"Regression to the mean" is a phrase that gets tossed around a lot in baseball in the post-Sabrmetrics world, but what does it mean?

I got to thinking about that this weekend, when a discussion about Chris Colabello came down the Twitter timeline. Colabello is enjoying a huge breakout season, going from a .229/.282/.380 line in 2014 with the Twins to a .318/.362/.494 line and counting for Toronto this season. That breakout screams "regression," but the question that is worth answering  -- and it doesn't just apply to Colabello -- is "what does he regress to?"

It hasn't been hard to find discussions about Colabello's likelihood to regress. There was this one from June, followed by this one from July, and yet the 31-year-old has just kept on hitting. In fairness, Colabello has regressed since that second article, but his .282/.318/.462 line since the All-Star break would still represent a massive improvement over his career to date.

So, what is he going to regress to, if or when it finally happens?

Luck-based regression

This is the easier of the two factors to account for, and it the one that we can talk about a bit more generally here. One of the great achievements of the sabermetric movement has been the more widespread acceptance of one fundamental fact: baseball players are more similar than we often think.

There are a nearly infinite number of possible outcomes every time a pitcher puts the ball in play towards a hitter, so it is kind of amazing how relatively narrow the likely range of outcomes is. This season, for example, 2.9 percent of all at-bats have ended in a home run; Mike Trout has had 8.3 percent of his at-bats end with a home run and Angel Pagan has yet to hit one. Those are the extremes.

Similarly, the league-wide BABIP last season was .299, with Starling Marte leading the way among qualified hitters with a .373 mark and Brian McCann bringing up the rear at .231. Again, that's a pretty massive gap within the context of the sport -- 14.2 extra hits per 100 balls in play -- but it really only represents a gap of 14 parts out of 100. From 1985 through 2014, 1,291 non-pitchers have at least 1,000 major-league plate appearances and 90 percent fall between a .261 and .332 BABIP for their careers. 

The range of possibilities isn't huge. (USATSI)
The range of possibilities isn't huge.

BABIP is influence by talent level, to be sure -- there's a reason Luis Valbuena ranks among the lowest career marks for this time period while Bryce Harper is near the top -- but there isn't a ton of year over year consistency. BABIP is one of those stats that can be influenced as much by luck or random chance as talent level, with the more extreme seasonal outliers usually the result of random fluctuations rather than any change in ability.

When it comes to Colabello, this is one place we can say he has obvious regression coming. He is sporting a .395 BABIP for the season, and that's fueling a lot of his apparent improvement. Based on historical precedent, we should expect him to fall between that .261-.332 range; there's a chance he's a true outlier, but outliers are by definition exceptional, so you shouldn't expect them.

Another place Colabello could have some regression coming his way is his HR/FB ratio. HR/FB comes about at the confluence of luck, randomness and talent, and for what it's worth, the park he hits in could play a part in boosting power. However, there has tended to be a pretty strong correlation between HR/FB and hard-hit average, and that is where what Colabello is doing looks unsustainable.

Colabello's HR/FB rate of 20.8 percent would rank 14th in the majors if he qualified, right ahead of hulking sluggers like Paul Goldschmidt and Albert Pujols. However, Goldschmidt consistently packs a wallop, with a hard-hit average of 41.3 percent, while Colabello has a 31.3 percent mark. That isn't bad, but it also isn't great by any means, and it does suggest that he won't be able to keep hitting homers at this rate.

From 2010 (when Baseball Info Solutions' current hard-hit average stat went into usage) through 2015, there have been 881 individual player seasons that qualified for a batting title; 180 of those seasons have featured a hard-hit average between 33.0 and 30.0 percent. The average HR/FB for that group was 12.21 percent, with only seven of the 180 posting a mark better than Colabello's. Again, maybe this is something he can sustain, but it isn't something we should expect. 

The point here is less about Colabello -- we'll get to him even more later -- but rather about what is realistic to expect from any hitter. Sure, some players can sustain a very high BABIP or HR/FB ratio -- Mike Trout and Giancarlo Stanton come to mind -- but most players aren't outliers. Most players fall within a relatively narrow range of potential outcomes, and when you have someone who falls outside of that range, natural regression should be expected, regardless of talent level. 

Talent-based regression

Now let's take a look at Colabello a bit more specifically, because his breakout is worth talking about.

With Edwin Encarnacion and Jose Bautista developing rather unexpectedly into star hitters late in their careers with the Blue Jays, just writing Colabello off entirely would seem like a mistake. Encarnadio was a fine player, with a .789 career OPS to his name, until he broke out as a huge power source in 2012. Bautista's breakout came even more out of nowhere, as he had just 59 career homers to his credit in 500 games before he broke out with 54 as a 29-year-old in 2010. 

Encarnacio and Bautista became significantly better power hitters in Toronto, but it wasn't just because they tapped into some newfound raw power. Both also became better functional power hitters, increase their rate of balls hit into the air. Encarnacion went from a 43.4 percent fly ball rate before joining the Blue Jays to a 46.1 percent mark since, while Bautista's fly ball rate jump from 43.0 to 47.1 percent since his 2010 breakout. 

That isn't the case for Colabello. He's traded some groundballs for fly balls and line drives, but still isn't anywhere near Bautista or Encarnacion's level, with a paltry 26.4 percent fly ball rate. In terms of batted ball locations, he's become less pull-heavy, but not significantly so, another sign that his apparent power breakout doesn't have a lot underlying it.

Worse still, his strikeout rate is barely improved, and his hard-hit average has actually fallen from 33.8 percent last season to 31.3. The massive jump in line-drive rate should be enough to convince you that Colabello really is an improved hitter -- 13.7 percent to 26.4 percent is a massive jump -- but it's hard to find much in his underlying numbers that hint at a true breakout.

So, Colabello is undoubtedly an improved hitter. He wouldn't be the first player to enjoy a late-career resurgence with the Blue Jays, and he probably won't be the last. Their organizational track record alone garners him some benefit of the doubt. However, if you're one of the players in 40 percent of CBSSports.com leagues who has picked up Colabello this season, just know that this won't last.

Where he ultimately regresses remains to be seen, but we might have just witnessed the best 70 games of Colabello's life.