clock menu more-arrow no yes mobile

Filed under:

With the first month over, Yankees team statistics are stabilizing

It has only been 22 games, but that is just when team statistics start to become reliable.

Brad Penner-USA TODAY Sports

April is a tough time to write about baseball. Just about everything comes with the small sample size disclaimer, and we as analysts are forced to either assess performance based on scouting individual games, or by ending every claim with the wishy-washy, "This player has looked good/bad, but there is still time".

I want to be able to draw conclusions. Baseball is a game of a lot of uncertainty, and we are now entering the portion of the season where we get a bit more certainty. I mean, this is essentially the equivalent of seeing a tiny dot of light at the end of the tunnel, but it is something. We can't make any sense of most individual player statistics (plate discipline statistics are more reliable, though), but team statistics? Yes, those are more reliable.

In a piece from Baseball Prospectus in April of last year, Rob Arthur wrote: "How Quickly Do Team Results Stabilize?". Through exhaustive analysis, he found that "Between games 20-25, where most teams presently lie, the combined model is accurate to within ~.25-.2 runs per game, for both RS [runs scored] and RA [runs allowed]". This combined model he proposed was through mixing runs scored and runs allowed through 20-25 games into the season, and rest-of-season PECOTA projections.

This model he created is not available to the public, but we can try to figure this out by shorthand. PECOTA thinks that the Yankees should score 4.29 runs per game, and they so far have scored 4.95 runs per game. No matter how you weight that, the sample so far is significant enough to pull them around to 4.50 runs per game in the future. And if we consider runs allowed, PECOTA has them allowing 4.35 runs per game, and they so far have allowed 3.82 runs per game. If we say that sits around 4.1 runs per game allowed (by a rough estimate) moving forward, then you're looking at a rest-of-season run differential of 56 runs.

This is probably a bit optimistic, and the data is still very noisy. One bad game screws all of this up. There is also research from Russell Carlton last May that run differential stabilizes at about 39 games, and they are about 17 games away from that mark. They currently have a +25 run differential, the fourth best in baseball. Once again, one bad game changes this, and that is why we are just shy of this team statistic becoming reliable.

This recent research, though, shows that the Yankees preseason projection of 80-82 wins undershoots their true talent by a few wins. There is of course the chance they do not benefit from good sequencing and squander their possible boon in run differential, but there is the chance that runs scored and runs allowed so far has pushed their true talent needle up a bit. If this means that they're really a true talent 85-win team, then they are in a better position to compete than the past couple of years, easily.