Tag Archives: war

WAR Roster Dump

Today we’ll do a WAR roster dump table like we generated a few days ago here as part1 of our playoff horse race series.  WAA has additive properties and WAR does not which we showed here back in 2013.  Here is the WAR table listing total WAR for hitters, pitchers, starters, relief and total.

TeamID Hitters Pitchers Starters Relief Total W-L
LAN 22.80 10.30 4.20 6.10 33.10 51
WAS 15.20 19.00 17.90 1.10 34.20 30
HOU 27.60 12.20 7.90 4.30 39.80 26
CLE 15.00 19.40 13.50 5.90 34.40 20
BOS 12.60 15.00 9.10 5.90 27.60 19
ARI 12.40 22.80 18.00 4.80 35.20 17
CHN 17.00 9.40 3.80 5.60 26.40 12
COL 12.90 17.80 12.10 5.70 30.70 11
NYA 22.60 15.90 9.80 6.10 38.50 8
MIN 14.70 6.00 4.60 1.40 20.70 6
MIL 16.90 13.70 7.80 5.90 30.60 5
ANA 15.70 8.00 3.90 4.10 23.70 4
BAL 17.50 7.50 0.70 6.80 25.00 3
TEX 17.50 10.50 7.20 3.30 28.00 0
SLN 16.00 10.20 7.90 2.30 26.20 0
MIA 20.40 2.70 1.90 0.80 23.10 0
TBA 18.10 5.80 4.40 1.40 23.90 -1
SEA 16.10 6.20 1.80 4.40 22.30 -2
KCA 12.80 11.20 6.00 5.20 24.00 -2
PIT 8.00 13.30 8.40 4.90 21.30 -8
TOR 6.90 15.50 9.20 6.30 22.40 -11
ATL 8.60 10.40 4.10 6.30 19.00 -13
SDN 5.30 5.30 2.20 3.10 10.60 -15
NYN 0.30 6.80 4.70 2.10 7.10 -16
DET 11.80 9.30 6.20 3.10 21.10 -16
OAK 9.20 3.30 2.60 0.70 12.50 -17
CIN 21.30 1.60 0.10 1.50 22.90 -21
CHA 10.50 -1.00 0.20 -1.20 9.50 -27
SFN 4.50 7.70 4.80 2.90 12.20 -29
PHI 5.20 8.10 4.40 3.70 13.30 -34

Blue highlights are top of the league in that column, red bottom according to WAR.

WAR doesn’t go negative very often and we showed in this post how the sum of all hitters/fielders adds to exactly 600 and pitchers add to 400.  WAR folds subjective fielding math into hitters.  We suspected that WAR treats hitting 40%, pitching 40% and fielding 20%.  This model treats fielding as a separate class.   We measure fielding at a team level using the only reliable and official play by play metric, errors and the unearned runs caused by them.  Scorekeepers have been tabulating errors since the beginning of baseball.

Scanning the WAR Total column you notice there is very little disparity in WAR totals in the top 2/3 of MLB.   WAR is a fan value stat that tries not to upset fans of players not playing well.  Since WAR treats hitting as 60% and pitching 40% the hitting column is very high for all teams.  The relief column is low.  WAR does not handle part time players well and relievers could be considered part time compared to starters.   We have shown in past posts here how relievers have helped both Cleveland and the Cubs be in contention for the playoffs.

The Dodgers have an interesting line.  According to this table they have better hitters than pitchers.  That runs counter to their seasonal team status line.

BAT PITCH Rs Ra W L UR LR TeamID
32.2 159.1 655 439 91 40 20.1 4.6 LAN

Runs not let up due to pitching (PITCH) far exceeds their runs scored for the season (BAT).  The roster this dataset uses have Kershaw and Wood on DL which are WAR=4.4 and WAR=3.1.  If you add 7.5 to LAN’s starter total they don’t have the best pitching staff in MLB which is not true.  They don’t even have the best starting staff.   Their team status clearly shows their +51 is due to pitching.    In order to fold fielding into hitting, WAR takes value away from pitchers and gives it to hitters under the guise its formulae can discern who deserves what with respect to fielding.  They can’t.  We have shown time and time again how massive errors get introduced and this is why.  The Dodgers have the best pitching staff period.  We have Kershaw ranked #3 in MLB both pitchers and batters and that’s after not playing for a bunch of weeks.  WAR has him ranked #26.

And finally, this caught my attention when modifying some scripts to do these WAR tables.

Rank WAR Name_TeamID Pos
-002- -2.0 Albert_Pujols_ANA DH

WAR does not go negative often.  Even Edwin Jackson stayed positive in some version of WAR back in 2014 when he had a terrible season of historic proportions.  How did Pujols of all players get a -2.0 ranked the second worst player according to WAR?  This has to be another big mistake in whatever algorithm used to compute WAR.  Sometimes I think there are people pulling levers to favor certain players and not favor players they simply don’t like.  Perhaps someone is blaming Anaheim’s demise on Pujols.   These WAR numbers came from baseball-reference.  We don’t know how to compute it.

Rank WAA BA OBP PA RBI R Name_TeamID Pos
XXXXX 0.36 0.232 0.278 526 83 47 Albert_Pujols_ANA DH

Above is Pujols’ full line according to this data model.  We have him slightly above average and unranked.  For Pujols this is his worst year.  In no way does he suck this year.  WAR is simply wrong once again.

Edit: For a player making $30 or so million per season a 0.36 is well below what ANA expects for a top tier MLB salary.  Pujols should be at +3 at the lowest for the season.  Pujols is well below what is expected from him but as an anonymous MLB player he’s slightly above average according to his seasonal run production.   Albert Pujols is a first (possibly unanimous) ballot HOFer 5 years after whenever he decides to retire.  I’m starting to think his WAR listed at baseball-reference may be a joke.

WAR vs WAA Pitchers 2013

This post will compare WAR and WAA rankings for pitchers in the 2013 season. The only team analyzed in WAR vs WAA comparisons was CHA (Chicago White Sox) so those pitchers will be excluded from the pertinent stats table.  As mentioned in the previous post, WAR and WAA values are included in the pertinent stats table for informational purposes.  They are very different numbers with much different ranges.  This analysis compares how each system ranks players using their own values.

WAR overrated or WAA underrated?

This posts takes the top 50 ranked WAR players and lists the biggest differences with WAA ranked players. Thus, WAR thinks these pitchers are highly valued, WAA not so much. It is interesting to note that the differences for pitchers aren’t as great as the differences for batters in the previous post.

Diff WAR WAA Name_tm
175 17 192 Chris_Tillman_BAL
167 21 188 Doug_Fister_DET
159 49 208 Jon_Lester_BOS
110 16 126 Cole_Hamels_PHI
110 11 121 Jose_Quintana_CHA
100 19 119 Jorge_De_La_Rosa_COL
WAR WAA IP ERA G W L Name_Tm BAT/PITCH
4.4 0.8 206.3 3.71 33 16 7 Chris_Tillman_BAL PITCH
4.1 0.9 208.7 3.67 33 14 9 Doug_Fister_DET PITCH
3.0 0.6 213.3 3.75 33 15 8 Jon_Lester_BOS PITCH
4.6 1.7 220.0 3.60 33 8 14 Cole_Hamels_PHI PITCH
4.3 1.8 167.7 3.49 30 16 6 Jorge_De_La_Rosa_COL PITCH

WAR underrated or WAA overrated?

The next table shows the biggest differences with the highly ranked WAA players (less than 50) and WAR.  It is interesting that with both pitchers and batters in the previous post, the differences between WAR and WAA are much less with highly ranked WAA players.  In other words, WAR agrees more with WAA when WAA is ranked higher than WAA agrees with highly ranked WAR.

Diff WAR WAA Name_tm
71 88 17 Mark_Melancon_PIT
69 117 48 Craig_Breslow_BOS
67 110 43 Justin_Wilson_PIT
58 98 40 Kevin_Siegrist_SLN
57 91 34 David_Carpenter_ATL
57 90 33 Luke_Hochevar_KCA
WAR WAA IP ERA G W L Name_Tm BAT/PITCH
2.0 4.2 71.0 1.39 72 3 2 Mark_Melancon_PIT PITCH
1.6 3.0 59.7 1.81 61 5 2 Craig_Breslow_BOS PITCH
1.7 3.1 73.7 2.08 58 6 1 Justin_Wilson_PIT PITCH
1.9 3.3 39.7 0.45 45 3 1 Kevin_Siegrist_SLN PITCH
2.0 3.4 65.7 1.78 56 1 4 David_Carpenter_ATL PITCH
2.0 3.4 70.3 1.92 58 5 2 Luke_Hochevar_KCA PITCH

WAR vs WAA Batters 2013

In past posts we highlighted certain teams that were either under or over rated according to WAR or WAA. This post will highlight the largest differences between WAR and WAA with respect to batting. The tables below will pick the top 5 differences of under and over rated players not mentioned in previous posts. Thus, SLN, DET, PIT, and LAN players will be shown in the rankings but not in the pertinent stats table.

WAR underrate or WAA overrate?

Unlike in previous posts on this topic only players with greater than 100 PAs were included in the rank.  This cleans up a lot of noise around zero for WAA.  Since all players start at WAA=0 a player with a couple of plate appearances will rank higher than a player who goes into negative territory.  This skew is confusing.  Eliminating short time batters, including most all pitchers, cleans up the ranking numbers.

This sort is based on WAA so only WAA ranked players less than 50 were considered.  The next section, will sort based on WAR ranking.  The table following each ranking table will list the player’s pertinent stats including their WAR and WAA values.  The actual WAR and WAA are included for informational purposes only as there is no direct comparison between the two numbers.  We are concerned with how each system ranks players with each other.  That is the only way to compare the two systems.

Diff WAR WAA Name_tm
144 164 20 Brandon_Phillips_CIN
131 157 26 Prince_Fielder_DET
116 139 23 Michael_Cuddyer_COL
115 158 43 Yoenis_Cespedes_OAK
115 125 10 Brandon_Moss_OAK
109 126 17 Allen_Craig_SLN
106 128 22 Mark_Trumbo_ANA
WAR WAA BA OBP PA RBI R Name_Tm BAT/PITCH
1.6 4.8 0.261 0.310 666 103 80 Brandon_Phillips_CIN BAT
2.0 4.7 0.331 0.389 540 84 74 Michael_Cuddyer_COL BAT
1.7 3.5 0.240 0.294 574 80 74 Yoenis_Cespedes_OAK BAT
2.2 5.8 0.256 0.337 505 87 73 Brandon_Moss_OAK BAT
2.2 4.7 0.234 0.294 678 100 85 Mark_Trumbo_ANA BAT

WAR overrate or WAA underrate?

This table sorts the top 5 largest differences of less than 50 ranked WAR players.  Either WAR overrated these players or WAA underrated them.  You make the call.

Welington Castillo is a catcher for CHN.  Since WAR folds defense into their number it’s quite possible that led to their high rating for Wellington.  The catcher is the most important defensive asset in fielding so perhaps they have a point.  This model treats defense as a separate class with its own rating.

Diff WAR WAA Name_tm
431 37 468 Welington_Castillo_CHN
379 16 395 Gerardo_Parra_ARI
311 8 319 Andrelton_Simmons_ATL
263 24 287 Starling_Marte_PIT
224 44 268 Russell_Martin_PIT
186 29 215 Ben_Zobrist_TBA
185 25 210 Joe_Mauer_MIN
WAR WAA BA OBP PA RBI R Name_Tm BAT/PITCH
4.4 -2.2 0.274 0.349 428 32 41 Welington_Castillo_CHN BAT
6.1 -1.2 0.268 0.323 663 48 79 Gerardo_Parra_ARI BAT
6.8 -0.7 0.248 0.296 658 59 76 Andrelton_Simmons_ATL BAT
5.1 0.1 0.275 0.355 693 71 77 Ben_Zobrist_TBA BAT
5.4 0.1 0.324 0.404 508 47 62 Joe_Mauer_MIN BAT

Update 2/8/2014

I had noticed an anomaly with Joe Mauer’s stats in the last table above.  How did a player batting 0.324 with a 0.404 OBP end up in the top either overrated WAR or underrated WAA list?  Isn’t this a clear case of WAA underrating a player?

All tables in this model get spit out automatically based upon certain guidelines I define.  It  surprises me when something doesn’t seem right so I  trace back and check to make sure there isn’t a software bug somewhere (that happens).  First I check RISP tables.  RISP tables are described here.  We have shown in this post that 3/4 of all runs occur in Runners In Scoring Position situations.  The fundamental difference between the WAR and WAA systems is WAR focuses on hits and walks whereas this model focuses on run production.  Below is Joe Mauer’s RISP record.

RAA BA OBP PA Name_team
0.9 0.272 0.405 126 Joe_Mauer_MIN

Although Mauer maintained his OBP in RISP situations, his BA is 50 points lower.  He still is above league average with a 0.9 runs above average with runners in scoring position.  A batter brings value to his team by knocking in runs and/or scampering around the bases so others can hit him in easier.  Mauer’s WAA=0.1 places him slightly above an average MLB player which is appropriate here.  Had Mauer hit 0.324 in RISP situations he would have produced a lot more wins for his team and scored a higher WAA value.  In other words; the data here indicates many of Mauer’s hits came in situations that didn’t matter. The WAR math does not have the capability to discern this.

In the end, baseball rules dictate that the team with the most runs wins a ball game — not the team with the most hits and walks.

WAR Pitching Evaluations for CHA

The differences aren’t as striking with regard to pitching evaluations between the two systems. WAR ranked the Chicago White Sox the #2 staff in the league even though they gave up 49 runs above league average. This should have made them a below average pitching staff — not top tier as WAR numbers suggest. The White Sox gave up almost 30 unearned runs above the league average which means they had bad fielding. Perhaps this skewed WAR rankings across the board for the White Sox.

I did not find any large ranking discrepancies in the top 100 ranked WAR and WAA pitchers for CHA. There were only two and they are shown in the tables below.

Chicago White Sox

Diff WAR WAA Name_tm
117 56 173 Hector_Santiago_CHA_PITCH
110 11 121 Jose_Quintana_CHA_PITCH
WAR WAA IP ERA G W L Name_Tm BAT/PITCH
2.8 1.1 149.0 3.56 34 4 9 Hector_Santiago_CHA PITCH
5.4 1.7 200.0 3.51 33 9 7 Jose_Quintana_CHA PITCH

I suspect the large amount of unearned runs CHA gave up caused an across the board increase in WAR pitching evaluations for all CHA pitchers that incrementally caused the anomaly giving them such a high team pitching rating.

WAR Batting Evaluations for PIT and LAN

As shown in this previous post WAR ranked LAN and PIT #2 and #3 in the league for batting even though both scored quite a few runs less than league average.  This is quite an anomaly.  This post will examine WAR player batting evaluations when WAR over rates a team.

This time we took the top 100 ranked players from the WAR and WAA systems and sort players based upon the biggest ranking differences.  Although we can’t directly compare the numbers for the two systems we can certainly compare their ranking.  We want to see the most egregious differences in the below tables.

Los Angeles Dodgers

Zack Greinke is highlight below because WAR rated him +1.3 for his batting as a pitcher.  His 0.328 batting average is quite impressive but he only had 72 plate appearances.   He didn’t generate enough data to indicate that 0.328 batting average isn’t anything more than a statistical anomaly.   WAR rates pitcher Zack Greinke higher for his batting than Matt Holliday and Matt Adams mentioned in the previous postGreinke is ranked #88 amongst all batters in MLB.

Like in the last post, the first table in the below sets lists the ranking differences between WAR and WAA sorted by the highest rank difference.  The next table lists their WAR,  WAA, and other pertinent stats.

Update: The ranking for Zack Geinke is in error and probably much lower due to a bug in the script. 

Update 2/3/2014 Rankings below have been updated.  They didn’t change much but Greinke went from 88th to 184th.  This is still quite a high ranking for a pitcher that only had 74 plate appearances so the original commentary still stands.

Diff WAR WAA Name_tm
1368 94 121 1462 Nick_Punto_LAN_BAT
1327 98 90 1425 Andre_Ethier_LAN_BAT
1265 76 83 1341 Mark_Ellis_LAN_BAT
1178 88 184 1266 Zack_Greinke_LAN_BAT
122 39 50 161 Juan_Uribe_LAN_BAT

Note: The heading for these tables are overloaded which means columns represent different types depending upon PITCHer or BATter.

Update WAR numbers in below player tables were all corrected on 1/30/2014

WAR WAA BA/IP OBP/ERA PA/G RBI/W R/L Name_Tm BAT/PITCH
2.2 -1.8 0.255 0.328 335 21 34 Nick_Punto_LAN BAT
2.7 -1.4 0.272 0.360 553 52 54 Andre_Ethier_LAN BAT
3.0 -1.0 0.270 0.323 480 48 46 Mark_Ellis_LAN BAT
1.3 -0.7 0.328 0.409 72 4 5 Zack_Greinke_LAN BAT
4.1 0.6 0.278 0.331 426 50 47 Juan_Uribe_LAN BAT

Pittsburgh Pirates

Note how WAR ranked Starling Marte #25  24 out of all MLB batters yet he produced only 35 RBIs.

Update 2/3/2014Rankings below have been updated to reflect WAR.  In this case the rankings didn’t change that much which is expected since baseball-reference WAA is highly correlated with their WAR number. 

Diff WAR WAA Name_tm
1137 25 24 1162 Starling_Marte_PIT_BAT
1090 53 59 1143 Neil_Walker_PIT_BAT
1065 40 44 1105 Russell_Martin_PIT_BAT
WAR WAA BA/IP OBP/ERA PA/G RBI/W R/L Name_Tm BAT/PITCH
5.4 -0.4 0.280 0.343 566 35 83 Starling_Marte_PIT BAT
3.9 -0.4 0.251 0.339 551 53 62 Neil_Walker_PIT BAT
4.3 -0.3 0.226 0.327 506 55 51 Russell_Martin_PIT BAT

Under the WAR rating system LAN and PIT won based on their hitting and not their pitching. They were #2 and #3 in MLB for hitting posted here. LAN and PIT ranked #11 and #18 in MLB for pitching.  The next post will explore how WAR ranked pitching at a player level that led to  this evaluation of LAN and PIT.