Pages

12/14/22

Michael Freire - Run Differential Revisited

 A few years ago, I wrote a series of articles on this very blog detailing the different statistical models that have become more "mainstream" in the modern baseball era.  I tend to be a statistical geek at times, so the subject matter and how teams are using this data to improve their performance is interesting, to say the least.

Now, before any "old school" fans get concerned, I am only going to focus on one specific statistical model in this article and I have no plans to revisit the aforementioned series. 

As the title of this article states, one of the models I most admire uses a team's run differential (runs scored minus runs surrendered) to produce an expected win total.  The fancy name for this formula is The Pythagorean Win Expectation and it is calculated by taking the number of runs a team scored (squared) and dividing that figure by the runs scored (squared) plus the runs allowed (squared).  To keep things reasonably simple, the resulting figure is the team's expected win percentage, which can then be multiplied by the number of total games played to generate an expected win/loss record.  Not so bad, right?

It does have a few drawbacks, namely it needs a fairly large sample size to be accurate and it is most effective when analyzing how a team performed after the fact (not as effective as a predictive model).

For example, from 2017 through 2021 (including the weird 2020 season), the Mets' run differential and resulting records were as follows;

2021:  -32 (77-85)

2020:  -22 (26-34)*

2019: +54 (86-76)

2018: -31 (77-85)

2017: -128 (70-92)

First observation is that there was some pretty bad baseball played during that window!  Second observation is that the only winning record in that time period was in 2019 when the team produced a positive run differential. In 2022, the Mets had a very successful regular season, as we all know (despite the sour ending).  The team produced a record of 101-61 and a run differential of +166, which is no coincidence and also very impressive.

How did some other teams do last year?

The Atlanta Braves had a run differential of +180 and a similar record to the Mets (101-61).  The LA Dodgers had an absurd run differential of +334 and a record of 111-51 (how they failed to win the World Series last year is still mind boggling, but it also shows the difference between regular season baseball and the quirks of the post season).

Sometimes teams can under or overachieve their run differential, but typically the statistics/formula are really close to the final results. 

The formula uses two basic components, which are runs scored (offense) and runs allowed (defense, if you will).  Making improvements to the overall differential can be the result of improving one or the other.  However, if you improve both, it has an even more profound effect.

The 2022 Mets improved their offense by a whopping 136 runs scored compared to the previous year, while reducing the number of runs allowed by a respectable 62 runs.  Combined, they went from a -32 differential in 2021 to a +136 differential in 2022.  It's no wonder why the team also improved by 24 games.

The Mets had the third best offense in the NL in 2022 (772 runs scored or 4.77 runs per game) and the second best "defense" in the NL (606 runs allowed or 3.74 runs per game).  As stated earlier, that was quite a jump compared to 2021, but can they do something similar in 2023?

There's a lot of hand wringing going on about the need for more offense in 2023, but I am not so sure.   We return largely the same cast from last year (thanks, Carlos), so is it crazy to think the team can't produce another 772 runs scored?  Maybe a bit better if we have better injury luck and a few of the kids turn out as good as advertised.  Remember, this team put up the listed runs with voids behind the plate, at third base (mostly) and at the DH spot.

From a run prevention perspective, I would argue that the rotation and bullpen are a step above where they were last year (especially if we get 50 to 60 combined starts from Scherzer and Verlander).  The 2022 Mets had similar potential, but could never seem to get everyone healthy at the same time (outside of the stretch run in September, but that's a different article for a different day).  Is it crazy to think the team cannot match the 606 runs allowed from last year?

I see incremental improvement on both sides of the equation, honestly.  Sure, the NL looks stronger this year and especially the teams in our division.  But, with health being the big caveat, I think the 2023 Mets could produce a similar run differential and a similar record.





No comments:

Post a Comment