Monday, October 19, 2015

How Good Were The 2015 Astros?

The Astros just ended a memorable run, a couple series before the ultimate prize. They snuck into the playoffs by winning the 2nd Wild Card by one game, finishing 86-76, the worst record for 2nd Wild Card since the new format was introduced. Almost no one saw this coming, and it would be easy to say they overachieved, and will need to get significantly better to maintain success.

Is this true though? Looking deeper at the underlying stats paints a much different picture, and bodes well for the future.

Just looking at Pythagorean Record, which takes basic runs scored and runs allowed to determine expected won loss records, the Astros should have finished 93-69. There is a lot of evidence to indicate Pythagorean Record is a better predictor of future performance than actual record. But there is an even deeper look, which paints an even better picture.

Fangraphs calls it BaseRuns. Baseball Prospectus calls it 2nd and 3rd Order Winning Percentage. I'm sure the inputs differ slightly, but they are both methods to strip away sequencing and clutch performance, which have almost no predictive power, to get to an expected won loss record based on underlying stats. Both methods agree the Astros were something close to a 97 win team.

In other words, the Astros hit and pitched like a 97 win team. So, how did they only win 86? They finished 29th of 30 in "Clutch" hitting and 28th in pitching in "Clutch".  I'm sure none of us can remember a time the Astros were "unclutch," can we?

But sequencing and clutch has been shown over and over to be, essentially, random. Good clutch teams one year become bad clutch teams the next, and vice versa.

In August of last year, the Cubs were under-performing their BaseRuns record by 8 games. Dave Cameron looked at the history of teams which under performed by a similar amount on JABO. He found
On average, these 11 teams underperformed their BaseRuns expectation by 53 points of winning percentage, the same mark as the Cubs are underperforming this year. Their overall winning percentage improved by 38 points in the next season, meaning that their record the next year came pretty close to their expected record in the prior season. Ten of the 11 teams improved their record in the next season. This is the kind of thing that is often referred to as regression toward the mean.   
Even if all the Astros do this offseason is bring back who they had, they should improve based on better distribution of events. But no one really expects them to stand pat. They have money to spend, prospects to call up, and maybe even some growth from the young players. But that BaseRuns numbers should give us a lot of optimism. Because they aren't really adding to a 86 win team. They are adding to a team that performed like a 98 team win in 2015, and could be on the verge of creating a juggernaut.


ntxlfty said...

This is very interesting. Perhaps it explains why we felt let down so often. Are there any examples that go the other way?

(Not Hank) Aaron said...

Of teams that outperformed baseruns? Well, the Rangers and Angels both were +7, and Twins +10. Essentially, sequencing created a interesting race down the stretch.

Chaz R said...

That's interesting, but doesn't that assume the same level of performance from the same players next season? You just never know from year to year who is definitely going to regress and who is going to take a step forward. Sure, we can make assessments, but we're not likely to be accurate. you've got hedge your bets by trying to improve known weaknesses.

Do we really want Jake playing everyday in the OF? Carter/Singleton at 1B? What happens with Valbuena? Is McHugh a legitimate #2 starter?

(Not Hank) Aaron said...

Sure, regression is possible, even likely, but no one on the roster really had an extreme best case scenario season, save perhaps for Keuchel. I think individual regression will likely work both ways, whereas there is almost no where else for the baseruns regression to go but up.

ntxlfty said...

How are these numbers affected by acquisitions? For instance, if the Rangers signed Price and Darvish returns to his old form?

Reuben said...

Does Base Runs factor in batter strikeouts? Obviously, you won't score as many runs as the average team does with a man on third and 1 out, if you strike out 10% (or whatever the number is) more than the average team.

(Not Hank) Aaron said...

Yes, they most definitely include strikeouts. The Cubs had a higher k-rate and were +3. Padres and Orioles were 3 and 4 in k rate and were +3 and +2, respectively. The A's had the second lowest k-rate and were -12. Don't see any correlation.

And all of the between season changes will obviously effect this. But, the point is, the expected base might not be 86 runs, like it seems.