SoxProspects News
|
|
|
|
Legal
Forum Ground Rules
The views expressed by the members of this Forum do not necessarily reflect the views of SoxProspects, LLC.
© 2003-2024 SoxProspects, LLC
|
|
|
|
|
Forum Home | Search | My Profile | Messages | Members | Help |
Welcome Guest. Please Login or Register.
Don't Say "Clutch": The Reality of Situational Hitting
ericmvan
Veteran
Supposed to be working on something more important
Posts: 8,925
Member is Online
|
Post by ericmvan on Nov 9, 2022 5:37:28 GMT -5
One thing that's happened to baseball in the last decade is that there is now a strong correlation between batting in high leverage and facing the best pitchers. Win Probability Added, when compared to straight-up, unadjusted WAR from hitting, is thus largely a measure of how well a hitter hits good pitching. This table lists career WAR per 600 PA, then the same figure ("Adj" for Adjusted) with WPA substituted for Batting Wins Above Average. The final number, HV, is "Hidden Value," expressed as a percentile. An HV of 25% means the true WAR, including situational hitting, is 25% more than what you'll see at FanFraphs. Note that a slightly negative HV is probably the likeliest outcome. That is, the average good hitter hits relatively better against bad pitching than good. Any number here might be subject to interpretation. A LHB who can't hit lefties, like Pederson, is going to get his HV killed if the manager has him doing that, because the good lefty pitchers will own him. And in fact he has a -9% career HV. After typing that I ran his 2022 numbers for the first time: 2.9 WAR / 600, 3.4 adjusted, 18% HV. Which is terrific. Note that the overall WAR, as his writeup in FG's top 50 FA list points out, was dinged by having to play too many innings in the OF. Name PA WAR600 Adj HV
Hosmer 6877 0.9 1.6 83%
Drury 2276 0.9 1.4 49%
Verdugo 1957 2.2 3.0 37%
Swanson 3387 2.9 3.9 36%
Andrus 8197 2.6 3.4 34%
Rizzo 6540 3.2 4.0 26%
Laureano 1640 3.7 4.6 26%
Frazier, A. 3045 2.3 2.8 23%
Peralta, D. 3908 2.5 3.0 20%
Hernandez, K. 2995 2.4 2.9 17%
Brantley 6092 2.8 3.3 15%
Segura 5611 2.7 3.1 14%
Wong 4043 2.9 3.3 14%
Margot 2512 2.1 2.4 13%
Carpenter 5370 3.7 4.1 10%
Peterson 2323 1.1 1.2 9%
Benintendi 3163 2.5 2.7 8%
Profar 3102 1.2 1.2 7%
Kiermeir 3351 4.2 4.4 5%
Devers 2958 3.7 3.8 3%
Turner, Jus 5146 4.0 4.2 3%
Turner, Trea 3737 5.1 5.2 2%
Haniger 2437 2.9 3.0 2%
Conforto 2980 3.9 3.8 -2%
Tatis Jr. 1175 6.9 6.8 -3%
Longoia 7969 4.1 4.0 -3%
Bogaerts 5389 3.8 3.7 -4%
Story 3532 4.0 3.8 -5%
Kepler 3361 2.9 2.7 -5%
Vazquez, C. 2633 2.8 2.6 -6%
Yastrzemski 1726 3.2 2.9 -8%
Judge 3161 6.9 6.2 -10%
Pederson 3431 2.3 2.1 -10%
Nimmo 2368 4.5 4.0 -11%
Correa 3813 4.9 4.3 -12%
Abreu 5506 3.0 2.5 -16%
Bell 3406 1.5 1.2 -18%
Navarez 2083 2.6 2.1 -20%
Sanchez, G. 2665 3.2 2.3 -30%
Gallo 2811 3.2 2.1 -33%
Happ 2438 2.6 1.7 -35%
Contreras, Will 2839 3.3 2.1 -35%
Martinez, J 5891 2.7 1.6 -40%
Naquin 1811 1.2 0.6 -50%
Mancini 3119 1.4 0.4 -68%
I'll take requests.
And this is ridiculous. I'm looking at Hosmer and thinking, this is Mitch Moreland V2.0. So I ran Moreland's numbers. 0.9, 1.6 = 83% Hosmer 0.9, 1.6 = 76% Moreland
|
|
|
|
Post by bcsox on Nov 9, 2022 20:54:10 GMT -5
Do we have any way to retroactively look at Papi’s “clutch” numbers. I recall reading something years back that his clutch numbers which somehow were mostly his playoff numbers were not much different and a tad lower than his regular season numbers. My eyes told me a whole different story on this topic.
|
|
ericmvan
Veteran
Supposed to be working on something more important
Posts: 8,925
Member is Online
|
Post by ericmvan on Dec 7, 2022 19:26:26 GMT -5
So, "clutch" hitting is supposedly random. Well, you have to measure it correctly before you make that assertion. And they are not. I've fixed that. (Explanation in the next post). Here are JDM's Situational Hitting numbers since he broke out in 2014, in Wins per 600 PA: 2014 -0.97
2015 -1.54
2016 -1.52
2017 -1.39
2018 -0.18
2019 -2.38
2020 -0.24
2021 -0.36
2022 -1.74
That's not random. And his 26.5 total WAR, based on the notion that situational hitting is random, becomes 17.6 when you factor it in. As I ran the numbers for 14 other guys, I noticed that a lot of them got markedly better after 2 seasons (sometimes 3).
Here's everyone's Situational Wins per 600 PA, first two years and after: Name Y1-2 Post Diff SSS? Note
Murphy -3.80 -0.71 3.09 Y
Verdugo -1.56 1.22 2.78 Y -.65, 2.03 after 3
Haniger -1.85 0.84 2.69
Laureano -0.08 2.03 2.11
McGuire -2.41 -0.43 1.98 Y
Conforto -1.16 0.39 1.55
Bogaerts -0.94 0.19 1.13 But .72 up to 2019
Devers 0.13 0.79 0.66 -.20, .84 after 3
Abreu -0.43 -0.21 0.22
Story 0.09 -0.07 -0.16
Hosmer 1.17 0.82 -0.35
Hernandez 1.37 0.49 -0.88
Contrears -0.49 -1.61 -1.12
Happ 0.06 -1.10 -1.16
The small sample guys have big differences, and that immediately suggests that the whole phenomenon involves relaxing into your status as a big-leaguer. You might break down the first and second halves of first and second seasons and find a meaningful improvement in a majority of hitters.
Fun fact: Hosmer's Sit-Hit adjustment almost doubles his career WAR (10.3 70 21.4).
Who else besides Murphy and Laureanoshould I run full numbers on? |
|
|
|
Post by trajanacc on Dec 7, 2022 19:37:03 GMT -5
Thanks as always for your unique posts and contributions. Always fun stuff to read.
Perhaps not as relevant as potential FA signings, but I’d be interested in the numbers for some of the guys we have moved on from in the last few years. Betts, JBJ, Renfroe, Vazquez, Benintendi etc.
|
|
ericmvan
Veteran
Supposed to be working on something more important
Posts: 8,925
Member is Online
|
Post by ericmvan on Dec 7, 2022 19:55:18 GMT -5
Thanks as always for your unique posts and contributions. Always fun stuff to read. Perhaps not as relevant as potential FA signings, but I’d be interested in the numbers for some of the guys we have moved on from in the last few years. Betts, JBJ, Renfroe, Vazquez, Benintendi etc. Thanks for the praise. I do this for myself (out of curiosity and often against my will  ), so sharing is a natural next step.
That would be very interesting. And they might bring CV back!
I've got a scrumptious spreadsheet that can spit out a player's year-by-year numbers in something like 30 seconds ... putting them here is much more timr-consuming.
|
|
|
|
Post by xdmo on Dec 7, 2022 20:02:43 GMT -5
Moved this tweet here.
|
|
|
|
Post by Guidas on Dec 7, 2022 21:38:02 GMT -5
Thanks as always for your unique posts and contributions. Always fun stuff to read. Perhaps not as relevant as potential FA signings, but I’d be interested in the numbers for some of the guys we have moved on from in the last few years. Betts, JBJ, Renfroe, Vazquez, Benintendi etc. Thanks for the praise. I do this for myself (out of curiosity and often against my will ), so sharing is a natural next step. That would be very interesting. And they might bring CV back! I've got a scrumptious spreadsheet that can spit out a player's year-by-year numbers in something like 30 seconds ... putting them here is much more timr-consuming. My hagiography: Manny always seemed to get a hit with a man on second when they needed it. Was he good in the clutch or just average? P.S. I always believed Manny could fall out of bed from a dead sleep, stand up, grab a piece of barbed wire and get a hit off any pitcher in baseball, so I may be a bit biased. |
|
|
|
Post by lonborgski on Dec 19, 2022 7:22:54 GMT -5
“Who else besides Murphy and Laureanoshould I run full numbers on?“
Justin Turner
|
|
ericmvan
Veteran
Supposed to be working on something more important
Posts: 8,925
Member is Online
|
Post by ericmvan on Dec 21, 2022 6:44:07 GMT -5
Thanks for the praise. I do this for myself (out of curiosity and often against my will ), so sharing is a natural next step. That would be very interesting. And they might bring CV back! I've got a scrumptious spreadsheet that can spit out a player's year-by-year numbers in something like 30 seconds ... putting them here is much more timr-consuming. My hagiography: Manny always seemed to get a hit with a man on second when they needed it. Was he good in the clutch or just average? P.S. I always believed Manny could fall out of bed from a dead sleep, stand up, grab a piece of barbed wire and get a hit off any pitcher in baseball, so I may be a bit biased. Just saw this. It's a bit of work to create the background data for a given MLB season, so extending my spreadsheet to cover Manny's career would be a lot of work.
Manny's career OPS+ by leverage, where 100 is himself overall, 105, 103, 93. He was 83 Late and Close, so he didn't have the tough-relievers-no-problem thing going, but he was 90 with the bases empty, 106 with a man in 1st, and 112 with RISP.
.297 / .385 / .562 career bases empty .327 / .454 / .594 career RISP
That certainly backs up your memory.
|
|
ericmvan
Veteran
Supposed to be working on something more important
Posts: 8,925
Member is Online
|
Post by ericmvan on Dec 25, 2022 4:11:57 GMT -5
I've seen enough raw data to be fairly certain that the Situational Hitting adjustment -- which, keep in mind, actually does measure actual player value, while unadjusted WAR does not -- can be predictive.
I've run the yearly numbers on 26 guys. Seven of those have 8 or more seasons of data. If this were random, you'd expect the majority of players to have roughly as many seasons where they had a positive differential as a negative, right? What we have is this (I'm including 2020 if it's part of a streak, or huge) Eric Hosmer 11 - 1 (the exception being his walk year)
Elvis Anrdus 9 -3 Juan Segura 5-4 Xander Bogaerts 4-5 Jose Abreu 1 - 6 (and a season of exactly 0.0) Carlos Correa 2-6, with a current 6 year negative streak J.D. Martinez 0-8 (starting after the Astros released him)
That's nothing like the expectation. It's very tough to show this is real, in part because this is a stat that centers at zero, which means that outlier seasons muck up year-to-year correlations. But I've come up with 2 approaches that should work around this. More thoughts on this in the Clutch thread I started, at some point soon.
How? I have no clue what you're actually doing, you aren't posting formulas. So I can't say your formula does what you think it does. You know this, you know the leg work that needs to be done. You're one of the few people that likely understands the crazy work that done for war and what changing it would actually require. You know Fangraphs isn't doing 26 players and changing the war formulas. Your talking whole league over years I'm not changing WAR at all. I'm simply adding an extra component of the game that isn't being measured. I've mentioned what it is I'm doing, but here's the whole method.
First, Win Probability Added as measured by FG and B-Ref is incorrect. Last year all the hitters had a WPA that was .27 wins too low per 600 PA. I've corrected it, a different correction factor for each year. The explanation for this is at the end.
Second, I determine each season's Runs per Win from FanGraph's data, in order to be able to convert them into one another.
Third, I grab FG's Batting Runs above Average, and convert it into Wins Above Average. [1]
This gives us the situation-neutral Batting Wins Above Average.
WPA gives you the situation-adjusted Batting Wins Above Average (and a bit more; see below).
WPA - BWAA gives you the Situational Hitting differential.
Now, if you add this number to WAR you are double-counting the easily measured components of base running -- stolen bases and caught stealing, pick-offs and advances on passed balls and wild pitches. But WPA already credits hitters with the baserunning performances of guys on base (more on that below). These inaccuracies are very likely smaller than the error bars in fielding ratings. Fixing them is a job for someone getting paid to do it.
When you use (adjusted) WPA instead of Batting Wins Above Average, there's no double-counting, and you have included not just the win values of the aforementioned baserunning plays, but their situational importance.
[1] Fun fact: last year, according to FG, all the hitters in baseball were collectively 62.2 Runs Above Average! What a neat trick. Fortunately, that amounts to just .02 wins per 600 PA, so I decided to not bother correcting it. My guess is that they do recalculate the run value of each event each season -- a staple of the linear weights methodology -- and have never noticed that they need to go to an extra decimal point for each run value to avoid this obvious error. It looks much worse than it is in practice.
----
Finally, I am sympathetic to the plight of the hitter who starts a rally from down a bunch of runs and gets just a tiny WPA, even if his hit is crucial. I actually had a conversation about this with Clay Davenport years ago at a SABR conference. Nothing was concluded, IIRC.
My fix is this: whenever a run or runs are scored in an inning, you also take the run value of each event in the inning, and use that to divide up the increase in Win Probability of the inning. So if there are two outs and nobody on in the bottom of the ninth, down two runs, and you win the game with 5 straight singles, each of the five guys who hit a single gets, at this stage, equal credit for the win, just under .2 wins each. WPA has it as .03, .05, .08, .44, .38 (with no runners going first to third).
You then average this with WPA and call it "Team-Context WPA." In this example the five guys would get .11, .12, .14, .32, .29, which seems to track our subjective sense of the importance of each PA better than the actual odds! (This is the first time I ever ran numbers on this idea, which has to be close to 15 years old).
It is of course partly a measure of the hitters hitting after a given hitter, but it still might identify hitters each year who had a good or bad season at starting rallies.
----
Why all WPA figures are off, and how to correct them.
The sum of the league WPA for both hitters and pitchers needs to be zero. Offense and defense are equally important by definition, and so they must have the same total WPA at season's end, and since every game has a winner and loser, the sum of each must be zero.
When the pitchers start to dominate the game a bit and scoring drops, that makes the value of hits greater, and the value of preventing hits less great, because the former has become harder and the latter easier. As a result, the win values of each event need to be adjusted to conform to the level of offense in a given year.
That has not been done. As a result, all the pitchers are combining to have a supposed positive effect on wins and all hitters are combining to have a negative effect. We know that's impossible. And it makes no sense; the improved pitchers are not extra responsible for teams winning, they're just responsible for less scoring. If the former were true, pitchers would be getting better contracts and hitters worse ones.
So my first step is to sum the supposed pitching WPA of each season (the hitting WPA is always the negative of that) and calculate how many runs are being mistakenly credited per 600 PA. The WPA of every hitting season is adjusted based on the year's error size and the hitter's PA.
This is not a precise adjustment. There should be a new set of Win Probabilities for every situation, created at the end of each year. That will change the measured WPA of every hitter ... but the differences are again, likely small. The adjustment for baserunning is likely quite a bit larger -- i.e., a double with a man on first should get the average WPA of that event with an average baserunner and the actual fielder handling the ball, and the baserunner gets the difference from average -- the positive value of scoring when it was unexpected, or the negative of failing to do so when it was easy. You'd track that separately.
|
|
ericmvan
Veteran
Supposed to be working on something more important
Posts: 8,925
Member is Online
|
Post by ericmvan on Dec 25, 2022 7:28:17 GMT -5
In 2021 Alex Verdugo was 2/20 (both singles) in his 20 least important PA ....
And he hit .448 / .476 / .658 in his most important 42 PA. That's excluding 4 IBB, which boost his OBP to .522.
This last year Verdugo went 1/31 in his 31 PA with a Leverage Index of .04 or less ...
And he hit .429 / .471 / .429 in his most important 17 PA.
|
|
manfred
Veteran
Posts: 11,397 
Member is Online
|
Post by manfred on Dec 25, 2022 9:00:50 GMT -5
In 2021 Alex Verdugo was 2/20 (both singles) in his 20 least important PA ....
And he hit .448 / .476 / .658 in his most important 42 PA. That's excluding 4 IBB, which boost his OBP to .522.
This last year Verdugo went 1/31 in his 31 PA with a Leverage Index of .04 or less ...
And he hit .429 / .471 / .429 in his most important 17 PA.
Man, if I was his agent I’d be pissed. Guy has average numbers because he can’t keep motivated? Maybe he shouldn’t dog it half the time? Does he run full soeed when it doesn’t matter? |
|
jimoh
Veteran
Posts: 3,975
|
Post by jimoh on Dec 25, 2022 9:06:45 GMT -5
In 2021 Alex Verdugo was 2/20 (both singles) in his 20 least important PA ....
And he hit .448 / .476 / .658 in his most important 42 PA. That's excluding 4 IBB, which boost his OBP to .522.
This last year Verdugo went 1/31 in his 31 PA with a Leverage Index of .04 or less ...
And he hit .429 / .471 / .429 in his most important 17 PA.
“you can step right up step right up that's right, it filets it chops it dices slices never stops lasts a lifetime mows your lawn and it mows your lawn and it picks up the kids from school it gets rid of unwanted facial hair it gets rid of embarrassing age spots it delivers a pizza and it lengthens and it strengthens and it finds that slipper that's been at large under the chaise lounge for several weeks and it plays a mean Rhythm Master it makes excuses for unwanted lipstick on your collar and it's only a dollar step right up it's only a dollar step right up 'cause it forges your signature if not completely satisfied mail back unused portion of product for complete refund of price of purchase” |
|
|
|
Post by soxin8 on Dec 31, 2022 13:13:18 GMT -5
|
|
|
|
Post by umassgrad2005 on Dec 31, 2022 13:31:19 GMT -5
How? I have no clue what you're actually doing, you aren't posting formulas. So I can't say your formula does what you think it does. You know this, you know the leg work that needs to be done. You're one of the few people that likely understands the crazy work that done for war and what changing it would actually require. You know Fangraphs isn't doing 26 players and changing the war formulas. Your talking whole league over years I'm not changing WAR at all. I'm simply adding an extra component of the game that isn't being measured. I've mentioned what it is I'm doing, but here's the whole method.
First, Win Probability Added as measured by FG and B-Ref is incorrect. Last year all the hitters had a WPA that was .27 wins too low per 600 PA. I've corrected it, a different correction factor for each year. The explanation for this is at the end.
Second, I determine each season's Runs per Win from FanGraph's data, in order to be able to convert them into one another.
Third, I grab FG's Batting Runs above Average, and convert it into Wins Above Average. [1]
This gives us the situation-neutral Batting Wins Above Average.
WPA gives you the situation-adjusted Batting Wins Above Average (and a bit more; see below).
WPA - BWAA gives you the Situational Hitting differential.
Now, if you add this number to WAR you are double-counting the easily measured components of base running -- stolen bases and caught stealing, pick-offs and advances on passed balls and wild pitches. But WPA already credits hitters with the baserunning performances of guys on base (more on that below). These inaccuracies are very likely smaller than the error bars in fielding ratings. Fixing them is a job for someone getting paid to do it.
When you use (adjusted) WPA instead of Batting Wins Above Average, there's no double-counting, and you have included not just the win values of the aforementioned baserunning plays, but their situational importance.
[1] Fun fact: last year, according to FG, all the hitters in baseball were collectively 62.2 Runs Above Average! What a neat trick. Fortunately, that amounts to just .02 wins per 600 PA, so I decided to not bother correcting it. My guess is that they do recalculate the run value of each event each season -- a staple of the linear weights methodology -- and have never noticed that they need to go to an extra decimal point for each run value to avoid this obvious error. It looks much worse than it is in practice.
----
Finally, I am sympathetic to the plight of the hitter who starts a rally from down a bunch of runs and gets just a tiny WPA, even if his hit is crucial. I actually had a conversation about this with Clay Davenport years ago at a SABR conference. Nothing was concluded, IIRC.
My fix is this: whenever a run or runs are scored in an inning, you also take the run value of each event in the inning, and use that to divide up the increase in Win Probability of the inning. So if there are two outs and nobody on in the bottom of the ninth, down two runs, and you win the game with 5 straight singles, each of the five guys who hit a single gets, at this stage, equal credit for the win, just under .2 wins each. WPA has it as .03, .05, .08, .44, .38 (with no runners going first to third).
You then average this with WPA and call it "Team-Context WPA." In this example the five guys would get .11, .12, .14, .32, .29, which seems to track our subjective sense of the importance of each PA better than the actual odds! (This is the first time I ever ran numbers on this idea, which has to be close to 15 years old).
It is of course partly a measure of the hitters hitting after a given hitter, but it still might identify hitters each year who had a good or bad season at starting rallies.
----
Why all WPA figures are off, and how to correct them.
The sum of the league WPA for both hitters and pitchers needs to be zero. Offense and defense are equally important by definition, and so they must have the same total WPA at season's end, and since every game has a winner and loser, the sum of each must be zero.
When the pitchers start to dominate the game a bit and scoring drops, that makes the value of hits greater, and the value of preventing hits less great, because the former has become harder and the latter easier. As a result, the win values of each event need to be adjusted to conform to the level of offense in a given year.
That has not been done. As a result, all the pitchers are combining to have a supposed positive effect on wins and all hitters are combining to have a negative effect. We know that's impossible. And it makes no sense; the improved pitchers are not extra responsible for teams winning, they're just responsible for less scoring. If the former were true, pitchers would be getting better contracts and hitters worse ones.
So my first step is to sum the supposed pitching WPA of each season (the hitting WPA is always the negative of that) and calculate how many runs are being mistakenly credited per 600 PA. The WPA of every hitting season is adjusted based on the year's error size and the hitter's PA.
This is not a precise adjustment. There should be a new set of Win Probabilities for every situation, created at the end of each year. That will change the measured WPA of every hitter ... but the differences are again, likely small. The adjustment for baserunning is likely quite a bit larger -- i.e., a double with a man on first should get the average WPA of that event with an average baserunner and the actual fielder handling the ball, and the baserunner gets the difference from average -- the positive value of scoring when it was unexpected, or the negative of failing to do so when it was easy. You'd track that separately.
So you backed check the data for the whole season making sure positional players got 570 fwar? That's fangraphs war system based on replacement players and their win percentage. 1,000 fwar per season, 570 to positional players, 430 to pitchers. You can't just adjust numbers because you think it makes more sense. It has to fit into the war system and calculations. |
|
ericmvan
Veteran
Supposed to be working on something more important
Posts: 8,925
Member is Online
|
Post by ericmvan on Jan 8, 2023 3:27:00 GMT -5
I'm not changing WAR at all. I'm simply adding an extra component of the game that isn't being measured.
So you backed check the data for the whole season making sure positional players got 570 fwar? That's fangraphs war system based on replacement players and their win percentage. 1,000 fwar per season, 570 to positional players, 430 to pitchers. You can't just adjust numbers because you think it makes more sense. It has to fit into the war system and calculations. Assuming that the average for Batting (Runs >) Wins Above Average is close enough to zero (and I checked, and it's very close, off by accumulated rounding errors), then the situational hitting adjustments sum to zero as well. They have to. Win Probability Added also needs to sum to zero because it's relative to average. So all I'm doing here is measuring two diffeent sum-to-zero measures, WPA and BWA, and crediting or debiting the hitters who have more or less WPA (which is situational) than BWA (which is not).
It's really simple. And everything in the WAR system works the same way except the translation from Wins Above Average to Wins Above Replacement. It's just a sum of measurable skills, above or below average, and then you convert the resulting Wins Above Average to Wins Above Replacement by adding positional, playing time, and league quality adjustments. You don't have to consider the Average to Replacement element at all when you are measuring a new skill to include in WAR, like, e.g., pitch framing.
|
|
|
|
Post by umassgrad2005 on Jan 8, 2023 7:31:04 GMT -5
So you backed check the data for the whole season making sure positional players got 570 fwar? That's fangraphs war system based on replacement players and their win percentage. 1,000 fwar per season, 570 to positional players, 430 to pitchers. You can't just adjust numbers because you think it makes more sense. It has to fit into the war system and calculations. Assuming that the average for Batting (Runs >) Wins Above Average is close enough to zero (and I checked, and it's very close, off by accumulated rounding errors), then the situational hitting adjustments sum to zero as well. They have to. Win Probability Added also needs to sum to zero because it's relative to average. So all I'm doing here is measuring two diffeent sum-to-zero measures, WPA and BWA, and crediting or debiting the hitters who have more or less WPA (which is situational) than BWA (which is not).
It's really simple. And everything in the WAR system works the same way except the translation from Wins Above Average to Wins Above Replacement. It's just a sum of measurable skills, above or below average, and then you convert the resulting Wins Above Average to Wins Above Replacement by adding positional, playing time, and league quality adjustments. You don't have to consider the Average to Replacement element at all when you are measuring a new skill to include in WAR, like, e.g., pitch framing.
Yes you do, you have to backcheck and make sure it does what it's suppose too. You know this, a few small rounding errors when talking every player in the league can be huge. You want to redefine war and create new numbers do the leg work you know has to be done! Look at your adjusts for certain guys they are huge and war is tied to wins. Heck at minimum do a full team and see if it works for even one single team! You know if you worked for Red Sox and took this to them, they'd say show me it works, do the leg work. |
|
jimoh
Veteran
Posts: 3,975
|
Post by jimoh on Jan 8, 2023 8:05:21 GMT -5
Not sure how there is a new thread with all of Eric's December claims about clutch and EricWAR (and EricWPA?--it's not clear) and few voices of dissent.
As argued, still not convincing or trustworthy.
|
|
|
|
Post by pappyman99 on Jan 8, 2023 10:33:17 GMT -5
Clutch is definitely very real, some people can’t handle large pressure situations and others can. That is a reality in all of life not just sports.
I’d take Ortiz and Jeter over Arod in a late situation any day of the week.
With that said, I don’t trust or believe in anything that tries to measure that. Not possible to measure human emotions
|
|
|
|
Post by seamus on Jan 9, 2023 12:43:55 GMT -5
Are you able to provide a step-by-step for how you calculated a single player as an example? That might make it easier to assess the metric and either make constructive critiques or start to buy-in.
|
|
jimoh
Veteran
Posts: 3,975
|
Post by jimoh on Jan 9, 2023 13:16:06 GMT -5
Clutch is definitely very real, some people can’t handle large pressure situations and others can. That is a reality in all of life not just sports. I’d take Ortiz and Jeter over Arod in a late situation any day of the week. With that said, I don’t trust or believe in anything that tries to measure that. Not possible to measure human emotions We all know that "some people can’t handle large pressure situations and others can." But those who are skeptical of clutch hitting in mlb think that most people who cannot handle pressure are weeded out in the minor leagues, when the pressure is "if you can't play you go home." |
|
|
), so sharing is a natural next step.
