In an upcoming issue, we’ll be adding a proxy for mean-reversion potential to the lineup of risk metrics. Here’s a preview.
The rationale for monitoring mean reversion — specifically, an econometric-based guestimate that mean reversion is likely in the near term — is that selling and buying sometimes, perhaps often, go too far. Trees don’t grow to the sky and holes have bottoms. Deciding where the sky and holes end are forever open for debate, but in the cause of providing more context for assessing funds, benchmarks and strategies we’ll be introducing an estimate of mean reversion (MV).
The methodology is using the current 5-year return and putting it into historical context by way of percentiles. As an example, consider the S&P 500 Index. The chart below shows how a rolling 5-year return (with a 1990 start date) compares when transformed into percentiles. The maximum percentile reading (99%) in the chart below equates with a trailing 5-year return that’s the highest for the sample period. By contrast, the lowest return is at the 1st percentile.
The assumption here is that an asset that’s posted its highest return for a given trailing period faces a relatively high(er) probability for downside mean reversion risk. The opposite condition applies to an asset that’s performed at the bottom range of results.
By that standard, our MV score also doubles as a measure of expected return from a value risk-factor perspective. That is, an asset at the 1st percentile mark may be undervalued and therefore its expected return is relatively elevated (and vice versa at the 99th percentile).
The obvious caveat: high-flying assets can continue to outperform, sometimes for longer than expected. Meanwhile, poorly performing assets can remain dogs for an extended period.
Nonetheless, the ability to quickly compare how a set of assets stack up on a relative basis vis-à-vis MR is useful. All the more so when you can easily cross reference the MR scores against the other risk metrics that routinely show up in our data tables — see here, for instance.
Alas, there are (still) no silver bullets. But providing a broader scope for evaluating risk never hurts, and sometimes it’s unusually valuable. With that in mind, MR data will soon become part of our regular updates for our ETF opportunity set, strategy benchmarks and proprietary strategies.