User Comments, Suggestions, or Complaints | Privacy Policy | Terms of Service | Advertising
Page rendered in 0.7432 seconds
84 querie(s) executed
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand
— The Best News Links from the Baseball Newsstand
Tuesday, October 13, 2009
YESNetwork: Goldman: Five years, what a surprise
that TBS trotted out the Buster Elias Sports Bureau’s Productive Outs thingee?
PRODUCTIVE OUTS
I was disgusted to see a list of “productive outs” pop up towards the end of TBS’s broadcast last night. It just cemented TBS’s status as a network that broadcasts baseball but doesn’t pay enough attention to the game in any of its aspects to be successful. How the heck do we kill this concept that making outs can be a good thing? Check out the stats: A team with a runner on first with no outs has the expectation of scoring .88 runs, but a team with a runner on second and one out will score just .69 runs. Even though the runner moved over, the chances of scoring went down. Similarly, a runner on second with no outs meant that teams scored 1.14 runs on average, whereas hit a grounder to the right side and “productively” move that sucker over, and the run expectation drops to .97. Now, it is preferable to have the runner at second with one out (.69) then it is at first with one out, that is, having received a “non-productive” out (.53), so the productive out would be worth .16 of a run. That’s nice, but it’s such a small thing that it doesn’t really mean anything, doesn’t add up into anything you can see in the final record.
If the Yankees can be credited with having a high total of such outs, it is because they had a high total of runners on base. Scoring is the result of reaching base and making extra-base hits, not making outs. All this stuff about productive outs is purely imaginary corn for suckers, and TBS embracing it is just one more embarrassment for an amateurish production.
Repoz
Posted: October 13, 2009 at 07:02 AM | 25 comment(s)
Related News: General, Sabermetrics, Television
Related News: General, Sabermetrics, Television
My Bookmarks
You must be logged in to view your Bookmarks.
Hot Topics
Newsblog: Hardball Talk: Gleeman: Lenny Dykstra is back with some more can't miss investment advice
(131 - 10:51pm, Feb 09)
Last: Zuvella!
(131 - 10:51pm, Feb 09)
Last: Zuvella!
Transaction Oracle: 2010 ZiPS Projections - Toronto Blue Jays
(94 - 10:45pm, Feb 09)
Last: RollingWave
(94 - 10:45pm, Feb 09)
Last: RollingWave
Newsblog: Sam Hutcheson's Top 11 Sabrenerd Baseball Dork's* Basements
(20 - 10:43pm, Feb 09)
Last: baseball chick (now, with NEW blog)
(20 - 10:43pm, Feb 09)
Last: baseball chick (now, with NEW blog)
Newsblog: Cashman: No new pacts for big three
(14 - 10:01pm, Feb 09)
Last: Kyle C welcomes back our OBP Savior
(14 - 10:01pm, Feb 09)
Last: Kyle C welcomes back our OBP Savior
About Baseball Think Factory | Write for Us | Copyright © 1996-2008 Baseball Think Factory
User Comments, Suggestions, or Complaints | Privacy Policy | Terms of Service | Advertising
User Comments, Suggestions, or Complaints | Privacy Policy | Terms of Service | Advertising
Reader Comments and Retorts
Go to end of page
Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.
NO!. The chance of scoring 1 run did not go down (well it might have. I don't' have the data, but you can't make that conclusion from a run-expectancy matrix). The expectancy of total runs scored that inning went down. Big difference.
We must win and we must know how to win rather than win because we have statistical people. We have to win because we have baseball players that know and can understand the game.
It is, but only in certain situations. Showing a productive outs leaderboard, where the not-so-productive productive outs are lumped in with those that were helpful is ... unproductive. Invariably, there are bad teams on that leaderboard, so what does it tell you? Not that productive outs are particularly useful taken as a group. Also, no hitter should be *trying* to make productive outs. They should be trying to get hits; any truly "productive" out, aside from sacrifices, should be an accident.
I liken it to eating White Castle hamburgers, at least for me. There are certain situations/moments when a White Castle hamburger is a good thing. Ideally, I'd have a meal at a four-star restaurant, but White Castle does the trick in a pinch. Indiscriminate White Castle eating, however, is a recipe for intestinal revolt.
I'm pretty sure Ron Gardenhire woulda been doing the same thing, BTW. And why not. If you've got a runner on 1st and nobody out, and you've got Young, Punto and Tolbert coming up, what are your odds of scoring. Pretty much .00. If you've got a runner on second with 1 out and Young, Punto and Tolbert coming up, your odds are probably .01. I mean, any one of 'em could hit a single. You're just not going to get two base raps in a row. So maybe in that case you move him over.
I don't know any situation where a team would want to score exactly 1 run. What teams actually want is to "not score 0 runs". Bottom of the ninth, tie game, runner on 1st. The team should do what it takes to not score no runs that inning. If that means a home run to score 2, that's fine. The run expectancy of the chart does take than into account, whereas scoring exactly one run does not.
But in bottom of the 9th tied game isn't "scoring 1 run" just as effective as "not score 0 runs", and probably easier?
If teams went by just the run-expect matrix, they would never bunt (now this would probably improve a lot of team's run scoring). Bunting has it's situational uses, esp in a tied game bottom of 9th, runner on 1st, someone like Nick Punto up.
The run-expect matrix is a completely different animal then "we need to score right now, what are our chances in this situation" matrix.
*graphic appears on screen*
[20:30] <JH> NOOOOOO!!!!!
[20:30] <JH> NOOOOOOOO!!!!!!!
[20:30] <Alex> sigh
[20:31] <Alex> Productive outs?
[20:31] <Alex> seriously?
[20:31] <Joe> .ol
[20:31] <Tuf> Oh for ##### sake
[20:31] <Joe> is that list basically the teams with the most runners on base this year?
(snip)
[20:33] <JH> I'm sorry - I'm still flabbergasted by the Productive Outs thing.
[20:34] <JH> Flabbergasted. Flabbergasted.
(names altered)
When I saw the NOOOOO!!! I whipped my head to the TV screen expecting to see a chopper landing with armed gunmen or something. The reality was so much worse.
You can't measure run expectancy in the scenario you posed because the inning ends before measuring the number of runs scored can be completed. Run expectancy in innings with three outs can't be profitably compared to run expectancy in innings without three outs.
What this is trying to measure is a "second place scenario" thing.
In some elections worldwide, there are runoffs. So if three people are running, the third-place finisher is dropped and the other two battle it out. The reason this is fair is that it legitimizes a second-place vote. You have the ability to vote your conscience, but also if your guy isn't in the final running, you can say "at least I want this person" in the end. Otherwise you have Ross Perot and Ralph Nader possibly influencing election results for mainstream candidates.
Yes, every hitter wants a big hit to score some runs and to help the team win. But if they aren't able to do it, this stat (and it's imperfect for this application) tries to measure how often they do something useful with their time at bat.
There have been times over the past couple of years when Hideki Matsui is up in a big spot with a man on first. I want him to do something dramatic with the bat, but I really don't want him to ground weakly to second and have a double play turned against the team. At least he could strike out, or ideally hit the ball the other way to advance the runner, giving Cano a better chance to drive him in.
Is different than this: Run expectancy in innings with three outs can't be profitably compared to run expectancy in innings without three outs.
You *can* measure run expectancy, just with a more limited data set and a higher MOE. Take whatever set you were going to use, and only look at games where the team needed to score at least as many runs as the team you're calculating for. The only missing data will be those who needed to score less (whether they did or not) and the final run expectancy will be tamped down very slightly by the fact that some of the top end was cut off by the end of their games.
I don't think it's a bad stat because it is trying to measure something that need not be measured. I think it's a bad stat because it takes data and attempts to bend and twist it away from its purpose.
That would have a huge MOE.
You simply can't measure run expectancy in "little ball" situations by extrapolating from non-little ball situations. Closers aren't in the game in the 4th inning, managers don't pinch hit looking for one run in the 4th inning, managers don't pinch run looking for one run in the 4th inning. Run expectancy is unmeasurable in those sitations because innings get cut off before completion.
Maybe those things matter to "true run expectancy in little ball situations," maybe they don't. Empirical evidence for the proposition is and will always be lacking.(**)
(**) I still wonder why anyone cares. Miguel Cabrera loafed home in a lead run situation in Game 163. Almost every other Tiger would have scored. Does it really matter whether Miguel Cabrera loafing is or isn't the "real" or the "true" Miguel Cabrera?
Immaterial. Goldman's claim was "Even though the runner moved over, the chances of scoring went down."
That is not necessarily correct. In theory, the mean number of runs could decrease while the probability of scoring at least one run goes up. In some situations a team does want to maximize its chances of scoring at least one run.
Now chances are there are few if any such base-out situations where the mean scoring goes down while the probability of scoring at least one run goes up with an out/base advance but Goldman's statement is unsupported until somebody shows that.
The fallacy of productive outs is that they aren't particularly under the control of the batter or the manager (other than sacrifice attempts ... which, when they work at all, work mainly because enough of them go for hits or errors as to not cost an out). Players don't choose between productive outs and non-productive outs. They "choose" at some level between trying to make contact vs. not. LHB make more productive outs because, duh, they're more likely to ground out to 2nd, moving a runner on 2nd to 3rd.
The correlation this season between productive outs and baserunners was 0.13.
However, the correlation between productive outs and PO opportunities was 0.67.
Now chances are there are few if any such base-out situations where the mean scoring goes down while the probability of scoring at least one run goes up with an out/base advance but Goldman's statement is unsupported until somebody shows that.
I understand the points above about situations such as closer-in-the-game, bad-hitter-coming-up, and a hundred other things that can influence the decision. But the statements made in the article and in post #5 are referring to the run expectancy matrix. The problem is that you can't look at just the "score exactly one run" outcome. Because the matrix includes innings where more than one run scored. But one run had to score before the second (or third, etc.) run scored, so you still win those games. Therefore you have to count those.
The matrix for man on first, nobody out includes the time where the runner stole second and the next batter got a single, then three straight outs. That scores a run. It also includes the next two batters making outs and the next guy hits it out of the park, then the third out. That counts under the "scores two runs" part of the matrix, but you'd take that in the bottom of the ninth, tie game.
To address Walt's point about the mean going down while at least one goes up, I am under the impression that it does not, but that would be from my memory of other posts so I have no real proof.
Again, I'll re-iterate that the scoring matrix is merely input to a decision and doesn't take into account speed of the runner, which pitcher is on the mound, who's batting next, percentage of chance of a sacrifice succeeding, etc.
I wouldn't go so far as to say hitters have "no" control over productive outs - I would say only that they have far less control than the media pundits typically assume when they fawn over the stat. Hitters do, as a rule, try harder to make contact in a situation where a productive out would be valuable (part of the reason why power output tends to be lower in such situations even though in-play hit rates are virtually identical), but they normally can't (and aren't normally expected to) change their basic hitting approach just to put the ball in play. They aren't going to swing at a pitch that is significantly outside their normal hitting zone, where the odds of making an out go up dramatically - more often than not when they do so they make an unproductive out anyway.
-- MWE
I seem to remember a study that showed that fly balls to the OF were no more common in an AB with a runner on 3rd and less than two outs than at any other time. In other words, no evidence of skill and/or attempts to hit a sacrifice fly. Anyone else know?
Run expectancy matrix: http://www.tangotiger.net/RE9902.html
Runner on 1st, 0 out: run expectancy is 0.95, chance of scoring at least once is 44%.
Runner on 2nd, 1 out: run expectancy is 0.73, chance of scoring at least once is 41%.
This is from 1999-2002, when run scoring was higher. Right now instead of 44% to 41% it is probably close to a dead tie.
Runner on 2nd, 0 out: run expectancy is 1.19, chance of scoring at least once is 63%.
Runner on 3rd, 1 out: run expectancy is 0.98, chance of scoring at least once is 66%.
So in this situation you do get the phenomenon Walt is talking about.
I looked at the 2007 AL. All PA, excluding bunts, IBB, and defensive interference.
Overall frequencies (86123 PA):
K: 17%
BB+HBP: 9%
Contact: 74%
-- GB: 33%
-- LD: 14%
-- FB: 21%
-- PO: 6%
Runner on 3B, less than 2 out (5045 PA):
K: 15%
BB+HBP: 9%
Contact: 76%
-- GB: 34%
-- LD: 15%
-- FB: 21%
-- PO: 6%
So there is a little more contact, but not more fly balls. Also with differences this small, you have to start worrying about sample size, quality of pitcher and batter, etc.
Of course, just as the batter is trying to hit the ball in the air, the pitcher is trying to prevent that from happening. It is possible that the two effects cancel out.
You must be Registered and Logged In to post comments.
<< Back to main