Mathematical Discussion: The Fog of WAR and Understanding the Components of WAR

Think back to the close MVP race 20 years ago. I’ll go with 2007, because that was the first year I played good baseball, which means it was also the first year I followed baseball statistics closely. The National League MVP debate comes down to Matt Holliday and Jimmy Rollins. Matt Holliday was coming off an offensive season, hitting .340/405/.607 with the Rockies while hitting 34 HR and a league-leading 137 RBI on a Rockies team that made the postseason after taking down the Padres in Game 163 (no comments on home plate touching). Jimmy Rollins, on the other hand, played in all 162 NL East winning games for the Phillies, joining the “four 20 club” (not the 420 club, which I belong to), hitting +20 homers, +20 doubles, +20 triples, and stealing +20 bases. It’s actually even more impressive because he hit 30 homers, 38 doubles, 20 triples, and stole 48 bases, so it was a 30/30/20/40 season, Rollins 2007 being the only record. In the end, Jimmy Rollins received 16 of the 30 first-place votes (Holliday received 11) and became the fourth Phillies player ever to win MVP, and interestingly, we haven’t seen a short MVP win since.

This article is not a history lesson. The thing to focus on is the *thought process* that went into deciding who deserved to be the MVP between the two. Remember this was 2007; Before the iPhone, emo music was all the rage, Fangraphs wasn’t a popular website, Razzball didn’t really exist, and most importantly, WAR wasn’t part of baseball’s common parlance. People would look at the statistical leaders in each division, and use their big brains to think “well… Rollins hit 30 homers and Holliday hit 34 HR… but Rollins is a shortstop so I’ll give him the magic bean… but Holliday leads the league with a .340 average… A stat that was based almost entirely on vibes, and when pressed, we probably couldn’t even put it on paper. endless debates in bars all over the country, finally, the geniuses were able to calculate how many magic beans to add to be a shortstop, and how many magic beans to take to the Colorado bar boy.

A lot has changed since 2007, everyone has iPhones, emo music is dead (but never gone), Fangraphs and Razzball are celebrities in the online baseball world, and WAR rules the MVP voting. Many debates that have existed over the years have been resolved even before the end of the season, because WAR shows us who should be the finalists, and if there is any doubt about who will win the award. Keystone baseball statistics, how to compare players from different positions, different home ballparks, and even different eras. WAR isn’t just about award voting, it controls the conversation about trades, free agent signings, and ultimately who gets into the Hall of Fame.

With so much influence and control over the entire baseball world, one would think that the average baseball fan would have a very detailed sense of how their favorite player’s WAR is made. There’s nothing like sitting down with the guys and saying “I think Bobby Witt was good for the Royals as he led the team with 1.1 WAR, *even*, 0.6 of that came from his hitting”. Nobody talks like that, and it’s safe to assume that, reading this, you don’t know that most of Shohei Ohtani’s NL-leading 1.5 WAR is mostly WAR or hitting WAR. The nebulous nature of the WAR statistic often feeds skepticism about its value as a metric. People will say, “I don’t care what WAR is, nobody knows how it’s calculated, blah blah blah.” The thing is, this is not true! At least in Fangraphs WAR calculations, components are listed under a player’s “Value” table and are shown as Run components.

This table is crazy if you can read more runs

The reason this table isn’t useful to most people is that it shows WAR components like “Runs” that need to be converted to “Wins” (usually around ~10 runs per win). Once you convert parts, you can see that WAR is simply the total value of the part you get from it:

• Change = This is simply the value you bring to play
• League = The extra value you bring to the league you are in
• Position = The amount you add (or lose) based on the position you play
• Base Running = Value from being a good/bad keeper
  • This is not just stealing but taking extra bases on singles, scoring from first doubles, all the fun stuff
• Fielding = This used to be based on Defensive Runs Saved, but not based on Statcast’s Outs Over Average
• Hitting = This is the most common hitting you see, adjusted for the ballpark and the hitting location

Taking these parts and drawing them creates this:

This graph is from A WAR partition tool I created. It will allow you to draw – of any player and season since 2002 – a breakdown of a player’s WAR per segment. In this example of Bobby Witt Jr, he was a hitter wise position player, and all of his value came from driving (he was a great player), some from running, and some from positional (as he is a shortstop). You might see this graph and think, “Wait, you’re getting a third of the wins from just playing? Does that mean you need to be *worse* than replacing to have a bad WAR?” and the answer is yes. To illustrate this, take a look at Luisangel Acuna’s breakdown:

He gains a fifth of MPI just by playing, but then loses all of that and more by having a .422 OPS, then loses even more by being a poor center fielder (although he does play centerfield in the minors), and then gets to go back to being an elite runner and playing the “premium” position (center field) with a WAR total of -0. The purpose of this is to make the WAR number more important by understanding how it relates to the total.

Naturally, once I had the classification tool set up, my next thought was, I can pull this for every player at once and show the leaderboard, and that’s what you can find in the leaderboard tab. Here is a breakdown of the current Top 15 players by position player WAR:

The legend below shows you how the value adds up for each player, but note that this is not the same as a traditional stacked bar chart, where higher “bars” represent higher numbers. James Wood is in the middle of the chart, although his bar is about the same as Ben Rice, who is in third place, and that is because Wood has those negative numbers at the bottom of his bar. You lose -0.2 WAR for being an infielder, and an additional -0.1 WAR for being a left fielder. Likewise, Yordan and Ohtani both lost the position WAR to being a full-time DH. Nico Hoerner is #5 in baseball even though his blue bar for hitting is much lower than those around him, and that’s because he basically gets direct value from every angle, not just hitting well but being a good hitter, base runner, and fielder.

The third “Compare” tab in the tool allows you to set this classification aside. Here’s a comparison of two players on the team last season, Juan Soto and Francisco Lindor, where Lindor had the higher WAR even though Soto had an .878 OPS and Lindor had an .811 OPS:

This brings us back to the 2007 NL MVP debate. Instead of counting magic beans as we did then, we can see a precise calculation of the value they brought:

It turns out that even with his demotion at Coors, and Rollins’ premium playing shortstop, Holliday still slightly leads Rollins in total WAR. So did the voters almost get this right? No! Just look at how far they both trailed each other in the NL WAR leaders!