rolling sabermetrics and statistics thread

Pirates hopeful Cole Figueroa and his pursuit of analytics

http://triblive.com/sports/pirates/10149683-74/figueroa-ball-spring

http://www.baseballprospectus.com/article.php?articleid=28697

https://pbs.twimg.com/media/Cegi7j4XIAAIauV.jpg

The Indians are projected as the winningest AL team by PECOTA.

I read at least some of every BP Annual team chapter while i had an i.v. drip in yesterday. Some real good stuff despite two different references to Flo from Progressive... Russell Carleton (I think) wrote a breakup letter to WAR ("Remember when we stayed up all night making fun of Murray Chass?"), and the guy who wrote the Angels chapter calls the team, given its defiant-ostrich culture, The Comments.

on DRA

http://www.cbssports.com/mlb/news/mlb-analytics-guru-who-could-be-the-next-nate-silver-has-a-revolutionary-new-stat

https://fivethirtyeight.com/features/goose-egg-new-save-stat-relief-pitchers/?ex_cid=538twitter

seems cool!

I like that stat, more telling than a save. It raises a bit of the bar on John Franco and Lee Smith, both of which's career straddled between the old 'Fireman' and the 'Eckersley - 9th inning only' style of closers.

i'll have to read the article again closer to understand the applied stats a little better but when it comes to ranking pitchers based on value added, i don't see why you wouldn't just use WPA. as far as basic box score stats go tho, the goose egg definitely seems preferable to the save

an issue with using goose eggs to retrospectively evaluate relievers tho, as silver notes, is that reliever roles have revolved around getting saves for the past 30 or 40 years, so guys like rivera miss out on opportunities. also seems a bit unfair to LOOGY guys

Win Shares does this as well as LI (Leverage Index). But it may not be as general public friendly as goose egg.

I think the idea of it being only 2 run lead is what is interesting in the figure. A 2 run lead to me is a standard of the such, as it is a 'bloop and a blast' from the opposition being back in the game.

It's a good stat to do the comparison Nate set out to do, which is look at the classic 70s to mid 80s 'Fireman" compared to the LaRussa/Duncan closer works the 9th setup. I think this stat would perhaps be more an illustration of worth for some long time setup men too.

The type of relievers that are perhaps not as easy to compare would be old school long man types that would often come in early with the club to hold a team in a game. Big Red Machine's pitching staff was anchored by a couple of those types with Pedro Borbon and Clay Carroll. Those guys threw alot of innings held teams in check and let those big bats get them come back wins.

some of the differences here are rather small, but it does suggest the dodgers have found a metric they like for pitchers: average exit velocity (here filtered for a minimum of 150 batted ball events)

https://baseballsavant.mlb.com/statcast_leaderboard?year=2017&abs=150&player_type=pitcher

https://www.fangraphs.com/blogs/putting-war-in-context-a-response-to-bill-james/

dave cameron responds to bill james' nonsense

I'm at work and can only skim, but judging from the tone, he hardly seems to dismiss James's piece as nonsense. They see the issue differently.

yes, the issue is mostly one of what we value when we use statistics to evaluate players. that said, many of james' points were pretty ill-informed -- he even mentioned that he never really took the time to understand how WAR worked until now, which was pretty shocking to read considering...he's bill james!

I could be wrong about this--and if true, I don't think James would ever concede the point--but I do think he's always had a certain amount of resentment that WAR was adopted industry-wide and Win Shares never really went anywhere. (As opposed to the Pythagorean Formula or RC/27, among other things, that became foundational.) Did you read Posnanski's response to James? That, and a NoTime post on the awards thread, helped me understand his piece better.

i did read the posnanski piece, and he seemed to fall into the same trap james did. as cameron says, it more or less comes down to what question you want to answer. if you really want to reflect “what happened”, then just use WPA and be done with it (although as cameron points out, that solves one problem while not addressing another, if the objective really is to tie the statistics to wins). better yet, just use RBI, lol

it’s 2017. we accept at this point that performance in clutch situations is due largely to chance. WAR essentially says that if this player put up the same numbers over a million simulated seasons, he would add this number of wins to the team relative to a replacement player. the timing is irrelevant — but of course, that is a value judgment

The Yankees and Astros both scored almost 200 runs than they allowed. But the Astros won 101 games while Yanks won 91. When adding the individual contributions of the players do you want the Astros to be even with Yanks or way ahead?

— Tangotiger (@tangotiger) November 20, 2017

Can I answer "I have no desire to add the individual contributions of the players together?"

— Voros McCracken (@VorosM) November 20, 2017

I think the 1,000 simulated seasons would quite probably (though not with absolute certainty) prove that Altuve hit in luck in high-leverage situations this year, and that Judge had very little. But I still don't think that's particularly useful in determining 2017's MVP.

we’ve been over this, but to put it briefly: the “question i’m trying to answer”, to use dave cameron’s phrase, is — which player performed best in skills over which we can be reasonably certain they have signficant control? i think it’s fine to use as a tiebreaker for voting purposes, but in 10 years from now i’m going to look back at this season and conclude that judge and altuve had comparably productive years. the fact that it was an up year in leverage situations for altuve and a down year for judge won’t matter

One of the respondents to Posnanski's piece makes a good point, maybe the same one you're making (John Autin, who I think is one of the High Heat Stats guys...never look at that site anymore):

— New York’s 9-win shortfall from Pythagorean Wins is one of the 40 largest of all time, placing in the 98th percentile for absolute distance from expected wins.
— Judge’s “clutch” shortfall is also extreme. For instance, his high-leverage OPS was .188 less than his overall mark, which ranks near the 4th percentile in the last 5 years (of those with 100 hi-lev PAs in a season).

Basically, that James got lucky that this one very extreme case fits his argument.

Cameron talked about it on the EW podcast

https://www.fangraphs.com/blogs/effectively-wild-episode-1139-the-war-we-want/

Cameron isn't disagreeing with James, he's emphasizing what WAR is and what it isn't. He even says that he wouldn't rely solely on WAR when filling out an MVP ballot.

The MVP criteria are whatever you want them to be, so if you value players solely on their skills devoid of context that's fine. But applying some context to those skills is fine too, if you're looking to assign credit to who won the actual games. An RBI single in the bottom of the ninth makes you the hero, but a single with nobody on in the third doesn't. It's the same skill with different results, and it's OK to consider both if you're choosing the MVP of the game or for the entire season.

This is as good a place as any to revisit Judge's consistency vs Altuve's, which points to another problem with WAR -- it's strictly cumulative. Take this extreme example, who is more valuable over a four game stretch, a player who hits four home runs in the first game but does nothing during the next three games, or a player who homers in four straight games? WAR would say they're the same, but they're clearly not. In the first case, the player is doing nothing to help the team during three of the four games, in the second case he's contributing in every game. By the same notion, you'd rather have a player who performs consistently well over the entire season rather than a player who puts up three bad months and three great months, even if their season stats turn out to be identical.

i don’t agree the player who homers every game is more valuable...

https://www.baseballprospectus.com/news/article/35455/prospectus-feature-bill-james-vs-noise

very good article

very important assumption that James implicitly makes, but does not discuss: that the sole events worthy of consideration are the outcomes that actually occurred

Not sure where he implies this. And what type of outcomes are we talking about? It's not like he's arguing that you got a single but you get no credit because you didn't score.

According to Position A, the only thing that matters about Joey Votto’s walks is how the other Reds hitters capitalized on them.

Same thing. Don't think so.

if he only gets credit for them in games the reds win, then that's true

I think James was making a general point about context mattering and that stats that purport to show a player's overall worth might take context into consideration. To extrapolate from this that someone like Bill James doesn't understand that it's valuable to have a player on your team that gets hits even when his teammates don't happen to come through and bring him around to score is silly.

I completely agree, "position A" is a bad mischaracterization of James' views.

The three-HR game vs. three-games-with-a-HR question...I guess it comes down to a) the first guy greatly increases the likelihood you'll come away with at least one win in the series, vs. b) the other guy increases your chances in three games, but you still could get swept. I'd rather have the three games with a HR myself, but I understand the argument that they're of equal value.

Judge's slump...I wonder what his WAR was for those two terrible months. If a guy just missed two whole months, his chances of winning MVP would be close to nil (Trout this year might support that--best player in the league again, but voters thought he missed too much time). I have to believe Judge was at, maybe even below, replacement level for those two months, the walks and the HR aside--a .180 batting average makes for a whole lot of outs. If that's true (and I don't know that it is), he may have been actively reducing his team's chances of winning games. Which to me has to count as even worse than simply being out of the lineup.

Again, not two months' worth of games, but two actual calendar months.

One thing I remember from James in the 80s is the notion that a team's, let's say, 3rd run scored in any given game was more valuable than, say, their 10th run in any game that they happened to score that many, because a 10th run is generally less necessary for a win. Wouldn't that argue that the player who homers in four straight games is more valuable in that four-game stretch than the player who hits four in one game and then nothing?

I am bad at statistics and know it, so

Judge's slump...I wonder what his WAR was for those two terrible months.

from 7/14 to 9/2, his wRC+ was 82. obviously nowhere near as good as his first half of 197, but he was still putting up a 19% walk rate and playing solid defense. people act like he suddenly became a worthless player. i can't figure out how to split it up to get only the second half of july and all of august, but he put up +0.2 WAR in august, which extrapolated out over a whole season would come out to a little over 1 WAR. (it's 1.169, but you can dock him a little because the second half of july was his nadir.)

Thanks--I don't know how to figure that stuff out for myself. (I'm a whiz at RC/27...) The two months I isolated were slightly different: 7/13 - 9/12. Anyway, so he was a little above a replacement player. That helps his case. A little. I should also mention that I don't know how that measures up against MVPs historically. Maybe what I'm treating as this unprecedented slump for an MVP-candidate isn't in fact unprecedented--maybe other MVPs have had two-month stretches just as bad. I highly doubt it, but I can't say for sure.

(xpost) I think that's the basic argument, WC. In a way, Judge's season is a variation on that. It's a three-game series, though. He hit a home run in game 1, and overall went 3 for 8 in the first two games, with a couple of walks; in the third game, he went 1-6. Altuve went 2-5, 2-5, and 1-4, and he did other Altuve-like things in each game. (You can't do this precisely...that has him hitting .357, and you can't give him half-a-HR.) Who would you rather have?

clem, you can see WAR leaders over certain splits on the fangraphs leaderboard. they just do monthly/yearly/halves, so to go from specific dates i had to use the splits tool

just . . . everyone . . . stop with 'valuable'

if we're going to go this deep we should also factor in salary, and i *hope* no one wants to do that

make it the best fucking player over the course of a season award and let's go from there

I think there have always been such awards; The Sporting News' Player of the Year comes to mind.

We love to argue, though, so in 1931 some people got together and said "Let's create an award just ambiguous enough that we'll always have something to argue about." The arguments were so good, they did it all over again a few years later with the HOF.

One thing I remember from James in the 80s is the notion that a team's, let's say, 3rd run scored in any given game was more valuable than, say, their 10th run in any game that they happened to score that many, because a 10th run is generally less necessary for a win. Wouldn't that argue that the player who homers in four straight games is more valuable in that four-game stretch than the player who hits four in one game and then nothing?

Yes, that's exactly what I was getting at.

I guess it comes down to a) the first guy greatly increases the likelihood you'll come away with at least one win in the series, vs. b) the other guy increases your chances in three games, but you still could get swept.

Right, the counterargument would be that a 4 HR game basically guarantees you the win, whereas HRs in four straight games will score you some runs but won't guarantee a win. But to me that's kind of like claiming that a HR is equal to four singles, i.e. the HR guarantees you at least one run, whereas four singles gives you four chances to score runs but doesn't guarantee you'll score. And I probably don't have to explain why that's a fallacy (e.g. acc. to linear weights, a single is worth 0.4 runs on average, whereas a HR is worth 1.4 runs).

One thing I remember from James in the 80s is the notion that a team's, let's say, 3rd run scored in any given game was more valuable than, say, their 10th run in any game that they happened to score that many, because a 10th run is generally less necessary for a win. Wouldn't that argue that the player who homers in four straight games is more valuable in that four-game stretch than the player who hits four in one game and then nothing?

this is the sort of trivia that was probably interesting or even groundbreaking back in james’ day when no one had really given it serious thought before. but in 2017 it doesn’t really address any issues that are interesting to most sabermetricians

Except, it would seem, to the guy who invented sabermetrics.

game done changed

but in 2017 it doesn’t really address any issues that are interesting to most sabermetricians

I have no idea what this is supposed to mean. WPA, run expectancy, and pitcher leverage indices aren't interesting to most sabermetricians?

The basic point is still the same: context is relevant for evaluating past performance, but not for predicting future performance. The people arguing against that point are the ones jumping to silly conclusions, like in that BP article. Literally nobody is saying that Votto's walks are meaningless unless someone drives him in, that's a strawman argument. Bill James isn't the problem here.

context is relevant for evaluating past performance only if you wish to evaluate past performance in context :)

WPA and the like are fine for that, but those are a good deal more sophisticated than what i was commenting on. sorry, didn't mean to imply context-dependent stats don't have currency in the current sabermetric world -- of course they do. i personally don't care for them much, but that is just due to the questions i find interesting ("who are the best players?" rather than "who got luckiest this year?"). i agree with mookie in that i wish the award would just go to the player who played best that year (although not necessarily the "best player")

i will admit that i don't understand the granularity of win shares well enough (for some reason it seems to be impossible to find a good article on this...) to comment on it for certain, but my assumption is that because it is derived from total team wins, players on teams with better records might have an advantage. maybe that is incorrect

https://deadspin.com/major-league-baseballs-statcast-can-break-sabermetrics-1820987737

finally getting around to reading this

cameron to the padres

https://www.fangraphs.com/blogs/the-one-i-never-thought-i-would-write

goodness gracious

a team shd hire me

damn, this one hurts. not great for the saber community when all its best writers get scooped up by MLB teams and their work becomes proprietary

which is partly what that article I posted above is about

this isn't new info, but the visualization is pretty cool: https://twitter.com/search?l=&q=radar%20from%3AMattEddyBA&src=typd