Projected standings
It’s been a blast making this projection system and seeing it progress, seeing how changes I make affect different players. I’m going to have even more fun seeing how what I predicted compares to other systems.
I wanted to kind of take this a step further though. I wanted to see if I could project teams win/loss records based on the predicted WAR. I remembered reading that a good estimation would be to use a teams total WAR and add 47.8, giving you wins. Using that equation would help me get to the end result.
To do that though, since my predictions had 2449 players, which comes out to 61 different 40 man rosters, I would likely need to narrow that down some. What I chose to do is use FanGraphs predicted 26-man rosters. I didn’t go with 40 man rosters because most likely the remaining 14 players for each roster would make negligible differences in the finals. I used that dataset and merged it with my projection dataset, that then entered my predicted WAR, which in turn allowed me to get full team WAR for each team. The results looked something like this:
Team Predicted_WAR
COL 18.2
PIT 19.8
BAL 22.9
CLE 23.2
MIA 24.7
KCR 27.3
CHC 26.9
OAK 27.4
CIN 29.6
ARI 29.8
MIN 30.6
DET 31.8
WSN 32.8
BOS 32.4
TOR 32.2
TEX 32.9
TBR 34.8
SEA 34.3
SDP 34.3
HOU 35.5
CHW 37.4
NYY 38.1
MIL 39.7
ATL 42.8
PHI 41.9
STL 43.5
LAA 47.2
NYM 53.8
SFG 54.7
LAD 58.9From there I was able to add together the 47.8, to give me the following win/loss expectations:
Team Wins Losses
COL 66 96
PIT 67 95
BAL 71 91
CLE 71 91
MIA 72 90
KCR 75 87
CHC 75 87
OAK 75 87
CIN 77 85
ARI 77 85
MIN 78 84
DET 79 83
WSN 80 82
BOS 80 82
TOR 80 82
TEX 81 81
TBR 82 80
SEA 82 80
SDP 82 80
HOU 83 79
CHW 85 77
NYY 86 76
MIL 87 75
ATL 90 72
PHI 90 72
STL 91 71
LAA 95 67
NYM 101 61
SFG 102 60
LAD 107 55Success! It actually looked really good. I really liked it. But adding it together, instead of a .500 record we would see more wins than losses. But also this is assuming that all teams face all other teams. I think it should weigh heavier for a team that might have an easier schedule to benefit from that.
I created a matrix for all teams that would show a percent chance of winning a game based on teams WAR. Taking Baltimore (22.9 WAR) and the Yankees (38.1 WAR) as my examples. If I wanted to figure out how often Baltimore might win, I would do 22.9/(22.9+38.1) which would equal .3754 or 37.54% of the time. So I could reasonable say that for every 3 meetings between these two teams, Baltimore could expect to walk away with 1 win.
Next I did the same matrix, but this time I included the number of matchups for each set of teams. Once that was completed, I could multiple the expected chance of winning by the number of matchups to get wins. Using the same example as above. I would take .3754*19 = 7.13 or 7 wins.
After getting that all set, I was ready to create my new standings:
Team Wins Losses
COL 53 109
PIT 59 103
BAL 66 96
CLE 67 95
MIA 65 97
KCR 74 88
CHC 72 90
OAK 72 90
CIN 77 85
ARI 75 87
MIN 79 83
DET 82 80
WSN 78 84
BOS 82 80
TOR 81 81
TEX 81 81
TBR 85 77
SEA 82 80
SDP 81 81
HOU 84 78
CHW 89 73
NYY 89 73
MIL 89 73
ATL 89 73
PHI 88 74
STL 93 69
LAA 96 66
NYM 99 63
SFG 101 61
LAD 103 59Added all together is a .500 record. Something that I noticed rather quickly is the AL Central and how much that changed. With Cleveland and Kansas being lower on the list, they made room for the better teams to improve their records. But the same with the Dodgers and the Giants in the NL West. They have a tougher division with the Padres, so their win totals dropped (moreso for the Dodgers than the Giants, however).
So with that, my predicted playoff picture for the AL:
Divison Winners:Team Wins Losses
LAA 96 66
NYY 89 73
CHW 89 73Wildcard: Team Wins Losses GB
TBR 85 77 0.0
HOU 84 78 0.0
SEA 82 80 2.0
DET 82 80 2.0
BOS 82 80 2.0
TOR 81 81 3.0
TEX 81 81 3.0
MIN 79 83 5.0
KCR 74 88 10.0
OAK 72 90 12.0
CLE 67 95 17.0
BAL 66 96 18.0
And the NL:
Divison Winners:Team Wins Losses
LAD 103 59
STL 93 69
NYM 99 63Wildcard:Team Wins Losses GB
SFG 101 61 0.0
MIL 89 73 0.0
ATL 89 73 0.0
PHI 88 74 1.0
SDP 81 81 7.0
WSN 78 84 11.0
CIN 77 85 12.0
ARI 75 87 14.0
CHC 72 90 17.0
MIA 65 97 24.0
PIT 59 103 30.0
COL 53 109 36.0
As a quick disclaimer — this is data for the 26-man rosters as of 3/21/2022. So there is a lot that can change, players will move, data will be updated. And that’s fun that I’m looking forward to.
