Next: Matt Maloney
From: Ron Johnson on 2 Jun 2007 00:25
On Jun 1, 5:04 pm, David Short <David.no.Sh...(a)Spam.Wright.Please.edu>
> 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.
That said, we're just starting to truly break down how much of this
is the pitcher and how much is the defense. The guys at the Hardball
Times are doing some truly fascinating work.
> 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.
Right, but they carry *exactly* the same weight. The correlation
coefficients are the same to about the 7th decimal place.
From: John Kasupski on 2 Jun 2007 02:49
On Fri, 01 Jun 2007 20:57:50 -0700, Ron Johnson
>On Jun 1, 6:24 pm, John Kasupski <kc2...(a)wzrd.com> wrote:
>> On Fri, 01 Jun 2007 17:04:17 -0400, David Short
>> 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)
One thing that's occurred to me in this area is the quality of the
opposition. Last time I sniffed around the concept, somebody told me
the scores of individual baseball games during a season meant nothing,
which surprised me coming from a guy who I know is a football fan,
because the NFL watchers pay very much attention to the quality of the
opposition when looking at a team's W/L record. But if I look a a
team's game logs, I'm going to be less impressed by a winning record
produced by a team that went 10-3 against the cellar-dwelling teams on
their schedule and 6-7 against teams that were legitimate contenders
for postseason play, as opposed to a team that has a winning record
against almost every opponent regardless.
>> 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
I guess what I'm looking for here is this: You've explained how there
is a correlation between runs scored and wins, but is there a similar
correlation between runs allowed and wins, and if not, how is this
explained? To me, if one team scores 800 runs and allows 850, while
another team scores 800 runs and allows only 750, I'd expect more wins
from the team that allows the fewer number of runs?
I guess I'm looking at the flip side of the same coin.
>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
>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.
....which is more likely to happen with your ace out there presumably
giving up fewer runs than your #5? Or not?
John D, Kasupski, Tonawanda, NY
Reds Fan Since The 1960's
From: John Kasupski on 2 Jun 2007 02:58
On Fri, 01 Jun 2007 21:25:42 -0700, Ron Johnson
>On Jun 1, 5:04 pm, David Short <David.no.Sh...(a)Spam.Wright.Please.edu>
>> 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.
>That said, we're just starting to truly break down how much of this
>is the pitcher and how much is the defense. The guys at the Hardball
>Times are doing some truly fascinating work.
>> 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.
>Right, but they carry *exactly* the same weight. The correlation
>coefficients are the same to about the 7th decimal place.
That's what I was looking for. It seemed to me like they *should*
carry the same weight since the number of runs you allow pretty much
dictates the number of runs you need to score in order to win.
John D, Kasupski, Tonawanda, NY
Reds Fan Since The 1960's
From: Bob Braun on 2 Jun 2007 10:37
"Ron Johnson" <johnson(a)ccrs.nrcan.gc.ca> wrote in message
> On Jun 1, 11:29 am, "Bob Braun" <oxspo...(a)hotandsunnymail.com> wrote:
>> "David Short" <David.no.Sh...(a)Spam.Wright.Please.edu> wrote in message
>> > coachros...(a)hotmail.com wrote:
>> >> On May 31, 6:02 pm, Ron Johnson <john...(a)ccrs.nrcan.gc.ca> wrote:
>> >>> Put simply, if Ks were important in modelling team runs
>> >>> scored, our models wouldn't work. There's no room for
>> >>> Ks to matter more than a couple of runs per team per
>> >>> year.
>> >> MODELING??? I thought teams were actually trying to score runs. K's
>> >> only cost a team a couple of runs a year, huh? Stay with your fanasty
>> >> league all you want; I'll watch real baseball where it is ALWAYS more
>> >> important to put the ball in play than not.
>> > This is one of the fundamental chasm's that sabremetrics cannot cross.
>> > There are people who do not believe in math. They do not understand it.
>> > They don't know what it does. When the math doesn't fit what they think
>> > they know, it MUST be the math is wrong.
>> > dfs
>> I understand the math. I understand the models. I understand the
>> I don't think K's are as costly as some may think. BUT........
>> when we are talking about ground balls, fly balls, moving runners and the
>> relative percentages, I still prefer a ball in play. How many times does
>> runner advance on a K?
>> The numbers are also skewed by the selfishness of modern day baseball
>> players. They simply don't adjust.
> People have been saying that since ... well I suspect it
> goes back to the day after baseball became a clearly
> distinct game.
> Al Spalding was talking about fans not being
> able to identify with players making vast sums of money
> in the 1880s (and how all that money was making players selfish)
> and the first "damned selfish players not playing the game the
> right way" quote that I'm aware of goes back to Dickie
> Pearce (Who was 35 when we first got organized leagues)
>> Situational hitting is a lost art, and
>> it's often times not discernable on a stat sheet.
> Well we've got PBP data going back to the mid 50s (and beyond.
> We have most of 1911 for instance)
So in 1974 I lined out to the second baseman, with a runner at second and
nobody out. The runner was caught off the bag for a DP.
Your data will show that I was attempting to move the runner?
From: RJA on 2 Jun 2007 11:28
"John Kasupski" <kc2hmz(a)wzrd.com> wrote in message
> On Fri, 01 Jun 2007 09:54:49 -0400, David Short
> <David.no.Short(a)Spam.Wright.Please.edu> wrote:
>>> On May 31, 6:02 pm, Ron Johnson <john...(a)ccrs.nrcan.gc.ca> wrote:
>>>> Put simply, if Ks were important in modelling team runs
>>>> scored, our models wouldn't work. There's no room for
>>>> Ks to matter more than a couple of runs per team per
>>> MODELING??? I thought teams were actually trying to score runs. K's
>>> only cost a team a couple of runs a year, huh? Stay with your fanasty
>>> league all you want; I'll watch real baseball where it is ALWAYS more
>>> important to put the ball in play than not.
>>This is one of the fundamental chasm's that sabremetrics cannot cross.
>>There are people who do not believe in math. They do not understand it.
>>They don't know what it does. When the math doesn't fit what they think
>>they know, it MUST be the math is wrong.
> Would you pay $250 for a seat in the Diamond section at GABP to watch
> guys in three-piece suits sit down in front of tables with computers
> on them and run numbers through Excel to determine which team wins the
> championship every year?
> That's basically what fantasy baseball is about.
> In the real world, that's not what baseball is about at all.
> One of the things by which people are going to test sabermetrics is
> whether or not it agrees with reality. Which is a good test. In fact,
> Grabiner even says so in his manifesto.
> The reality in real-world baseball is that on each and every day when
> a game is played, each team has an opportunity to win a game that day.
> There's no guarantee that he team that math determines to be the
> better team is going to win any particular game. And at the end of the
> season, the teams that have won the most games in each division make
> the playoffs, along with the non-division winning team with the most
> wins. That's the reality. Like it or not, ultimately the only thing
> that really counts is the number under W in the standings.
> This is why I'm with coachrose13 on this one. A lot of what Saber
> works with sounds great for the fantasy baseball leagues where the
> players are just numbers on a sheet and the winners and losers are
> determined based on the math, but...well, who was it that said that
> every journey of a thousand miles begins with a single step? A team's
> journey to the playoffs begins with tonight's game. Math is simply not
> up to the task of determining who is going to win tonight and who is
> going to lose. That's going to be determined by factors that no
> equation can hope to quantify.
> I don't know offhand what the Reds' record was in one-run ballgames
> during the 2000 season. But in any one of those games, if a guy comes
> up with a runner on third and less than two outs and instead of
> striking out, he grounds out while the runner scores and ties the
> game, then what?
> The math that tells us the difference is only a couple of runs a year
> can be correct, but the contention that it therefore makes no
> difference whether the guy strikes out or puts the ball in play does
> not take into account the timing of WHEN that handful of runs is
> scored or not scored.
Bingo. It's simply common sense. A ball in play at the right time instead
of a K can make a world of difference, which is why the ball in play is a
better outcome in many cases. Unless of course, the argument is made that
the same ball in play at the wrong time causes the opposite and keeps runs
off the board as with a DP but I'm not sure how you could prove the latter
because the runs that score after the DP could be the result of newly
> That's where the math fails the test of whether or not the math agrees
> with reality. Perhaps the Reds win that game in extra innings. As a
> result of that one statistically insignificant play, the Reds get one
> more W and instead of finishing in a tie with the Mets, they win a
> postseason berth outright. And as we all know, once a team gets into
> the postseason the sample size for a 5-game or 7-game series is so
> small, neither sabermetrics nor anything else has any hope of
> accurately predicting the results.
> I remember a guy who hit one HR all year long who hit two in the same
> game in a WS. Based on stats the probability was that the guy wasn't
> going to homer at all during that series. But he did, however much to
> the consternation of the mathemeticians who must explain it away as
> "small sample size" because no equation can quantify all of the
> factors that will determine the outcome of tonight's game. Which is a
> Good Thing. Otherwise there would be no point in playing the games.
> John D, Kasupski, Tonawanda, NY
> Reds Fan Since The 1960's