New study finds 130/30 outperforms long-only in back tests

130/30 18 Sep 2007

According to a study in the forthcoming issue of the Journal of Alternative Investments, 130/30 funds would have outperformed equivalent long-only funds over the period since 1994 (a hat-tip goes out to the Advisor Perspectives newsletter for drawing our attention to this study today).

The author of the study is Lee Munder Capital Group’s Gordon Johnson.   While Johnson’s full paper has yet to be published, he released a white paper on the same topic earlier today.  In addition, a slide presentation on the topic is available here at the website of the esteemed quant organization QWAFAFEW (pronounced quaff-a-few, an allusion to the organization’s libationary founding principles).

Earlier today, Johnson told

“…our research shows that with a reasonable alpha model, 130/30 portfolios stand a very good chance of outperforming their long-only counterparts.  This holds true in a large-cap international universe just as in a domestic universe.  However, our results also show that there will likely be years where the shorts in a 130/30, measured in simple terms by themselves, may not add value.   In the long run, however, an investor in a 130/30 portfolio is likely to do better than an investor in a corresponding long-only portfolio.”

One of the great challenges of assessing 130/30 strategies has been the lack of any historical track records.  As a result, researchers have used theoretical rationale to support the idea (see our 130/30 research dossier for several examples).  While the results of these explorations make intuitive sense, they represent theoretical models running in a theoretical world.

Johnson builds on these results by putting theoretical models through the real world – specifically, the 1994-2006 U.S. and International equity markets.

First, he creates an “alpha model” using a number of common risk factors and fixed trading rules.  Then he runs two different versions of that model over the 13 year time period.  One is a traditional long-only version of the model and the other is a 130/30 version of the same model.

The results below show that the U.S. 130/30 fund beats the corresponding U.S. long-only fund and that the international 130/30 fund beats the corresponding international long-only fund.

Domestic (U.S.)…

International (non-U.S.)…

As you can see from the table below, moving the alpha model from long-only to 130/30 increases excess return and the information ratio.  According to Johnson, this corroborates the theoretical conclusions of papers such as this one by Clarke, de Silva and Sapra (free registration req’d.).

What’s the downside? Not much – assuming the original model (or manager) is able to produce positive alpha to begin with, of course.  Transaction costs, which Johnson says were not factored into the model, would not be large enough to seriously affect the general conclusion of this empirical test.

But how much of this out-performance (or under-performance) is the result of the long positions and how much is a result of the shorts?  Johnson says that no standards exist yet to make such delineation.  So he proposes one.

First, he addresses the issue of the appropriate benchmark for these hybrid funds:

“…a 130/30 strategy is more like a long-only strategy because it is managed to a benchmark and has 100% exposure to the market. Therefore, a 130/30 strategy’s performance should be evaluated similar to a long-only strategy and compared to its benchmark…the long and short portions of a 130/30 portfolio should be evaluated versus the benchmark like a long-only portfolio, not on an absolute basis as for a hedge fund. However, a slight modification is required to scale the performance of the long and short portions of the 130/30 strategy by their proportions of invested capital.”

His approach, which is laid out in detail in his paper, divides returns into contributions from long, shorts and what he refers to as the “long-short interaction” (basically, rebalancing).  He finds that over the ’94-’06 time period, longs contributed 9.1% of excess returns from the 130/30 version of the alpha model and the shorts added a further 2.1%.

But when he looks at each year in the sample, it becomes apparent that the excess return from longs and shorts in the 130/30 strategy deviates markedly from year to year.  The worst years for 130/30 excess returns from longs were ’98 and ’99 and the worst years for 130/30 excess returns from shorts were ’03 and ’04.

So while we’ll never have historical track records from real 130/30 funds, Johnson provides some empirical evidence for what 130/30 practitioners have long since argued: that in the presence of skill, 130/30 increases a fund’s information ratio and excess return.

Be Sociable, Share!


  1. ftse spread betting
    June 23, 2010 at 9:48 am

    Not only outperforms but significantly outperforms

Leave A Reply

← 130/30 a moving target according to originators of the craze An Empirical Analysis of 130/30 Strategies →