The Astros and Their "Expected" Wins
The Astros, even if they don't win another game this season, will have at least produced the same results as last season. Currently, the 'Stros are sitting at 73 wins.
However, a good number of people would be quick to point out that the Astros are significantly outperforming their Pythagorean expected win percentage. And they are. Based on a formula that considers only how many runs the team has scored (624) and how many runs the team has given up (653), the Astros "should" only be sitting at around the 66-67 win area (depending on what exponent you use in your formula). That Pythagorean win percentage still puts the Astros at a 77-win season, a 5-game improvement over last year (and if my memory serves me, roughly what the experts predicted for the team before the season began).
But what has been the reason for the Astros' overachieving? Some would just call it luck. And to be honest, that's probably an element in their success. But there may be some tangible things we can point to that would help explain this.
The Pythagorean formula looks only at runs scored and runs allowed, so the way those runs are distributed across the games can yield different results. Specifically, if you win a lot of close games and get blown out in a lot of games, you'll drastically outperform the formula. If you lose a lot of close games and only win by large margins, you'll drastically underperform the formula.
On the season, when Houston has won, they've done so by an average of 2.96 runs. When they lose, they've done so, on average, by 3.73 runs.
If we call a "blowout" winning by 6 runs or more, the Astros have won 6 blowouts by margins of 7, 6, 6, 6, 9, and 8. But they have been blown out 14 times by margins of 8, 6, 8, 9, 13, 10, 9, 6, 6, 7, 10, 6, 6, and 8. So you can see part of where the weird run distribution comes from...we lost a lot more blowouts than we won, and when we got blown out, it was by a lot. So it's probably safe to say that when our pitchers have been bad, they've been bad. But our hitters haven't done too much shaming of the opposing teams, either, so the blame can be spread around a bit.
Another factor that some people think cause teams to significantly outperform their expected wins include performing well in high-leverage situations (hitting with runners in scoring position, bullpens that don't give up many late-inning runs), but looking at those numbers is an exercise I'll leave to somebody with better spreadsheet manipulation skills than me.
0 recs |
18
comments
Comments
Here's a formula for you.
The team that scores the most runs in one game wins. So based on the number of runs we score in any particular game we win, our win percentage for that game is expected to be 100%.
by entropic soul on Sep 3, 2008 1:54 PM CDT 0 recs
Yea
You can get all of these fancy win expectancy and what the stats tell you the team should do… but 99% of the times those stats aren’t correct, and 100% of the times the actual results are ;)
Go 'Stros!
by Stros Bro on
Sep 3, 2008 6:45 PM CDT
up
0 recs
They might not always be correct, but there also usually pretty spot on
But they force you to think, well this is out of whack: why is that? If there isn’t a fool proof reason why a team is over performing itself by 5 or more wins, you have to think to yourself, “how good of team will this be next year.” I get that you guys have an aversion to numerically based analysis, but is it necessary to dismiss just because you don’t like what it’s telling you?
A difference of one or two games isn’t anything to blink at, but the fact of the matter is, that the Astros have a tepid offense and tepid pitching staff, which questions the validity of our performance to date. The value of the Pythag formulas is that they hold predictive power. Right now they’re predicting that the Astros better hurry up and figure out how to score more runs or prevent mores runs. In an ideal world, it would be both. The Astros have been a train wreck of a team this year: the offense has been hot or cold, the same with starting pitching. Thus, their ability to win games consistently is in jeopardy unless one side of the equation improves.
A good friend of mine used to say, "This is a very simple game. You throw the ball, you catch the ball, you hit the ball. Sometimes you win, sometimes you lose, sometimes it rains." Think about that for a while.
by DyingQuail on
Sep 3, 2008 6:53 PM CDT
up
0 recs
Ummm..
I get that you guys have an aversion to numerically based analysis, but is it necessary to dismiss just because you don’t like what it’s telling you
I have nothing against numerically based analysis… I just think stats that tell you what a team should just don’t hold much water.. Give me stats that tell me how good a player is… how impressive their season has been… something that actually means something.
Don’t get so defensive when someone doesn’t agree with your rationale.
Go 'Stros!
by Stros Bro on
Sep 3, 2008 8:18 PM CDT
up
0 recs
I have no problem with you don't agree with my rationale
I hadn’t even presented my rationale, I was defending the rationale of someone else. You said:
but 99% of the times those stats aren’t correct
Which isn’t true. I don’t think I’ve read a single statistically based post on this blog, that wasn’t deemed irrelevant — like Wins don’t do a good job of evaluating a pitcher’s ability. I wasn’t saying just you, it was using a generality in defense of a reader of this blog who did some good work and raised a timely point.
AstroAndy was able to pause and think “this team is playing great right now, is it legitimate?” It turns out that it might not be. Neither you nor entropicsoul even said, good points or something to consider. You just trashed the guy’s idea by completely dismissing his method of analysis. I don’t get that rationale at all, especially since neither of you bothered to offer any kind of explanation as to why we shouldn’t consider Pythag records as an indication of a teams sustainability of performance. It’s fine to hold that opinion, but if you’re going to trash the idea, why not offer something for him, or someone so inclined, to understand your complete dismissal.
A good friend of mine used to say, "This is a very simple game. You throw the ball, you catch the ball, you hit the ball. Sometimes you win, sometimes you lose, sometimes it rains." Think about that for a while.
by DyingQuail on
Sep 3, 2008 9:52 PM CDT
up
0 recs
Well
I don’t think I’ve read a single statistically based post on this blog, that wasn’t deemed irrelevant
You haven’t been reading here very long
Go 'Stros!
by Stros Bro on
Sep 3, 2008 10:11 PM CDT
up
0 recs
I trashed nothing
For what it’s worth.. I AM the one that moved this to the front page so it would get more views… I posted my opinion about the stat…. it has nothing to do with Andy.. I thought it was well written and gave his point well.. hence why I promoted it.
Go 'Stros!
by Stros Bro on
Sep 3, 2008 10:13 PM CDT
up
0 recs
Thanks for bumping me up, StrosBro… I was happy to see something I wrote on the main page of my favorite Astros site!
by AstroAndy on
Sep 4, 2008 9:41 PM CDT
up
0 recs
No problem
this site is nothing without the readers…
Go 'Stros!
by Stros Bro on
Sep 7, 2008 10:03 PM CDT
up
0 recs
I know what you mean
The whole reason I wrote the fanpost was because every week when Baseball Prospectus and ESPN put out their rankings of the teams, they’ve always had a half dozen teams with worse records ranked ahead of the Astros. I always just shrugged it off as people who used to shit on the NL Central can’t do that anymore with the Cubs and the Brewers doing pretty well, so they just focus it on the Astros now. But then I found out that both BP and ESPN use some form of Pythag to do their rankings…which led to me wondering why things were so off.
But I’m totally behind the sentiment that there’s a reason the guys play the actual game instead of comparing stat sheets to decide a winner.
by AstroAndy on
Sep 4, 2008 9:40 PM CDT
up
0 recs
think back
to May. the Astros were riding high. Here is our previous high point of the year. http://www.baseball-reference.com/games/standings.cgi?date=2008-05-27. Bear with me here. We’re gonna take a look at our runs scored vs. runs allowed. Looks like the Astros were at +10. Respectable, but not enough to merit a record that good. This was the Astros playing their absolute best baseball of the year, record wise. They were simply maxed out. The Cards were +23, The Phils were +51. Those are sustainable run differentials, as both of those teams have been competitive all year long. Scroll down the season schedule and remember how low we sunk. The numbers caught up to us. Sort of. We were still outplaying where we should be. We still are right now. Sure, your statement that “the team that scores the most runs in on game wins” is obviously true. It’s one of those comments that John Madden would before the coin toss on a Sunday Night Football game. What it doesn’t tell you is where a team’s true needs are. If the Astros finish the season with a 84-78 record, on its face, you may think that we’re only five games away from making a playoff run. Sign a big name free agent pitcher, and we’re in. But- if we’re still -40 in run differential, it makes you wonder….go back through the annals of baseball history and see how many teams had negative runs scored differentials and finished with more than 85 wins. 2007 DBacks are the only team I could think of. So yea, maybe there are simpler ways to look at things, but few things in baseball are as simple as they appear.
by HighLeveragePerformer on
Sep 3, 2008 7:50 PM CDT
up
0 recs
We've had this topic before....and I would agree that the Pythagorean Formula
is a valuable statistical tool. It’s best value is as an analytical method of converting runs allowed/scored into wins/losses if you are modeling the effect of personnel changes for a team. As a predictive tool, it is much more crude, IMO. Sure, I agree with Dying Quail, above, that looking att differences between a team’s pythag prediction and the actual record should make you think, and one should be aware of it. But also keep in mind that the Pythag is based on certain assumptions regarding run distribution and that those assumptions are capable of being changed by managerial tactics, unusual personnel combinations, or unusual consistency patterns.. One of the most likely ways that team should be expected to do better than its Pythagorean record is if it has excellent late inning relief work, and I think that we can agree that Valverde has been dominant during this second half run by the Astros. This article in HT suggests that there are reasons why the D-Backs should have been expected to outperform their Pythag:
Interestingly, Valverde is a big part of the explanation posed for the D-Back’s overperformance. Also the manager’s strategy of “giving up” on blow outs, keeping terrible pitchers in the game to give up more and more runs is discussed. I think Cooper seems to follow this tactic too. The article points out that the D-Backs were 15-22 in blow-outs, which is unexpected for a 100+ win team. That seems consistent with AstroAndy’s tabulation of a 6-14 record for the Astros in blow outs.
Another characteristic which has been shown to be related to Pythag overperformance occurs when a team’s pitching is much more inconsistent than its offense. I think we can agree that the Astros’ pitching has been inconsistent this season…whether the offense has been more consistent, well I’m not sure, but I suspect so.
If you are interested in the topic of “explaining” Pythagorean discrepancies, try this article:
http://mvn.com/mlb-stats/2007/09/06/what-do-pythagorean-residuals-really-measure/
One run game winning percentage is highly predictive of Pythag overpoerformance, and the article points out the obvious point that every run scored above the winning run is a wasted run, and that teams which waste fewer runs will overperform.
I’m not really arguing one way or other, with respect to the Astros’ situation, because I don’t know for certain how much, if any, the Astros should be expected to out perform Pythagorean W/L numbers. But it’s an interesting topic. And I do see some characteristics associated with the Astros which seem to be among those which accompany overperformance.
by clack on Sep 3, 2008 8:28 PM CDT 0 recs
Inspired
Inspired by Astro Andy’s diary and his talk of high leverage situations, I sat down and (armed with trusty Excel sheet) I took a look at the majors, its win percentages, its expected win percentages, and its winning percentages in one-run games (high leverage games, natch).
The Astros, in outperforming their Expected win percentage by 50 points, are second in the majors in that metric, behind only the Angels, who are at .066.
But more interestingly, the ’Stros are the only team in the majors to have outperformed their expected record, while still having a losing record in one-run games. Houston is 19 – 20 in one-run games, for a .487 percentage, which is 18th best in the majors.
That’s a statistical disconnect that no other team in the majors can show. In all other cases, a team’s having outperformed their expected record predicts a winning record in one-run games, and a team having underperformed its pythagorean record indicates a losing record in one-run games. (The Cards, the Mets and The Gnats all have records equivalent to their pythagorean records, and all have losing records in one-run games).
I made a graph putting the expected – pythagoran number along the horizontal and the 1-run percentage along the vertical, and the astros are the only ones in the second quadrant. Everyone else is in the lower left or the upper right.
So what’s my point?
Ha. I don’t know. Maybe I’m just reinforcing the idea that the ‘Stros recent (and otherwise) hot streaks are sheer anomaly. After all, we all know the Astros just aren’t that good.
But I’m seeing that for many of the other teams who have done better than their expected record, like the Angels, you can point very quickly and very easily at their unrealistic record in one-run games as the reason.
With the Stros, you can’t. I read this book where Bill James seemed to sugges that managers should be credited for their teams consistenly finishing better than their expected records. But that would mean giving credit to Coop . . . and I’m not sure I’m ready to do that, either.
Demote Bourn . . . Before Someone Gets Hurt
by rastronomicals on Sep 3, 2008 8:51 PM CDT 0 recs
Always good to see you post here Rastro :)
I listened to an interview with Drayton, and he seems to think(at least publicly) that Cooper has done an outstanding job this season…
Go 'Stros!
by Stros Bro on
Sep 3, 2008 9:02 PM CDT
up
0 recs
Rastro, for what it's worth...
Hardball Times shows that Astros are 35-23 in “close games;” but I don’t know their definition of close games…maybe it could be 2 run margin. I don’t know if that adds anything to the 1 run game stat or not.
http://www.hardballtimes.com/main/teams/#nlstats
Another interesting think from the HT team stats above…the Astros’ bullpen is best at “Win Probability Added” in the NL.
by clack on
Sep 3, 2008 9:22 PM CDT
up
0 recs
I don't think anyone is ready to give the credit to Coop.
Blind luck seems so much easier to swallow.
A good friend of mine used to say, "This is a very simple game. You throw the ball, you catch the ball, you hit the ball. Sometimes you win, sometimes you lose, sometimes it rains." Think about that for a while.
by DyingQuail on
Sep 3, 2008 9:54 PM CDT
up
0 recs
I suggested that Cooper's strategy of letting starting pitchers take a beating...
might be a factor. (Think about keeping Backe in the game long enough to give up 11 runs…and almost that many runs a couple of other times….or R. Hernandez kept in the game at Pittsburgh to give up 6 or 7 runs in 3 innings…or allowing Chacon to give up 6 runs in 1 inning, etc.) The HT article about the D-Backs suggested that similar strategy by Melvin was genius. I think that depends on whether the strategy really saved the bullpen enough to be a positive influence on later games. Another way of looking at it…maybe this strategy did nothing more than inflate the size of blow outs. In other words, the “runs allowed” part of the Pythag is “too high” relative to a more “normal” pattern of pulling the bad pitcher. So, perhaps the effect of the manager is to make the “runs allowed” higher than it should be….it’s not that the manager caused “overperformance,” but rather caused an artificial number of extra runs to be allowed.
by clack on
Sep 3, 2008 10:33 PM CDT
up
0 recs
















