Before introducing the top Cubs tomorrow I wanted to show the current top ten players according to WAR. After perusing them there aren’t many large differences yet. There will be as more data accumulates and the error in WAR becomes more profuse. At this stage of the season the error in WAR can more easily be seen from the team level.
One of the columns on the team lines is RAA (Runs Above Average). This is the exact concept used here yet their numbers are different because they estimate runs. This model counts the runs. Somehow RAA translates into a total team WAR for batters. I chose RAA to analyze because there is a direct comparison to the team status lines described here and here.
Here is the team status line for the Cubs up to and including 5/2/2017 games.
The BAT column is Batter runs above average which is runs. The MLB team average for runs scored is 117. The Cubs’ Runs scored against is exactly average. Since UR is not credited to pitchers it must be subtracted out and that’s how the Cubs’ PITCH=+5.9. That’s how team runs above average is calculated using real runs a team produces. That is 100% accurate.
The first thing to look at in the RAA column of this table is the total for all teams. It should add to zero if this is a true runs above average calculation. Right now it adds to +25. Last season it ended at around +100. This is across 30 teams. Getting the books to balance was my biggest challenge. This model’s RAAs add to 0 exactly and if it doesn’t there is corruption in the dataset being calculated.
So we know their total error for all teams. That doesn’t show the swings in error for each individual team. Navigate that table down to the CHC row and find the RAA column and you’ll see BAT(Team RAA)=+21. This is simply wrong. The Cubs’ runs scored – league average of 117 puts their BAT(Team RAA) = 15. The above status line shows 13.8 but let’s forget about Lucky Runs (LR) for now. The error in their calculation for just the Cubs is 6.
This table shows value for pitchers. The RAA for all teams add to 0 so they got that right. They have the Cubs pitching RAA at -10 and this model has them at +5.9 if you leave out unearned runs (which you should). That’s almost a 16 run error. The +5.9 represents real runs. Sorting by WAR puts the Cubs in the bottom tier of MLB for pitching. This model puts them mid tier. How runs get divvied up amongst players will be explored later. The team status lines presented here are factual data, not made up.
RAA is a component for calculating WAR. What can be seen as blatant errors like its evaluation of PIT in 2013 translates down to the player level. Players that do well under this system may do well for their Draft Kings team, but often they aren’t playing so well for their real team.
Tomorrow we’ll do the Cubs and perhaps more articles like this in between. I’m also working on a way to search this database. But that is for later. Until then….