So can the Reds' hot streak last? Well, we can use a team's Pythagorean record as an indication of how closely their record matches their performance--which then can indicate how they can reasonably be expected to perform in the future. The Reds are currently rated as one game BETTER than their Pythagorean Record, whereas they were five games under Pythag when Narron left. That means they're roughly 6 games over Pythag under MacKanin...which, if we do the thought experiment, drops his managerial record with the Reds from 29-19 (0.604) down to 23-25 (0.480).
A 0.480 team is certainly a heck of a lot better than what we had before, and is closer to how most of us expected the Reds to play this season (I predicted a 0.500 year in the Hardball Times Preview). Unfortunately, this also indicates that it's probably rather unlikely that the Reds can continue to play 0.600 ball for the rest of the season, so we may be disappointed if we start to allow ourselves a bit of hope.
These data also probably indicate that while MacKanin has done a good job with the Reds, the turnaround in performance probably hasn't been as dramatic as the record indicates. So no, I'm not really on the "Hire Pete" bandwagon right now.
Now you've lost me - how again does being 29-19 really translate into 23-25?
ReplyDeleteBaseball is a game of streaks. Just as you should not marry a girl when your first get the hots for her, so too should the Reds not marry Pete just yet.
Hi. Here's the math:
ReplyDeleteReds with narron: -5 games vs. Pythagoras
Reds currently: +1 games vs. Pythagoras
Therefore:
Reds with MacKanin = 1 - (-5) = +6 over Pythagoras.
Therefore, the Reds Pythagorean record with MacKanin is roughly 6 games below their current record, or 23-25 (rather than 29-19).
It's probably better to get the actual run scored/allowed values with MacKanin, but I just wanted to do a quick estimate. I'm sure it's not off by more than a game one way or another.
-j
You're actually off by two games. The Reds' run differential under Mackanin is 281-275 R-RA, which comes out to a 25-23 Pythagorean record.
ReplyDeleteOkay, disregard that. I'm not sure where, but I screwed up the R/RA totals. The real totals are 254-264 R/RA. Which matches up with your estimated record exactly, 23-25.
ReplyDeleteHi Bradley -- Thanks for looking up the real numbers! -j
ReplyDeleteSo you're saying that six of Mackanin's wins are probalby luck, of the good variety. I certainly saw that yesterday with a grand slam from a second string catcher and quality start from a 30 year old rookie. That doesn't generally happen.
ReplyDeleteI think luck, in this case, generally refers to timing of when runs are scored and allowed. The Reds certainly get credit for Jorgensen's slam yesterday under Pythagoras. I think, without sitting down with the data, what's most likely been happening is that they've had an exceptionally good record in one-run ballgames (that certainly was the case in July...haven't looked at the data in August yet), and a propensity to give up runs when it doesn't matter so much. I'll look at this more when I do my August review this weekend. -j
ReplyDeleteYes Justin, that is usually the case when a team is outperforming their Pythagorean record in the short term. For any individual team, that may not entirely be attributed to luck, but it certainly is to some extent. For example, if you have a drastic gap between the top of your bullpen and the bottom, and you only use your best relievers in close games, you're probably going to run well in one-run games and also get blown out a lot. That's essentially what is happening with Arizona this year. Though their record is probably attributable to some deal of luck, their bullpen dynamics probably add a bit of skill that isn't captured by Pythagorean record projections.
ReplyDeleteI haven't looked at the Reds games very closely, but they have won a lot of one-run games and been blown out a few times since Mackanin took over. I'm more interested in the team EQA splits since Mackanin, but I can't seem to find those.
Pete should only have three ahead of him on a short list Only LaRussa, Girardi, and him.....that's it.
ReplyDeleteBradley,
ReplyDeleteI saw Sheehan's article on the D-backs. While it is true that bullpen usage and composition can affect a Pythagorean record, using those factors to explain the the fact that they are 12 wins over their Pythagorean record is a case of trying to attach logical meaning to a statistical oddity--a trap that we all fall into now and then. Sheehan himself acknowledged the importance of simple luck in his article.
Some hard numbers help. For example, David Gassko's study on Pythagoras from earlier this year found that the quality of the bullpen, especially the better pitchers within the bullpen, can affect a team's deviation from Pythagoras by +-2.7 wins (=2 x stdev). That's certainly not insignificant, but there's an awful lot of unexplained variation around the Pythagorean record that is (mostly) due to luck (+- ~7 wins).
As for EQA estimates...I tend not to like that stat all that much (my strong preference is wOBA, though OPS is a fine approximation most of the time), but I did note that July OBP and (especially) SLG were rather low in my July review. I also took a look at record in 1-run games and extra inning games in that piece, and both were unusually strong--hence the 0.500+ month.
I'll take a look at the August numbers this weekend. Fortunately, the team's performance does seem to have improved in a significant way this month. Not 0.600-ball improvement, but (so far) it's probably been the best month of the season. -j
Yeah Justin I agree. I wasn't trying to say all or even most of the DBacks' deviation was due to bullpen factors. I was just using them as an example of factors not incorporated into Pythagorean estimates that can cause deviation which is not directly attributable to luck.
ReplyDeleteAs for EQA vs wOBA/OPS, I don't think EQA is a very good metric for evaluating individual batters, but team EQA is easily, and pretty accurately, converted into expected run output. When adjusted for quality of opponents and a similar analysis for opponents AEQR it provides, IMO, a slightly better prediction than Pythagorean based on actual R/RA.
You might find this interesting if you haven't seen this already.
ReplyDeleteGreat thread on how one might create an alternative to PythagenPat. The argument is that one should either convert the runs up to wins, or convert them down to a bases-like stat (obp, woba, etc).
SABRMatt has an interesting approach (highlighted in that Tango thread) in which he converts each day's game via pythagoras, and then averages all the games on the season to predict wins. He gets a tighter fit than traditional Pythagoras, and Dave Studeman later found (mentioned in one of his columns) that SABRMatt's approach was slightly better at predicting second half records given first half records.
FWIW, PythagenMatt predicts that the Reds will get 74 wins on the '07 season, whereas PythagenPat puts the Reds at 73. So I automatically like PythagenMatt better. :) -j
Thanks for the links, I'll get to them tomorrow sometime. I'm ashamed to say I haven't actually read THE BOOK yet (it should be delivered tomorrow). :)
ReplyDeleteI'm still working through it myself -- reading time has been sparse the past few weeks. :} -j
ReplyDeleteAfter Tuesday, I'm on the Don't Hire Pete bandwagon. Well, I was actually already on it, but Tuesday was the low-fat, sugar-free icing on the cake.
ReplyDeleteHe's getting cautious. Too much so, IMO. He's playing not to lose.
ReplyDelete