- and that fellow saying that the system doesn't work well during transitional periods (e.g., I assume, when transitioning from a historic secular bull market to our current secular bear market). That would have been good to know at the time.


My guess was that most of those anomalous stocks would do poorly, but an occasional one would be a huge performer, and that drove the outperformance of the all-cap portfolios. As I said, I don't know if that actually was the case or not. It would be interesting to know.