THE BOOK--Playing The Percentages In Baseball

"I would guess the vast majority of his gain is in his situational hitting.”

Right, and that’s more skill than luck. Normal linear weights do underrate Ichiro. Contact hitters in general are better situational hitters (as Jeff points out), and I think Ichiro is one of the best situational hitters with his ability to seemingly place the ball. Chone Figgins and Ichiro are 2nd and 4th in Clutch scores over the last three years, respectively. The Great Clutch Project continues?

Rally has Ichiro as +40 for his career on GIDP.  Since wRAA doesn’t count outs explicitly, his out-making events in wRAA should be less costly than average because he makes fewer outs with those events by hitting into fewer double plays.  Rally’s valuation of that aspect is over half the difference between wRAA and RE24.  So the double plays thing likely is pretty big, and probably close to what Jeff speculates at the end.

One thing he brought up that I don’t think RE24 addresses is if Ichiro grounds into a fielder’s choice, he is replacing whoever was on base with himself, and thus presumably upgrading the baserunner and cutting a small amount into the cost of the out.  Since RE24 just looks at what bases are occupied and not the identity of the runners, it won’t care if it is Ichiro or someone else on first.  I would be very surprised if that is a significant difference, though.  If you had a baserunner who was +10 runs per 200 times on first, that would be +.05 per time on first.  If he trades himself for an average runner 20 times by getting on via fielder’s choice, that’s just 1 run.  +.05 per time on first seems on the high side even for Ichiro, compared to the range of baserunning values I’ve seen, but maybe it’s not when you only look at times on first base.

Here’s what I came up with a while back, looking at LWTS versus actual change in RE for Ichiro (this was during the ‘09 season, so no ‘09 data involved):

Ichiro’s career LWTS value: 121
Ichiro’s career delta-RE: 132

Kincaid - Ichiro has hit into 11 fielder’s choices in 9 years.  Not really a factor.

Jeremy and Tango - Almost all of Ichiro’s value, whether by linear weights or Run Value Added, comes from PAs with no one on base.  This is mostly because the vast majority of his PAs come with no one on base, but also partially because he can utilize his speed and style of hitting more effectively when no one is one base to be forced out at 2nd or 3d.

Colin: your RE is at odds with Fangraphs.  I’m going to have to figure it as well to see.  B-R.com has RE24 now, though I don’t think Sean compiled it seasonally, did he?

Peter,

Don’t these two statements contradict each other?

“Ichiro has hit into 11 fielder’s choices in 9 years.”

“but also partially because he can utilize his speed and style of hitting more effectively when no one is one base to be forced out at 2nd or 3d.”

If he isn’t hitting into fielder’s choices, then it doesn’t seem that his style is really more beneficial with noone on.  I’d think the opposite, because he hits for a high average with a low walk rate.  When no one is on, a single has the same value as a walk. 

That being said, I can agree that most of Ichiro’s value is when no one is on base, because, as you said, he bats more with no one on.  But, wouldn’t you expect almost all of the difference between RE24 and wRAA to be with runners on?  An out is an out and a single is a single with no one on, and RE24 should basically match wRAA in these situations.

Anyway, this makes this large gap look even larger, if all of the difference is in the PAs when runners are on.

Tom,

Why do you think his RE24 is so high (+223), when his WPA/LI is only 14.62, or about 146 runs?

Are you getting the 11 times in 9 years from Retrosheet?  I get 15 event types classified as FC from there, but 4 were no out was recorded, which can be lumped with ROE instead, so that leaves 11 where Ichiro makes an out but replaces a runner with himself.  That number is way too small, though.  Most FCs are marked as general outs (event type 2, rather than event type 19) in Retrosheet’s files.  Just going back through last season on plays where Ichiro grounded into a FC with a runner on first and Ichiro taking the runner’s place at first at the end of the play:

Sept. 30, bottom of the 8th
Sept. 27, top of the 9th
July 30, top of the 9th
July 21, top of the 6th
June 28, top of the 7th
June 10, top of the 5th
June 2, bottom of the 6th
June 1, bottom of the 3rd

That’s 8 going back through June of last year just on one specific type of FC, and none of those were marked as FC in Retrosheet (that query turned up nearly 150 more of that type of play from Ichiro’s career, but those 8 are the only ones I double-checked agaisnt B-R’s gamelogs).  So while I doubt that the effect is any more than 1 run per year, this type of event is way more common than 11 times in 9 years.

It would actually not surprise me much if there is something off with FanGraph’s RE24 numbers, because now that the examples in this thread have been brought up, I recall seeing weird discrepancies like that pretty frequently when I’ve looked at them for things.  I’m curious to see what more investigation turns up regarding the differences noted here.

Kincaid - My error.  Retrosheet does not use the fielder’s choice category for force outs.  I should have realized from the context of your comment that you were referring to a more inclusive use of the term.

I also am in disagreement with Fangraphs calculations of Run Value Added (RE24).  I wish there were some place on their site where they explained in detail how they were calculating all of the statistics that they use.

But, wouldn’t you expect almost all of the difference between RE24 and wRAA to be with runners on? An out is an out and a single is a single with no one on, and RE24 should basically match wRAA in these situations.

Steven - The linear weight value of an event is the average of run value added for every occurence of that event in every baseout situation.  Since that includes both events with runners on base and not on base, the run value added of an event with no one on will always be closer to zero than the linear weight of that event.  Since Ichiro is a much better batter than average, his total linear weights with no one on will be higher than his total run value added (RE24) with no one on.

http://www.fangraphs.com/blogs/index.php/glossary/

(The link is the top right corner of the main page.)

RE24 (runs above average by the 24 base/out states): RE24 is the difference in run expectancy (RE) between the start of the play and the end of the play. That difference is then credited/debited to the batter and the pitcher. Over the course of the season, each players’ RE24 for individual plays is added up to get his season total RE24.

The only thing I can think of is that the coding of the 24 base out states was done incorrectly.  For example, code “32” might refer to man on 3b, 2 outs, or it might refer to man on 1b/2b 2 outs.  I’ll ask David to double-check.

Tango - I was aware of the glossary, of course.  I don’t consider it sufficientfor my needs.  I want to know where the RE table comes from, whether it is empirically or Markov derived, what years it was calculated on, how they handle game ending hits for RE24, what events they are calculating RE24 on, and probably some other things that I can’t think of right now.  For wOBA I want to know what formula they are using and whether in includes SB or not.  For wRAA I want to know what they are doing because the glossary doesn’t give a clue.  It says they are calculating it from wOBA, so I guess that means they are not using the normal methodolgy of linear weights by matching batter events to their linear weight values.  I would also need to know where the linear weight values come from, what years are used, how they are calculated and what events are included and which are left out.  I know most of these things are not of interest to most of Fangraphs users, so I don’t expect to find them in a glossary, but it would be helpful if the information were available somewhere on their web site.  And perhaps it is, and I have just not been diligent enough to locate it.  If that is the case, any help would be appreciated.

Sean has Ichiro at +199 runs, including baserunning, which is kind of consistent with Fangraphs’ numbers.  Colin, is it possible the gap is with basestealing / baserunning?

So wRAA is missing baserunning and/or basestealing compared to RE24?

***

The RE charts at Fangraphs and B-R.com come from me.  I posted the RE chart in a Google Docs somewhere in this blog a few years ago.  I’ll see if I can find that link.

wRAA at FanGraphs has basestealing too, just not non-SB baserunning.  Calculating his wRAA without SB, I get 120 wRAA, so 33 runs less than his wRAA with SB/CS included.  That covers the majority of the value Rally has for baserunning.

RE24 doesn’t credit Ichiro with non-SB baserunning, though, does it?  If he takes an extra base on a batting event and not an an explicit baserunner event (i.e. SB/CS), then doesn’t that just get credited to the batter as part of the resulting base-out state that results from his batted ball?  For example, if Ichiro goes first to third on a single, doesn’t the batter get the credit for the difference between runner on first and runners on 1st and third, since that’s just the difference in RE from the start of his play to the end?

It’s possible - I know I didn’t do anything with baserunning when I ran my figures. I honestly don’t remember enough about my calculations to tell you what’s going on - that was nearly a year ago, and I can’t find the raw data anywhere.

There are four baserunning categories:

1. limited to only SB/CS
2. includes all non-PA events (WP, PB, BK, PK, DI)
3. on batted balls
4. on other PA-ending events (K, BB)

I brought this up on LL but was confused.  Why can’t we use WPA/LI and then convert this to runs?  Is that a silly thing to do?  Shouldn’t WPA/LI and RE24 be roughly the same?

They should be roughly the same, yes.

RE24 is better converted to wins, though it doesn’t matter.  Fangraphs already shows it as REW.

Remember, the best currency is wins.

DiPernian analysis (stating that Ichiro is just a corner OF with a 118 OPS+) has pointed out that LW sometimes overrates Ichiro’s offense, since he hits so many infield singles which have less runner advancement potential than an OF single.  But as this post shows, there are situations that underrate him as well.

I’ve looked at it from many angles, but don’t think the +191 runs I give him for batter events (hitting, ROE, avoiding DP) is too high or too low, I think it’s in the ballpark.

Ok that makes sense but for Ichiro his REW and WPA/LI are not close.

REW=22.99
WPA=20.89
WPA/LI=14.62
(LI = 1.00 so it doesn’t make sense to me why WPA and WPA/LI are different but that’s a separate point)

Also, shouldn’t REW and WPA be basically identical?

WPA/LI and WPA are different because he’s been clutch.  His average LI is 1, but he’s been better in the high leverage situations.  He has 6.3 clutch wins, which obviously will make up the difference, because it’s calculated as WPA-WPA/LI.

REW also measures clutch, to some extent.  It doesn’t measure “win clutch”, like batting in the bottom of the 9th, but it measures “run clutch” like batting with the bases loaded.  I’m sure there are better terms for these, but those were the best I could come up with.

So basically, Ichiro has been awesome in situations that have a high “run leverage”. 

Situation OBP SLG OPS
 Bases Empty 0.361 0.438 0.799
 Runners On 0.409 0.426 0.834
 RISP 0.443 0.421 0.864
 RISP w/2 Outs 0.481 0.455 0.936
 Bases Loaded 0.433 0.635 1.068

The higher the “run leverage” (again, I don’t know if there’s a better name for this) the better he hits. 

So the best explanation for all this that I can come up with is that Ichiro is very “run clutch”, but not “game clutch” above what is all ready measured there.

Ah, of course.  I knew my mind was getting bogged down in something that was kind of simple.  That makes total sense with the WPA and WPA/LI.

I never thought about REW measuring “run clutch” instead of “win clutch” but that makes a lot of sense.  That’s cool though.

So if Ichiro is very “run clutch” we have to ask the question of whether Ichiro hits better in high run leverage situations (focuses more or sees better pitches) or if his style of play produces better results in high run leverage situations than we would expect (hits into few DP).

I have to think it’s mostly his style of play.  He puts the ball in play a ton, without hitting into many double plays.  He also ends up hitting better in those situations because he can take advantage of fielders being out of position very well.  I don’t watch him enough to know if the hit and run gets called very often, but that would be another situation that is perfect for his skill set.