Broken Bat - Discussion Forum: Calculating WAR

Forum >> Discussions >> Calculating WAR

Post ID	Date & Time	Game Date	Function

#36028	06/23/2016 10:37:36 am
Seca Joined: 05/05/2014 Posts: 5199 Waterloo Dinosaurs Legends	Has anyone tried to calculate WAR for their BrokenBat players? I'm taking a shot at it, and am open to feedback / suggestions. (Note: I'm trying to follow fWAR). WAR = BattingR + FieldingR + Base-runningR + Positional Adjustment + League Adjustment + Replacement Runs ) / Runs per win Batting runs is ok. I think I have a decent value for wRAA. I didn't bother trying to add home park factors or league factors (effect of these would be pretty small in BB ... I think). Fielding runs uses UZR. I think I could come up with decent values for this. But I was planning on basing them on player skills rather than defensive stats. It would be hugely subjective. Base running runs tho is a major problem. wSB can be done, but there is absolutely no way to figure out UBR. I did see an alternate formula for WAR that put some stolen base data into the wOBA calculation, and then did not use UBR in the WAR calculation. It was from a much less credible source, but I think it would be pretty doable with the data that is available.

#36055	06/24/2016 8:22:56 am
Meccanodonte Joined: 04/21/2014 Posts: 370 Inactive	Good luck, Seca. I tried something, but my not-so-good sabermetric's knowledge led me nowhere. I hope you will succed.

#36057	06/24/2016 9:41:51 am
Seca Joined: 05/05/2014 Posts: 5199 Waterloo Dinosaurs Legends	Hehe. Well I took a shot at using player defensive skills to represent UZR (in Excel). I started with an "average" skill set at each position. (Eg., 2B was 14F-14R-7A). This defines 0 UZR for the position. Next I got excel to find the difference between the player's skill and this average. I tapered the difference if it was really big or really small. (Ie., for our 2B going from 8->9 arm is more valuable than going from 16->17 arm). I took these skill differences and multiplied each by a scale factor representing how I judged the relative importance of the skills (ie., 2B was scaled 1.5 F : 2 R : .5 A). I summed the weighted differences, and applied a final scale factor to bring the sum in line with the UZR scale (so players I thought were really good at a position were scoring ~+15, and players I thought were awful ~-15). I'm fairly happy with the results, but its super subjective. Someone else may have very different looking average players or weight factors. The next problem is base running runs. Weighted stolen bases is pretty easy to calculate and I think quite valuable (due to the opaqueness of base stealing ability in BB). Ultimate base running is impossible, but I'm thinking a player's speed skill (scaled appropriately) could replace it. This assumes things like tagging up are not influenced by some hidden player trait.

#36060	06/24/2016 11:46:28 am
el_jefe_loco Joined: 02/25/2015 Posts: 73 Inactive	Seca, what do you use for BattingR? I've played around with weighting player ratings to show a WAR-like "overall" player rating, but not sophisticated enough to fully mimic real sabermetrics

#36110	06/25/2016 8:28:41 pm
Dcmrulz Joined: 02/14/2013 Posts: 588 Inactive	I could try calculating for my hitters, but a lot of the weights will be dependent on league stats, so my calcs would only work for II.1. I'm going to use Fangraphs' Guts and use their weights. EDIT: Did my hitters. used Fangraphs to get my formulas, but make sure to adjust for games played, we play 1,920 games in a season per league, so the same rate that gives 1000 WAR per 2,430 game MLB season would match about 790.1235 WAR per 1,920 Broken Bat games in a season. (Split would be 450.3704 hitters and 339.7531 for hitters) Also used the plain ol position ratings (weighted by how much time they played it), so Defensive runs are not factored in. I also adjusted all OFs to be the -7.5 positional weight used by LF/RF. I don't know any noticeable difference a CF in this game is relative to just shoving him to LF/RF, so I assumed all OF are equal. Anywho, here are the WAR for my hitters: 1. 2B Tyler Rodgers: 10.73 WAR! 2. SS Dean Delaney: 4.85 WAR 3. C Mario Ball: 4.63 WAR 4. OF Hiroshi Saitou: 4.61 WAR 5. 1B Justin Coffey: 3.52 WAR 6. OF Larry Riley: 3.36 WAR 7. 3B Tomas Carreras: 2.74 WAR 8. OF Pablo Morales: 1.96 WAR 9. OF Brandon French: 1.67 WAR 10. C Ricardo Cardenal: 1.633 WAR 11. 2B Dale Hensley: 0.765 WAR 12. OF Lester Lloyd: 0.68 WAR 13. SS Victor Ochoa: 0.19 WAR 14. 1B Martin Campos: 0.15 WAR 15. OF Jeff Davidson: 0.00 WAR 16. 1B Jose Rosario: -0.29 WAR 17. 2B Manuel Velez: -0.36 WAR 18. OF Harry Pagano: -0.50 WAR Hitting was solid this year, but having a guy like T-Rod made a huge difference. The Good: T-Rod. Delaney/Ball/Saitou were very good. Special note to Delaney and Ball, as I didn't have as high expectations for either (Delaney had an underwhelming 2026, and Ball was a FA pickup before the season.) 7 hitters above what would be an MLB average, for a team more known for its pitching, maybe that will change. Cardenal had a breakout season (1.05 WAR in 224 PA), giving me a tough decision on whether I should let Ball go and giving Cardenal the gig full-time (Orta may be ML ready next season). It will be interesting, but I guess there's too many unknowns to know just yet. The Bad: Lester Lloyd. 415 Plate appearances. Under 1 WAR. He is supposed to be my OF of the future, but man he is not good. Carreras: 3rd most PA on the Catfish (with 638), 7th in WAR (2.74). It's above average, but Saitou and Rodgers (the two with more PA) turned their campaigns into an All-Star and MVP performance. "Above average" by comparison is kind of a letdown. Too Soon to Tell: Jeff Davidson: He was able to get a positive WAR despite only 11 appearances at the plate, so that's encouraging. Trend: UP Jose Rosario: The other rookie of my team: he had a little more time, and trended down. But that's what happens when you only manage 5 hits in 33 appearances. Looking at the SI, I don't think this guy is good enough to replace Coffey as the 1B of the future. Trend: DOWN Harry Pagano: Dead last in WAR. 86 appearances in 54 games. I'm going to err on the lack of playing time as a why, but that doesn't look good. Trend: DOWN That's about it for the Catfish hitters. I'll do Pitchers tomorrow, probably. Updated Saturday, June 25 2016 @ 10:36:57 pm PDT