Basketball Court

As major market indexes waver in and out of bear market territory— shedding as much as 20 percent in a few short weeks — it is natural to fear that a financial meltdown is on the horizon. During a market downturn, some investors will ignore the noise and maintain a long-term outlook, while others look to point fingers. They blame quants for manipulating the market or take shots at passive investment strategies that take advantage of broad-based ETFs.

The truth is, there is plenty of blame to go around for the trillions of dollars in losses. Trying to boil down a complex system like the stock market into one or two catalysts is a futile task. Sometimes downturns happen because what goes up, must come down. Investors call this mean reversion or the tendency for things to even out over time. It appears not just in investing either, but most everyday situations like shopping and sports.

Let’s unpack an example from the NBA before diving into the stock market.

First a Definition

The idea of mean reversion first cropped up in the late 19th century, when Sir Francis Galton penned the Regression towards mediocrity in hereditary stature. He observed that extreme traits were not always passed down to each generation. What happens is certain characteristics like height drift higher or lower in the children. If two parents stand over 6 feet tall, the children, on average, will be a few inches shorter, and vice versa. Many fields later adopted the concept to describe different situations like asset price fluctuations and why professional athletes hit a rookie wall.

But finding mean reversion in the wild requires distinguishing between skill and luck. Extreme events will often precede a more moderate event when luck plays a large role in the outcome. Skill, however, can help create sustainable long-term success without as many ups and downs. Other factors to consider include sample size, correlations, and probabilities. Anything can happen in the short run, but over time, we should expect things to stabilize given enough observations.

Pass me the Rock

The NBA provides a good example of how mean reversion works. When a streaky player sinks multiple shots from deep, fans assume the shooter will sink more shots with great success or the next few will clank out. In the bestseller, “Thinking, Fast and Slow,” Daniel Kahneman refuted the former idea, “The hot hand is a massive and widespread cognitive illusion.” This has been the consensus opinion of statistical and mathematical journals for decades. Instead, shooting performance returns to a stable position represented by the mean, moving average, or another metric. Determining which mean to use is not an exact science, though, so we look at various metrics.

Linear Regression

Using player data from the 2019 season (above) demonstrates that average points from week to week mean reverts above and below 10 pts per game (ppg). In other words, players who average over 10 ppg one week tend to score less the following week, and vice versa. This seems like common sense, though. Bench players receive less frequent touches than a high usage first or second option. Most of their points come in garbage time or during favorable matchups. That’s why the second unit may score as little as one ppg one week and almost 10 the next. Conversely, team leaders—who control the pace and ball movement of a game—occasionally burst out for 50+ pt games and then fall back to more sustainable levels after fatigue sets in or defenses make adjustments.

Over/Under

Of course, there are other ways to observe mean reversion. Here we look at the number of players above and below their 7-game moving average, or about two weeks of games, each day. Something longer would not account for new changes in the data like Harden’s 7-game 40 pts streak while something shorter can be too fickle. The existing chart shows that about the same number of players score above and below their two-week average, meaning scoring constantly reverts over time.

Money Never Sleeps

Mean reversion in finance assumes stock prices or market factors move to an equilibrium level over time. This has happened since the 1800s, according to Wharton Professor Jeremy Siegel. The market experiences quick bursts of volatility that shake investor sentiment only for a wave of buying to restore somewhat normal conditions. That’s why some investors believe the recent choppiness will be short lived.

To understand the concept, let’s consider a few examples.

Linear Regression

Unlike the NBA example, stock returns do not mean revert from month to month (I looked up to 12 months). The trend between this month and next month’s returns hovers around zero , meaning there is no clear relationship. Perhaps annual data or something long would produce the desired outcome. That said, mean reversion signals often emerge in different length-moving averages. When price converges (or diverges) to the 100 day MA, it does so with a certain level of consistency (see below). 

Moving Averages

As noted above, other metrics can exhibit mean reversion. The rolling 30-day volatility of the SPY experiences stretches of turbulence that almost always return to a long term trend. 

Rolling Vol

Like any investment strategy, though, trying to time the market is just about impossible. Human nature takes over and causes us to act against our better judgement. Why? We have short attention spans and prioritize the present moment without considering long-term trends.

Interested in more content about finance and entertainment? Take me there