|
Post by bleedthefreak on Apr 11, 2024 15:06:23 GMT -8
Pitchers losing 3 or more war: Pitch Hit 2.0 Tot 1.0 Tot Diff Player 40.6 (1.3) 39.3 50.0 (10.71) Wilbur Wood 53.4 2.4 55.8 66.1 (10.30) Luis Tiant 38.7 0.0 38.7 45.3 (6.63) Brad Radke 31.6 5.3 36.9 43.0 (6.13) Ned Garver 87.1 2.7 89.8 95.9 (6.11) Phil Niekro 50.4 0.0 50.4 56.4 (6.02) Dave Stieb 44.3 5.7 50.0 56.0 (6.02) Orel Hershiser 45.0 (0.1) 44.9 50.5 (5.57) Kenny Rogers 29.3 0.0 29.3 34.4 (5.08) Tim Wakefield 26.2 4.2 30.4 35.1 (4.70) Woody Williams 45.2 0.2 45.4 50.1 (4.66) Mark Langston 36.2 0.3 36.5 41.1 (4.60) Tom Candiotti 32.6 6.7 39.3 43.9 (4.57) Carlos Zambrano 28.1 0.0 28.1 32.6 (4.51) Pat Hentgen 31.0 5.4 36.4 40.9 (4.47) Catfish Hunter 76.5 0.5 77.0 81.3 (4.28) Nolan Ryan 134.2 0.8 135.0 139.2 (4.23) Roger Clemens 50.5 (0.2) 50.3 54.5 (4.21) Kevin Appier 32.0 0.8 32.8 37.0 (4.20) Dutch Leonard 1 58.0 0.1 58.1 62.1 (3.95) Dennis Eckersley 26.8 0.0 26.8 30.3 (3.52) Teddy Higuera 44.6 1.7 46.3 49.8 (3.50) Eddie Rommel 35.4 (0.3) 35.1 38.4 (3.32) Charlie Hough 36.5 0.0 36.5 39.8 (3.30) Danny Darwin 28.3 0.1 28.4 31.7 (3.29) Aaron Nola 47.6 0.8 48.4 51.7 (3.28) Johan Santana 35.8 0.2 36.0 39.0 (3.02) Howard Ehmke
|
|
|
Post by chriscobb on Apr 11, 2024 15:19:55 GMT -8
nWAR 2.0 doesn't seem to like knuckleballers: Wood, Niekro, Wakefield, Candiotti, Hough. Almost 20% of the pitchers with the biggest drop are knucklers.
|
|
|
Post by Jaack on Apr 11, 2024 15:47:50 GMT -8
It's not a 1-for-1 thing, but some of the names on the pitching side seem very fWAR-ish. Paul Derringer, Bob Friend, and Tommy John in the top 10 gainers, the knuckleball guys on the losers (as well as Dave Stieb and Eddie Rommel). It's not a hard correlation, just an observation, but Paul Derringer is a name that will make me think fWAR every single time I see it.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on Apr 11, 2024 16:30:51 GMT -8
The biggest losers (among position players) seems to be a lot of guys with great defense, and some more guys with *no* defense.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on Apr 11, 2024 16:33:11 GMT -8
Whoa, Derringer is a guy I need to add to my ratings. He slipped through the cracks.
|
|
|
Post by bleedthefreak on Apr 11, 2024 17:59:29 GMT -8
It's not a 1-for-1 thing, but some of the names on the pitching side seem very fWAR-ish. Paul Derringer, Bob Friend, and Tommy John in the top 10 gainers, the knuckleball guys on the losers (as well as Dave Stieb and Eddie Rommel). It's not a hard correlation, just an observation, but Paul Derringer is a name that will make me think fWAR every single time I see it. It also jives largely with Tom Thress won loss records.
|
|
alex02
Hall of Merit Voter
Posts: 39
|
Post by alex02 on Apr 12, 2024 13:20:09 GMT -8
Thanks bleedthefreak for such a thorough review. Much to consider!
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on May 8, 2024 12:20:11 GMT -8
So, I find discussions about pitchers slightly more intriguing than hitters, for a number of reasons. And I think there's a general but muted consensus that HOM might be a little short on pitchers, but we're fractured on "who's next." Let me try my hand at laying out some cases for the top candidates according to last year's voting.
Tommy John: pros = high WAR totals, huge volume, comparable WAA (in aggregate) to other candidates despite the huge volume, good contextual performance, strong postseason performance cons = lack of peakiness, era adequately represented
Kevin Appier pros = strong peak and prime, strong by RA and FIP lenses, good WAA, era underrepresented, very good contextual performance cons = little + lackluster postseason, shortish career, played in an era with many all-time great pitchers, so not at the top of his contemporary cohort
Vic Willis pros = lots of innings, high peak and WAA (by bWAR) cons = weak by FIP, not long career for era, further from top of contemporary cohort, era overrepresented, extreme reliance on defense(?), no postseason, poorish hitting
Tim Hudson pros = standard peak/prime by bWAR, strong contextual performance, decent chunk of acceptable postseason performance, positive hitter, era underrepresented cons = weakish by FIP, overall negative cWPA in postseason, not a standout when viewing from any lense (but not bad from any)
Urban Shocker pros = strong peak and career by bWAR, good hitter, (small?) war credit cons = era overrepresented, weak by FIP, little + lackluster postseason, short career, not at top of contemporaries
Roy Oswalt pros = strong prime and good peak, strong WAA, extremely strong rates by RA and FIP, good postseason sample size of slightly above average, strong contextual performance, era underrepresented, good rank among contemporaries cons = short career, low WAR
Jim McCormick pros = innings + bWAR cons = era overrepresented, not close to top of contemporaries, not even a candidate by FIP, humongous reliance on defense, one of top seasons in UA (with more WAA than WAR that year lol!)
Tommy Bridges pros = war credit, good ERA+, good postseason performance cons = low peak, somewhat high reliance on defense, era overrepresented, mediocre WAA and FIP, war debit
Orel Hershiser pros = good prime, good batting, excellent postseason, era underrepresented cons = mediocre WAA and FIP, mediocre peak, weak contextual stats
Don Newcombe pros = war credit, minor league credit, excellent hitter cons = that's a lot of "what if" credit, awful postseason, no peak (terrible WAA), not a standout by RA or FIP performance
I am not the person to make the case for Tommy Bridges or Don Newcombe - if I've materially misrepresented those guys then correct me. Also, now that I write it all out, I'm not exactly sure how helpful this is except to point out why these guys are the borderline group. Maybe it will ignite some discussion?
|
|
|
Post by dlfrommn on May 8, 2024 14:21:58 GMT -8
Tommy Bridgespros = war credit, good ERA+, good postseason performance cons = low peak, somewhat high reliance on defense, era overrepresented, mediocre WAA and FIP, war debit Bridges missed all of 1944 and most of 1945. Not sure why people would debit him for the war. WWII cut his career off short. His WAA drive his case, he wasn't a bulk innings eater but the innings he did pitch were quite good with a high strikeout rate for his era.
|
|
|
Post by Jaack on May 8, 2024 17:14:41 GMT -8
Don Newcombepros = war credit, minor league credit, excellent hitter cons = that's a lot of "what if" credit, awful postseason, no peak (terrible WAA), not a standout by RA or FIP performance I think my addition here is that Newcombe's hitting is, for me at least, a part of his peak case. His peak as a pitcher is only so-so, and not HoM material, but as an overall player, those top seasons look quite a bit better. But overall, I think Newcombe is a case of extreme circumstance - he had basically everything go against him to get onto the field, played in a tough league that was just brutal on pitchers. Just in context - he wasn't Warren Spahn or Robin Roberts, but he's probably the third best pitcher in the NL in the 1950s, and that's before crediting in his batting or his war years. I think that sort of contextual consideration is key.
|
|
|
Post by chriscobb on May 9, 2024 4:55:52 GMT -8
A quick comment on kcgard's useful list. I note that most of the pitchers that he has listed tend to include the annotation, "weak by FIP," or something along those lines. I don't know how many voters actually have an FIP component in their systems, but I would suggest that those who do think carefully about the applicability of FIP as a measure of value for pre-1960 pitchers and especially for pre-1920 pitchers. I don't have time to review the data in this quick post, but the evidence is pretty clear that what Christy Mathewson called "pitching in a pinch" was an important, almost standard part of pitching strategy prior to the lively ball era, such that the large majority of successful pitchers from the era racked up a large number of what Fangraphs WAR calls "LOB-Wins," which it ascribes to the "sequencing" of events and calculates as the difference between RA/9 WAR and WAR, minus BIP wins. That is, it's not a number derived from analysis of individual events but simply the difference between what all event-based analysis shows and what run-based analysis shows.
Almost all excellent pre-lively-ball era pitchers have a large number of LOB-Wins, which implies (although in the absence of pbp data) that they were good at getting outs when they needed them (although I would guess that unstrategic stolen-base strategies in the period are probably another contributing factor to these numbers (suppressing stolen bases is a piece that contributes to left-handed pitchers generally doing better than right-handed pitchers in the LOB-wins category).
Overall, LOB-Wins account for an increasingly large portion of pitchers' value the farther back one goes in baseball history, so applying a modern-baseball lens to pitcher evaluation is going to be increasingly inaccurate.
I am in the middle of finals week right now, so I can't spare the time to work up a detailed presentation of the historical data on LOB wins, but I should be able to put the data together in the second half of May. But you can see it yourself easily enough by doing a "pitching leaders search" on Fangraphs for LOB-Wins and playing around with the starting and ending dates for groups of pitchers examined.
|
|
|
Post by Jaack on May 9, 2024 12:40:16 GMT -8
The additional aspect of LOB-WAR that's going to be more impactful on older pitchers, particularly the 19th century pitchers, is that some amount of RoE-related value is going to show up there - since RoE count as outs for the BIP calculation (a pretty questionable decision, but that's par for the course with anything to do with errors really).
That being said, I do agree that pitching out of jams is going to be relatively more important in the pre-relief pitching era - it's notable that some of the more recent over-performers in LOB-WAR are the 70s/80s firemen type relievers - Dan Quisenberry, John Hiller, Greg Minton all have high marks.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on May 9, 2024 13:34:40 GMT -8
Tommy Bridgespros = war credit, good ERA+, good postseason performance cons = low peak, somewhat high reliance on defense, era overrepresented, mediocre WAA and FIP, war debit Bridges missed all of 1944 and most of 1945. Not sure why people would debit him for the war. WWII cut his career off short. His WAA drive his case, he wasn't a bulk innings eater but the innings he did pitch were quite good with a high strikeout rate for his era. I generally think it is not very consistent to give war credit without war debit. Bridges missed 44 and much of 45, but 43 was also one of his better seasons and that is a war-weakened year. On balance, Bridges would/should be a gainer by incorporating war credit, but it's not always as straightforward as "missed some time, so needs more credit." In my opinion.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on May 9, 2024 13:53:46 GMT -8
A quick comment on kcgard's useful list. I note that most of the pitchers that he has listed tend to include the annotation, "weak by FIP," or something along those lines. I don't know how many voters actually have an FIP component in their systems, but I would suggest that those who do think carefully about the applicability of FIP as a measure of value for pre-1960 pitchers and especially for pre-1920 pitchers. This is interesting. Let's side aside McCormick, who threw 600 innings pitching underhand and so forth, and talk about the others. I have also thought about this at some length when evaluating pitchers from early 20th century. There were pitchers who were great by FIP back then, but then a big swath of guys who were mediocre at best by FIP who still look good or great by RA. I think it's a real question how much we should credit those pitchers versus their defenses. The issue of having a third of opposing lineups generally unable to hurt you in any way probably plays into it as well. But the issue can get quite complicated to adjust for such things. Defense simply played a larger role in run prevention in those eras than it does in the current era, but I'm not sure WAR actually/accurately captures that dynamic. Again, look at McCormick. I have experimented with different weightings of RA vs FIP WAR for different eras, but haven't found anything I like meaningfully better in the aggregate. A guy here and there I like the new rankings better, another guy worse. But I think it's defensible. I would even say defensible to ignore FIP altogether going back far enough, though I probably wouldn't endorse it unless you do something about defensive contribution of the nonpitchers. I wouldn't cry foul though for someone straight ignoring it, that's a choice a voter has to make. I guess all that to say, it's complicated (you're welcome for such an enlightening comment, guys). To your larger point, really only Willis, Shocker, and Hudson are "weak by FIP" with some of the others being I wouldn't say weak but it's something that maybe neither helps nor hurts their case (and having something not help I guess I'm counting as a de facto con if you're looking for HOM?). Hershiser, Bridges, and Newcombe are in that group. Which makes it 1/3 FIP helping, 1/3 FIP neutral, 1/3 FIP hurting. That in itself I find quite interesting; probably has something to do with how the electorate evaluates pitchers as a collective, and part of the reason it's hard to get consensus. Which makes sense. If all the voters used FIP, we would have resigned Willis to the dustbin by now. If nobody used FIP, he would have been elected 50 elections ago or something. Variety of viewpoints I tend to think is a positive for the HOM - one of its most positive aspects even.
|
|
|
Post by chriscobb on May 9, 2024 15:01:42 GMT -8
The additional aspect of LOB-WAR that's going to be more impactful on older pitchers, particularly the 19th century pitchers, is that some amount of RoE-related value is going to show up there - since RoE count as outs for the BIP calculation (a pretty questionable decision, but that's par for the course with anything to do with errors really). Wow. I had no idea that was the case. Where does Fangraphs explain that? I imagine that accounts for some of the discrepancies between the fielding support evaluations in BBRef's pitching WAR and BIP-Wins. I'm going to have to think about the modeling implications of this choice. My initial thought is that it is going to result in more BIP-Wins for earlier pitchers than may be warranted. I would think that it would make more sense to count them as hits. If you are looking at balls-in-play from an FIP perspective, once the ball is in play, the outcome is determined by the fielders, not the pitcher, so whether the batter reached on a hit or on an error doesn't matter from the perspective of what the pitcher accomplished, but if you want to accurately track how many wins on balls in play a pitcher picked up, I'd think you would want to be working with accurate numbers with respect to baserunners and outs!
Counting the errors as outs is going to increase the BIP-Wins, which would necessarily decrease the LOB-Wins, since LOB-Wins = (RA/9 Wins-BIP-Wins)-(FIP Wins)
Does that seem like an accurate interpretation of calculating BIP-Wins with ROEs converted to outs??
What a strange choice . . .
|
|
|
Post by chriscobb on May 9, 2024 15:04:37 GMT -8
That being said, I do agree that pitching out of jams is going to be relatively more important in the pre-relief pitching era - it's notable that some of the more recent over-performers in LOB-WAR are the 70s/80s firemen type relievers - Dan Quisenberry, John Hiller, Greg Minton all have high marks.
I noticed that as well. Are LOB-Wins documenting real value for these guys, then,--their ability to shut down trouble?
|
|
|
Post by esteban on May 9, 2024 15:24:33 GMT -8
I'm someone who doesn't use FIP in my historical evaluations of pitchers. I find that it makes sense as a predictive evaluation of how pitchers will perform but it does not make sense to me to use FIP for evaluating historical records. You don't pitch in a vacuum and the defense behind you is part of your arsenal. The only instance I'll use some measure of FIP as part of a historical evaluation is for relievers.
|
|
|
Post by bleedthefreak on May 9, 2024 18:30:36 GMT -8
So, I find discussions about pitchers slightly more intriguing than hitters, for a number of reasons. And I think there's a general but muted consensus that HOM might be a little short on pitchers, but we're fractured on "who's next." Let me try my hand at laying out some cases for the top candidates according to last year's voting. Tim Hudsonpros = standard peak/prime by bWAR, strong contextual performance, decent chunk of acceptable postseason performance, positive hitter, era underrepresented cons = weakish by FIP, overall negative cWPA in postseason, not a standout when viewing from any lense (but not bad from any) First, thanks for posting on pitchers, with less consensus on them, extra discussion would be enlightening for myself, and I would hope others.
I'm continuing to work through the Baseball Projection data, so I'll be able to post on that more not too long into the future, but I may be relatively quiet until I'm caught up on that front.
On Hudson, maybe I'm the only member that leverages Tom Thress Win-Loss records much, but Huddy does quite well.
From the group of pitchers, approximate Key stat value converted to 100 as a baseline HOFer: ~125 - Urban Shocker (with a healthy 1918 war credit bump)
120.7 - Tim Hudson 118.5 - Tommy John ~104 - Tommy Bridges (with moderate war/PCL credit)
98.2 - Roy Oswalt 97.2 - Orel Hershiser
89.2 - Kevin Appier
~88 - Don Newcombe (though it's quite a swag good or bad on the integration/war credit assessment.
My early interpretations of Baseball Projection values, setting peak/prime/career weight, with incorporating some league strength adjustments from Eric Chalek's work, along with some debits for pre-integration, expansion, war seasons, etc, and trying to come up with reasonable war and minor league credits, just regular season, no park or clutch adjustments... 6,676 - Urban Shocker 6,617 - Dwight Gooden 6,444 - Larry Jackson
6,395 - Don Newcombe *6,223* - Adam Wainwright
6,209 - Felix Hernandez 6,135 - Bucky Walters
6,116 - Kevin Appier
6,099 - Roy Oswalt *6,092* - Jon Lester 5,965 - Carl Mays
5,926 - Tim Hudson
5,919 - Jerry Koosman 5,917 - Chuck Finley 5,841 - George Uhle
5,810 - Eddie Cicotte 5,803 - Lefty Gomez
5,757 - Tommy John *5,755* - Cole Hamels
5,755 - Frank Viola 5,710 - Babe Adams 5,708 - Dolf Luque
5,701 - Dizzy Trout
5,648 - Orel Hershiser
5,635 - Jimmy Key
5,620 - Bob Friend 5,612 - Herb Pennock
5,605 - Claude Osteen 5,601 - Cliff Lee
5,592 - Jack Quinn 5,536 - Hippo Vaughn 5,533 - Jim Kaat 5,502 - Wilbur Cooper 5,443 - Bob Shawkey 5,440 - Mel Harder 5,422 - Burleigh Grimes 5,406 - Frank Tanana
5,348 - Mark Buehrle
5,336 - Tommy Bridges 5,330 - Ron Guidry 5,297 - Sal Maglie 5,243 - Camilo Pascual 5,234 - Curt Simmons 5,212 - Jose Rijo
|
|
|
Post by Jaack on May 9, 2024 20:35:01 GMT -8
Wow. I had no idea that was the case. Where does Fangraphs explain that? I imagine that accounts for some of the discrepancies between the fielding support evaluations in BBRef's pitching WAR and BIP-Wins. I'm going to have to think about the modeling implications of this choice. My initial thought is that it is going to result in more BIP-Wins for earlier pitchers than may be warranted. I would think that it would make more sense to count them as hits. If you are looking at balls-in-play from an FIP perspective, once the ball is in play, the outcome is determined by the fielders, not the pitcher, so whether the batter reached on a hit or on an error doesn't matter from the perspective of what the pitcher accomplished, but if you want to accurately track how many wins on balls in play a pitcher picked up, I'd think you would want to be working with accurate numbers with respect to baserunners and outs! Counting the errors as outs is going to increase the BIP-Wins, which would necessarily decrease the LOB-Wins, since LOB-Wins = (RA/9 Wins-BIP-Wins)-(FIP Wins) Does that seem like an accurate interpretation of calculating BIP-Wins with ROEs converted to outs??
What a strange choice . . .
It's a pretty unfortunate choice, but it's pretty clear based on the calcuation that they give for BIP-Wins here. Counting RoE as singles would be clearly the better way to go about it, but that's the type of change that probably should happen across all BABIP-adjacent stats. Counting RoE as hits would raise the league BABIP, so pitchers who had fewer RoE than the league wide rate are going to have some of what should be BIP-WAR counted as LOB-WAR, while pitchers who had more RoE than the league wide rate are going to see the opposite effect. Now that I'm thinking about it a bit more, nailing down the obvious mechanic that would do this is hard. They aren't getting credited directly for stranding runners (none of FIP, BIP, or RA9-WAR are going to 'see' inherited runners). But what it might be is that partial inning relievers are simply less likely to allow their own runners to score, since their runners will often only have one or two outs to score. If Dan Quisenberry enters a bases loaded, two out game, walks three guys, and then gets the last out, that is technically 1/3 scoreless innings for him by RA9, but pretty awful by FIP standards (and the same for BIP if it's three singles). That's the only thing I can see that would cause that genre of pitcher to show strongly by LOB-WAR.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on May 17, 2024 10:29:04 GMT -8
bleed, did your BP values include strike credit for any players? No park adjustments seems like...a somewhat serious missing component. Part of the reason Buehrle shows so poorly? Is Larry Jackson the only pitcher in the top, roughly half(?), of the list who played in a hitter's environment? Urban Shocker looks like the majority of his career was as well.
|
|
|
Post by bleedthefreak on May 18, 2024 10:19:28 GMT -8
bleed, did your BP values include strike credit for any players? No park adjustments seems like...a somewhat serious missing component. Part of the reason Buehrle shows so poorly? Is Larry Jackson the only pitcher in the top, roughly half(?), of the list who played in a hitter's environment? Urban Shocker looks like the majority of his career was as well. Apologies, wasn't clear. Strike credit estimates included. I had no additional park adjustments beyond what the baseline WAR calculation is. The example could be Wade Boggs, I didn't make any adjustments for the fact that he has huge home/road splits. Buehrle was just didn't have much for peaks seasons at BP.
|
|
|
Post by bleedthefreak on May 18, 2024 10:20:20 GMT -8
Buehrle has 56 BP WAR.
|
|
|
Post by chriscobb on May 18, 2024 16:28:05 GMT -8
Fielding-Dependent and Fielding-independent Wins Study
I'm going to take up the question of the meaning of FIP-WAR as a measure of pitcher effectiveness for pitchers prior to World War 2. I have never included FIP-WAR into my system, but I have been interested in FIP-WAR + BIP-wins + LOB-wins as a way of compartmentalizing pitcher value and as a way of cross-checking the fielding support adjustments incorporated in the BBRef's RA/9-based pWAR. What I set out to do is to establish the extent to which FDP-Wins in general and LOB-Wins in particular correlate to other measures of pitching value. To the extent that FDP-Wins are unaffected by the pitcher, then their value should be randomly distributed across pitchers. Insofar as they are not randomly distributed, the patterns of distribution should reveal something about how FDP-wins are produced. These patterns may not have anything to do with pitcher skills: for example, it’s clear that modern relief pitchers generally have positive LOB-win totals: these, as Jaack has suggested, do not point to pitcher skill but to the tendency of relievers to enter an inning when there are already 1 or 2 outs, thus reducing the likelihood that baserunners they allow in that inning will score. Conversely, in the relief pitcher era, it should be expected that starting pitchers, who frequently leave games with baserunners they have allowed on base, to have lower LOB-wins totals as a result. Because the Hall of Merit is interested mostly in starting pitchers, I set up a study that would focus largely on making cross-period comparisons of the distribution of FDP values in relation to other measures of pitching value. First, I set up two periods of 70 seasons: 1871-1940 and 1955-2024. For pitchers throwing 1500 or more innings in their career during those periods, I calculated the following: 1) the percentage of the pitchers who had positive and negative BIP-wins, LOB-wins, and FDP-wins overall in each period 2) the difference between the rate at which the better pitchers in the period cohort accrued FDP-wins across their careers and the rate at which the worse pitchers in the period cohort accrued FDP-wins, with rate expressed as FDP-wins/1500 IP. I calculated and compared rates sorting pitchers by three measure of pitching quality: innings pitched, career FIP-wins, and career FIP-wins/1500 IP. For innings pitched, I compared the pitchers in the cohort with 3000+ IP to pitchers in the cohort with 1500-1999.7 IP. For career FIP-wins, I compared the top X pitchers in the cohort to the bottom Y pitchers in the cohort, where X and Y were equal to the number of pitchers in the cohort with 3000+ IP and to the number of pitchers in the cohort with 1500-1999.7 IP, respectively. For career FIP-wins/1500 IP, I again compared the top X pitchers in the cohort to the bottom Y pitchers in the cohort. Finally, to get a more granular look at early pitchers, I did the same study for pitchers from 1871-1910. It would be fairly easy to do the same study for other periods for which a more precise assessment of the arrangement of pitching value in that period was needed. I’ll now share the results for each of the three cohorts studied. PART 1: DISTRIBUTION OF POSITIVE AND NEGATIVE LOB-WINS, BIP-WINS, AND FDP-WINS OVERALL IN EACH PERIOD COHORT LOB-Wins1871-1910, 1500+ IP cohort 79 Pitchers with positive LOB-wins 41 Pitchers with negative LOB-wins 65.8% of cohort positive From 1871-1940, 1500+ IP cohort180 Pitchers with positive LOB-wins 86 Pitchers with negative LOB-wins 67.7% of cohort positive From 1955-2024, 1500+ IP cohort174 Pitchers with positive LOB-wins 250 Pitchers with negative LOB-wins 41.3% of cohort positive Analysis: Clearly, the distribution of value here is different post-1954 from pre-1941. Several factors are likely to be involved: • Redistribution of LOB-wins from starters to relievers in the post-1954 era • Decline in the use of starters in relief as rigid rotational use of starters becomes the norm. • Decline in use of LOB-win “pitching in a pinch” strategy in post-1954 era as pitchers increasing go all out on every batter • Increase in “index of self-destruction” power-pitching styles as advantages of going all out on every batter begin to outweigh benefits taking a diversified approach to context management. • Redistribution of BIP-wins to LOB-wins in the pre-1941 period due to the treatment of ROE as outs in the BIP-wins formula BIP-wins1871-1910, 1500+ IP cohort 86 Pitchers with positive BIP-wins 34 Pitchers with negative BIP-wins 71.7% of cohort positive 1871-1940, 1500+ IP cohort185 Pitchers with positive BIP-wins 81 Pitchers with negative BIP-wins 69.5% of cohort positive 955-2024, 1500+ IP cohort248 Pitchers with positive BIP-wins 176 Pitchers with negative BIP-wins 58.5% of cohort positive Analysis. Compared to LOB-wins, there is much less indication of a decisive shift in value distribution. Instead, there appears to be a moderate and perhaps steady downward trend. This could be tested by looking at additional 40-year cohorts. I suspect here that the gradual decline may be largely attributable to increasing quality of competition. It is harder for both pitchers and defenders to be above average, so the average point moves closer to the center of the set. Starting pitchers remain, overall, better pitchers than relievers, but reliever usage helps to reduce the apparent quality gap. FDP-Wins1871-1910, 1500+ IP cohort 86 Pitchers with positive FDP-wins 34 Pitchers with negative BIP-wins 71.7% of cohort positive Overall, in this period 1500+ IP pitchers accrued 964.6 FDP-wins, 4.598 wins/1500 IP. 1871-1940, 1500+ IP cohort193 Pitchers with positive FDP-wins 73 Pitchers with negative FDP-wins 72.6% of cohort positive Overall, in this period 1500+ IP pitchers accrued 1747.7 FDP-wins, 3.906 wins/1500 IP. 1955-2024, 1500+ IP cohort222 Pitchers with positive FDP-wins 4 Pitchers with zero FDP-wins 198 Pitchers with negative FDP-wins 52.4% of cohort positive Overall, in this period 1500+ IP pitchers accrued 338.9 FDP-wins, 0.523 wins/1500 IP Analysis. Nothing all that unusual about the way BIP-wins and LOB-wins combine.
|
|
|
Post by chriscobb on May 18, 2024 16:29:59 GMT -8
PART 2. CORRELATIONS BETWEEN FDP-WINS/1500 IP AND OTHER INDICATORS OF PITCHING EFFECTIVENESS: CAREER IP, CAREER FIP-WAR, AND CAREER FIP-WAR RATE
1871-1910, 1500+ IP cohort
Career IP & FDP-wins/1500 IP Pitchers with 3000+ IP (29 total): 7.710 FDP-wins/1500 IP Pitchers with 1500-1999.7 IP (37 total): 0.916 FDP-wins/1500 IP (Pitchers with 3000+ IP in this cohort accrued 18.280 FIP-WAR/1500 IP) (Pitchers with 1500-1999.7 IP accrued 14.914 FIP-WAR/1500 IP)
Career FIP-WAR & FDP-wins/1500 IP Top 29 pitchers: 6.634 FDP-Wins/1500 IP Bottom 37 pitchers: 0.183 FDP-wins/1500 IP (Pitchers in the top 29 by career FIP-WAR in this cohort accrued 20.178 FIP-WAR/1500 IP) (Pitchers in the bottom 37 by career FIP-WAR in this cohort accrued 13.372 FIP-WAR/1500 IP)
Career FIP-WAR rate & FDP-wins/1500 IP Top 29 pitchers: 5.864 FDP-wins/1500 IP Bottom 37 pitchers: 3.425 FDP-wins/1500 IP (Pitchers in the top 29 by Career FIP-WAR rate in this cohort accrued 22.961 FIP-WAR/1500 IP) (Pitchers in the bottom 37 by Career FIP-WAR rate in this cohort accrued 12.544 FIP-WAR/1500 IP)
Analysis. In this period, a solid correlation exists between FDP-wins and all the other measures. It is notable that the correlation is stronger between FIP-WAR and FIP-WAR rate, suggesting that during this period, pitchers could sustain a long career without a top FIP-WAR rate, if their FDP-wins rate was sufficiently robust. Splitting this cohort between pre- and post-1893 would probably be instructive.
1871-1955, 1500+ IP cohort
Career IP & FDP-wins/1500 IP Pitchers with 3000+ IP (63 total): 5.800 FDP-wins/1500 IP Pitchers with 1500-1999.7 IP (94 total): 1.279 FDP-wins/1500 IP (Pitchers with 3000+ IP in this cohort accrued 19.450 FIP-WAR/1500 IP) (Pitchers with 1500-1999.7 IP accrued 16.143 FIP-WAR/1500 IP)
Career FIP-WAR & FDP-wins/1500 IP Top 63 pitchers: 5.491 FDP-Wins/1500 IP Bottom 94 pitchers: 0.756 FDP-wins/1500 IP (Pitchers in the top 63 by career FIP-WAR in this cohort accrued 21.284 FIP-WAR/1500 IP) (Pitchers in the bottom 94 by career FIP-WAR in this cohort accrued 13.912 FIP-WAR/1500 IP)
Career FIP-WAR rate & FDP-wins/1500 IP Top 63 pitchers: 5.104 FDP-wins/1500 IP Bottom 94 pitchers: 2.750 FDP-wins/1500 IP (Pitchers in the top 63 by Career FIP-WAR rate in this cohort accrued 23.708 FIP-WAR/1500 IP) (Pitchers in the bottom 94 by Career FIP-WAR rate in this cohort accrued 13.263 FIP-WAR/1500 IP)
Analysis: Similar patterns for 1871-1955 as for 1871-1910.
1955-2024, 1500+ IP cohort
Career IP & FDP-wins/1500 IP Pitchers with 3000+ IP (59 total): 1.682 FDP-wins/1500 IP Pitchers with 1500-1999.7 IP (193 total): -0.213 FDP-wins/1500 IP (Pitchers with 3000+ IP in this cohort accrued 24.990 FIP-WAR/1500 IP) (Pitchers with 1500-1999.7 IP accrued 18.260 FIP-WAR/1500 IP)
Career FIP-WAR & FDP-wins/1500 IP Top 59 pitchers: 0.978 FDP-Wins/1500 IP Bottom 193 pitchers: 0.659 FDP-wins/1500 IP (Pitchers in the top 59 by career FIP-WAR in this cohort accrued 28.809 FIP-WAR/1500 IP) (Pitchers in the bottom 193 by career FIP-WAR in this cohort accrued 15.046 FIP-WAR/1500 IP)
Career FIP-WAR rate & FDP-wins/1500 IP Top 59 pitchers: 0.079 FDP-wins/1500 IP Bottom 193 pitchers: 1.034 FDP-wins/1500 IP (Pitchers in the top 59 by Career FIP-WAR rate in this cohort accrued 31.863 FIP-WAR/1500 IP) (Pitchers in the bottom 193 by Career FIP-WAR rate in this cohort accrued 14.703 FIP-WAR/1500 IP)
Analysis: Here the patterns are quite different from the early period. There remains a clear correlation between longer careers and FDP-wins/1500 IP, suggesting that generating FDP-wins remains a pitching skill and that this skill is associated with longer careers, although the FDP-win rates are much lower than for the first 70 years of major-league baseball. The correlation between FIP-WAR and FDP-wins disappears, however, with top and bottom pitchers by career FIP-WAR producing similar rates of FDP-wins, and those rates are only a little bit above average. Most notably, the correlation between FIP-WAR rate and FDP-wins rate has flipped, with pitchers with lower FIP-WAR rates having higher FDP-win rates. This result corroborates the analysis of changes in LOB-wins post-1954 suggesting that power pitching, which is productive of the highest FIP-WAR rates, entails approaches that do not result in FDP-wins. Pitchers with lower FIP-WAR rates can compensate somewhat for reduced effectiveness in FIP by adding effectiveness in FDP. This compensation is relatively small, however.
Inferences I draw from this study
1) FDP-wins and FIP-WAR result in part from the same pitching skills. Good stuff and good control limit walks and produce strikeouts, leading to FIP-WAR. They also lead to weaker contact (good stuff) and to balls in play going where fielders expect them to go (control), which produce BIP-wins. LOB-wins result from pitchers who can produce a particular kind of result (strikeout or double-play ball) when it is called for. Thus, pitchers who are stronger in FIP-WAR tend to be stronger in FDP-wins as well.
2) FDP-wins and FIP-WAR result in part from different pitching strategies, especially with respect to LOB-wins. PitcherW using a “control the running game” or a “pitch to contact’ or a “pitching in a pinch” strategy manage baserunners, invite contact in order to influence its results, and savestheir best stuff for occasional use. This strategy will result in lower FIP-WAR or lower BIP-wins that may be made up by LOB-wins and more innings pitched. The power pitcher who goes as hard as possible at every batter without being distracted by externalities like baserunners may still generate BIP-wins due to weak contact, but they will not achieve sequencing results that produce LOB-wins because they don’t have a “higher gear” to shift into when they are “pitching in a pinch.” In fact, their FIP may overpredict their overall wins because they are generating “excess” strikeouts or pitching in a way that leads to a high “index of self-destruction” as Bill James called it. A high self-destruction index from stolen bases allowed, ill-timed walks, or wild pitches might show up in negative LOB-wins.
3) Employing a pitching strategy that played for FDP-wins was sensible for any pitcher who could learn to do it prior to the 1960s. The rarity of power hitters, the heavy use of counter-productive strategies for baserunner advancement, the expectation that a pitcher should be trying to throw a complete game, and the wide variance in pitching skills and strategic know-how gave significant advantages to the pitchers who applied their skills strategically to generate FDP-wins. The growing prevalence of power hitters, the gradual optimization of baserunning strategies, the growth in the strategic use of relief pitching, and improvements in coaching for pitchers gradually reduced the utility of FDP-wins strategies while increasing the importance of FIP-WAR skills, a trend that was recognized analytically when Voros McCracken introduced DIPS in 1999.
4) Because of the way pitchers’ skills (stuff and control) contribute to FDP-wins as well as FIP-WAR, FIP-WAR should not be relied upon as a direct measure of pitchers’ effectiveness prior to the 1960s, and even after 1960 its meaning should be carefully interpreted. Pitchers’ varied strategic approaches to the use of their skills will lead to varied distribution of their wins among FIP, BIP, and LOB measures. In general, in pre-1940 baseball, pitchers who were generating more FIP-WAR were generating more FDP-wins as well: they resulted from the same skills being employed variably within an overall pitching strategy.
5) The caution in using FIP-WAR for direct comparisons applies to both cross-period comparisons between post-1960 and pre-1940 pitchers and to comparisons between pre-1940 pitchers.
|
|
kcgard2
Hall of Merit Voter
Posts: 98
|
Post by kcgard2 on May 19, 2024 5:05:21 GMT -8
Pitchers with lower FIP-WAR rates can compensate somewhat for reduced effectiveness in FIP by adding effectiveness in FDP. This compensation is relatively small, however. Sounds like the known relationship with heavy groundball pitchers. Maybe you could check if there's a correlation here for the post-1960 pitchers.
|
|
|
Post by chriscobb on May 19, 2024 5:58:57 GMT -8
Pitchers with lower FIP-WAR rates can compensate somewhat for reduced effectiveness in FIP by adding effectiveness in FDP. This compensation is relatively small, however. Sounds like the known relationship with heavy groundball pitchers. Maybe you could check if there's a correlation here for the post-1960 pitchers. I can't readily build on the work that I've done to add in a look at groundball percentage, unfortunately. Fangraphs, which is my source, of course, for FIP-WAR, doesn't have groundball/flyball data for pitchers until 2002. Baseball-Reference gb/fb data picks up earlier, in 1988. I could probably put together a study from 1988 to the present, although it would take some time to fit the two data sources together.
My guess would be that, given the relatively small impact of FDP-wins in recent MLB history, a factor like groundball percentage will have too variable an impact to create a pattern in the overall data set. A groundball approach creates FIP-value by avoiding home runs, while losing FIP-value by allowing BIP. It loses BIP-value by allowing a higher percentage of hits, but gains BIP-value by lowering the percentage of extra-base hits. It gains LOB-value by leading to more double plays, while losing LOB-value by reducing timely strikeouts. With so many offsetting factors, groundball percentage may not actually correlate with lower FIP-wins or higher FDP-wins. If I can take a look at it, I will.
Looking for correlations with K rates and BB rates, on the other hand, would be easy to do and could be done for the entirely of baseball history. Seeing how/if/when FDP-wins correlate with relative production of strikeouts and walks might be interesting, as those are both key components of FIP-value.
|
|