Adam Dunn
in [Cincinnati Reds]
|
Next: Matt Maloney
From: John Kasupski on 1 Jun 2007 16:38 On Fri, 01 Jun 2007 16:22:45 -0400, David Short <David.no.Short(a)Spam.Wright.Please.edu> wrote: >John Kasupski wrote: >> On 31 May 2007 10:36:05 -0700, Ron Johnson <johnson(a)ccrs.nrcan.gc.ca> >> wrote: >> >>> Another way to look at this. We know that we can estimate >>> team runs scored to within about 20 runs most of the time >>> using the basic stats you can get from baseball-reference. >> >> What value is there in the ability to estimate that, considering that >> last season, the top six teams in the major leagues in terms of runs >> scored did not qualify for postseason play, one of the teams that did >> make the playoffs was 26th out of 30 teams in runs scored, and the >> team that won it all was 17th? >> >> Just wonderin'. > >Why...if we were better able to estimate how runs are scored, what >components go into producing runs generally, we might come to an >understanding of what offensive metrics were valuable and learn how to >construct a lineup that would score more runs. That idea occurred to me. I even went so far as to check if there was any correlation between the teams whose pitching staffs racked up the most Ks and the ones that made the playoffs, but at least for last season it didn't seem like it. >Is that what you were after John? Actually, no. Read my reply to Kevin. I was going off on a tangent, but a different one. John D, Kasupski, Tonawanda, NY Reds Fan Since The 1960's http://www.kc2hmz.net
From: David Short on 1 Jun 2007 17:04 John Kasupski wrote: > On Fri, 01 Jun 2007 15:46:32 -0400, Kevin McClave >> On Fri, 01 Jun 2007 15:38:29 -0400, John Kasupski <kc2hmz(a)wzrd.com> >>> What value is there in the ability to estimate that, considering that >>> last season, the top six teams in the major leagues in terms of runs >>> scored did not qualify for postseason play, one of the teams that did >>> make the playoffs was 26th out of 30 teams in runs scored, and the >>> team that won it all was 17th? >>> >>> Just wonderin'. >>> >> That's a strange question, John. If we don't care about runs scored, why >> should we wring our hands about productive outs and Ks and what not? > > The idea in the long run was to find out from Ron if they have the > same degree of success in predicting how many runs a team will *allow* > as they have in predicting how many runs a team will score (which Ron > said they can do within 20 runs or so most of the time). Given the same information ? sure, but it's far less useful than you would think. You give me the components of how many singles/doubles/walks/homers/tripples/batters faced/etc my pitchers will have and I can tell you how many runs my team will give up with the exact same kind of precision that I can for my hitters. There is a reasonably strong correlation from year to year in batting statistics. Much less so, for pitchers. Dramatically so. > That really has little to do with the productive outs thing, I'm just > wondering if there is more of a correlation between the teams that do > and don't make the playoffs based on how many runs are allowed by > those teams as opposed to how many runs they score. ........Are you being funny? What's important is the difference between those two numbers. You can back door winning percentage that way to get a guess at how many wins you'll get. Take Runs scored and square it. Divide that number by the square of number of runs allowed. plug it into a formula and it will give you a fairly accurate rendering of the teams winning percentage. For the most part, deviations from that will be small. I think there are folks who have fooled with the exponent so it comes out to something like 1 / (1 + (Runs Scored)**1.7 / (Runs Allowed)**1.7 ) > While I'm typing, I also need to make a correction, because I said > that the top six teams in runs scored failed to make the playoffs, and > that of course is incorrect since the Yankees, who led the major > leagues in runs scored during the regular season, made the playoffs. Glad to see I'm not the only one that gets things wrong now and then. dfs
From: John Kasupski on 1 Jun 2007 18:24 On Fri, 01 Jun 2007 17:04:17 -0400, David Short <David.no.Short(a)Spam.Wright.Please.edu> wrote: >John Kasupski wrote: >> That really has little to do with the productive outs thing, I'm just >> wondering if there is more of a correlation between the teams that do >> and don't make the playoffs based on how many runs are allowed by >> those teams as opposed to how many runs they score. > >.......Are you being funny? What's important is the difference between >those two numbers. Well, no, I wasn't trying to be funny...even though in a short series its not at all unheard of for a team to be outscored over the entire series but win the series nevertheless. Look at it this way: If you win a game, you have to have scored at least one run. Score only that one run and you win as long as the other team scores zero runs. If the other team scores a run, you now need to score two to win. It really makes no difference if you win the game by 1-0 or 13-2 except you pad your total runs scored for the season with those 11 additional runs (that you will eventually wish you'd saved for another night when you lose 2-1). You still only get one W in the standings. Similarly, it really makes no difference if you lose by 1-0 or 15-14, you get one L in the standings either way. So it seems to me that at some point, no matter how many runs you've scored, the number of runs you *allow* starts looking maybe even more important because that's what dictates how many runs you need to score in order to win. >You can back door winning percentage that way to get a guess at how many >wins you'll get. >Take Runs scored and square it. Divide that number by the square of >number of runs allowed. plug it into a formula and it will give you a >fairly accurate rendering of the teams winning percentage. > >For the most part, deviations from that will be small. I think there are >folks who have fooled with the exponent so it comes out to something like >1 / (1 + (Runs Scored)**1.7 / (Runs Allowed)**1.7 ) Now that's interesting. >> While I'm typing, I also need to make a correction, because I said >> that the top six teams in runs scored failed to make the playoffs, and >> that of course is incorrect since the Yankees, who led the major >> leagues in runs scored during the regular season, made the playoffs. > >Glad to see I'm not the only one that gets things wrong now and then. Heh...It's not like I really give a flip about the Yankees anyway, and they made such little noise against the Tigers in the ALDS, I hardly noticed they were even around last October. I realized as soon as I hit the send button that I shot from the hip and I was going to catch some flak for the ricochet.
From: David Short on 1 Jun 2007 20:57 "John Kasupski" <kc2hmz(a)wzrd.com> wrote in > On Fri, 01 Jun 2007 17:04:17 -0400, David Short >>You can back door winning percentage that way to get a guess at how many >>wins you'll get. >>Take Runs scored and square it. Divide that number by the square of >>number of runs allowed. plug it into a formula and it will give you a >>fairly accurate rendering of the teams winning percentage. >> >>For the most part, deviations from that will be small. I think there are >>folks who have fooled with the exponent so it comes out to something like >>1 / (1 + (Runs Scored)**1.7 / (Runs Allowed)**1.7 ) > > Now that's interesting. Hate to tell you that's something James was writing about 20 years ago...but it was. As you are aware, there are always errors associated with estimates. The sources of those errors are often called luck and that doesn't seem very satisfactory. Very, very smart people (including Ron and Dan) have spent a lot of time and computer cycles trying to nail down things that we can call the source of error that isn't luck. They've looked at things like strikeouts, and productive outs and managers and ...well, things just about ANYTHING that they could quantify, and the only thing that was able to improve the equation was adding in a fudge factor if a team had a great bullpen. If any of those things really mattered over the course of the season (and yes...They are darn well aware of what that limitation means) it would have factored out. If those effects exist, they are teeny-tiny itsy bitsy things. There are other sabre ways to go about things, you can add up win probabilities and such, but they get messy fast. Those methods are much better at telling us how valuable a player was (players get credit for "clutch" that way), but on the whole, those methods DON'T do a better job at telling us how a player will do in the future than the "spreadsheet" way. Wonder if the Cubs would trade Zambrano now? dfs
From: Ron Johnson on 1 Jun 2007 23:57
On Jun 1, 6:24 pm, John Kasupski <kc2...(a)wzrd.com> wrote: > On Fri, 01 Jun 2007 17:04:17 -0400, David Short > > <David.no.Sh...(a)Spam.Wright.Please.edu> wrote: > >John Kasupski wrote: > >> That really has little to do with the productive outs thing, I'm just > >> wondering if there is more of a correlation between the teams that do > >> and don't make the playoffs based on how many runs are allowed by > >> those teams as opposed to how many runs they score. > > >.......Are you being funny? What's important is the difference between > >those two numbers. > > Well, no, I wasn't trying to be funny...even though in a short series > its not at all unheard of for a team to be outscored over the entire > series but win the series nevertheless. > > Look at it this way: If you win a game, you have to have scored at > least one run. Score only that one run and you win as long as the > other team scores zero runs. If the other team scores a run, you now > need to score two to win. It really makes no difference if you win the > game by 1-0 or 13-2 except you pad your total runs scored for the > season with those 11 additional runs (that you will eventually wish > you'd saved for another night when you lose 2-1). You still only get > one W in the standings. Interestingly though teams one of the signs of a really good team is their record in blowouts. Dominant teams play better than .700 ball in blowouts and something like .520 ball in one run games (yes there are any number of great teams with excellent records in close games -- there are also great teams with poor records in close games. The defining characteristic of excellent teams is not their record in close games but the fact that they play fewer close -- relatively random -- games) > > Similarly, it really makes no difference if you lose by 1-0 or 15-14, > you get one L in the standings either way. So it seems to me that at > some point, no matter how many runs you've scored, the number of runs > you *allow* starts looking maybe even more important because that's > what dictates how many runs you need to score in order to win. That's all true. And every now and then a team will manage to do this with a fair degree of consistency. However, teams which win substantially more games than you'd expect given their runs scored and allowed are a *very* strong bet to decline. In other words, whatever magic they had working in a given year does not appear to carry forward to the next year. Same thing applies to runs scored. Sometimes a team will score a lot more runs than you'd expect given their basic stats. Again, these teams are a very strong bet to decline. > > >You can back door winning percentage that way to get a guess at how many > >wins you'll get. > >Take Runs scored and square it. Divide that number by the square of > >number of runs allowed. plug it into a formula and it will give you a > >fairly accurate rendering of the teams winning percentage. > > >For the most part, deviations from that will be small. I think there are > >folks who have fooled with the exponent so it comes out to something like > >1 / (1 + (Runs Scored)**1.7 / (Runs Allowed)**1.7 ) > > Now that's interesting. Standard error is just short of 4 wins. And we've found a couple of things that explain some of the error. Teams with an excellent bullpen tend to ot-perform their pythags (expected wins) by just under two wins. Teams with unbalanced bullpens (a few bad pitchers, a few good ones) can also slightly over-perform _provided they do a good job of identifying which group the pitchers fall in._ Logic is easy to understand, with relief pitcher you can decide who pitches by game situation. By contrast, starter pretty much pitch until they have to come out. Meaning that you can end up with your ace pitching with a big lead. |