March 29, 2021

2021 MLB Season Projections


I'll leave the detailed commentary to the professionals, but with opening day just around the corner, here's how my model (henceforth known as Satchel) has the 2021 MLB season playing out.1 For an overview of how the model works, check out my earlier post. I'll also talk about how the model has changed since I first introduced it at the end of this post.

Model Results

Wins Box Plot

AL West

Team Mean Wins Mean Losses Max Wins Min Wins Win Division (%) Make Wild Card (%) Make Playoffs (%) Win League (%) Win WS (%)
HOU 93.5562 68.4438 133 45 49.67 15.42 65.09 14.19 6.90
LAA 87.5138 74.4862 130 38 27.96 17.21 45.17 7.87 3.69
OAK 81.8608 80.1392 126 39 16.79 12.29 29.08 4.28 1.56
SEA 71.6376 90.3624 128 27 4.56 4.70 9.26 1.04 0.41
TEX 62.1421 99.8579 112 20 1.02 1.19 2.21 0.16 0.04


AL Central

Team Mean Wins Mean Losses Max Wins Min Wins Win Division (%) Make Wild Card (%) Make Playoffs (%) Win League (%) Win WS (%)
MIN 88.9595 73.0405 134 35 36.95 14.29 51.24 9.46 4.36
CHW 88.7453 73.2547 131 41 34.61 14.88 49.49 8.53 3.89
CLE 82.7298 79.2702 126 36 20.05 11.65 31.70 4.59 1.88
KCR 71.7534 90.2466 121 24 5.37 4.44 9.81 0.99 0.37
DET 67.3900 94.6100 114 25 3.02 2.27 5.29 0.52 0.22


AL East

Team Mean Wins Mean Losses Max Wins Min Wins Win Division (%) Make Wild Card (%) Make Playoffs (%) Win League (%) Win WS (%)
NYY 97.2213 64.7787 138 51 44.26 28.18 72.44 20.53 11.53
TOR 94.0163 67.9837 135 51 32.19 31.42 63.61 16.31 8.70
TBR 88.4970 73.5030 135 47 18.56 28.09 46.65 8.91 4.40
BOS 76.3196 85.6804 126 27 4.16 11.40 15.56 2.21 0.85
BAL 65.0224 96.9776 111 26 0.83 2.57 3.40 0.41 0.17


NL West

Team Mean Wins Mean Losses Max Wins Min Wins Win Division (%) Make Wild Card (%) Make Playoffs (%) Win League (%) Win WS (%)
LAD 106.9540 55.0460 143 62 69.81 22.50 92.31 35.28 21.61
SDP 95.7330 66.2670 136 51 26.29 44.79 71.08 14.35 7.60
ARI 75.6193 86.3807 118 36 2.35 12.69 15.04 1.41 0.67
SFG 72.3000 89.7000 118 31 1.40 7.88 9.28 0.95 0.42
COL 59.9863 102.0137 107 19 0.15 1.24 1.39 0.11 0.03


NL Central

Team Mean Wins Mean Losses Max Wins Min Wins Win Division (%) Make Wild Card (%) Make Playoffs (%) Win League (%) Win WS (%)
MIL 87.0170 74.9830 127 38 38.54 10.64 49.18 7.44 3.00
STL 83.7571 78.2429 128 37 28.86 10.25 39.11 5.36 2.16
CHC 78.8213 83.1787 128 34 17.08 7.45 24.53 3.04 1.25
CIN 75.2999 86.7001 133 31 11.47 5.98 17.45 1.91 0.62
PIT 67.6113 94.3887 117 26 4.05 2.64 6.69 0.60 0.18


NL East

Team Mean Wins Mean Losses Max Wins Min Wins Win Division (%) Make Wild Card (%) Make Playoffs (%) Win League (%) Win WS (%)
NYM 89.6429 72.3571 132 44 32.85 20.08 52.93 10.04 4.82
ATL 88.3486 73.6514 132 38 29.14 19.19 48.33 8.87 4.17
WSN 85.8509 76.1491 134 41 22.87 18.72 41.59 6.61 2.95
PHI 80.8530 81.1470 127 31 13.86 13.52 27.38 3.80 1.52
MIA 64.8403 97.1597 109 25 1.28 2.43 3.71 0.23 0.03


Most divisions project to be relatively close races. Although the Yankees have a three game edge over Toronto in the AL East, it wouldn't take much for Toronto to jump ahead. The NL Central is in a similar situation with the Brewers holding a slight lead on the Cardinals. The AL Central and NL East are both effectively toss up, though I have yet to add an "LOLMets" factor to the model so I suspect we may be underestimating Atlanta's chances of ending the season on top of the East.2

Neither the AL nor NL West look particularly competitive as Houston projects to bounce back from a subpar 2020 regular season and the Dodgers once again run away with it in the NL. This is unfortunate for the Padres as their projected 96 wins would be enough to land at the top of any other division except the AL East. Two factors explain why Satchel projects the Dodgers to be so dominant this season: first, based on our measure of talent, the Dodgers are head and shoulders above every other team. Second, they have the weakest schedule in baseball.3 The figure and table below help visualize these points.

Talent Figure

Team Schedule Strength
1 BAL 0.088691
2 BOS 0.062504
3 MIA 0.057176
4 COL 0.039973
5 TEX 0.025384
6 TBR 0.017969
7 TOR 0.015707
8 SEA 0.007129
9 NYY 0.006288
10 PHI 0.005476
11 SFG 0.004386
12 DET 0.004100
13 KCR -0.002096
14 ARI -0.004922
15 WSN -0.006177
16 NYM -0.011025
17 ATL -0.011275
18 LAA -0.017886
19 OAK -0.018126
20 PIT -0.028874
21 CLE -0.031035
22 MIN -0.044572
23 CIN -0.045486
24 CHC -0.045514
25 HOU -0.047139
26 CHW -0.049937
27 SDP -0.051220
28 STL -0.071507
29 MIL -0.076807
30 LAD -0.085321


Given how the model works, it isn't surprising then that the Dodgers are the World Series favorites, or that they're followed by the next most talented teams:

Team Win WS (%)
1 LAD 21.61
2 NYY 11.53
3 TOR 8.70
4 SDP 7.60
5 HOU 6.90
6 NYM 4.82
7 TBR 4.40
8 MIN 4.36
9 ATL 4.17
10 CHW 3.89
11 LAA 3.69
12 MIL 3.00
13 WSN 2.95
14 STL 2.16
15 CLE 1.88
16 OAK 1.56
17 PHI 1.52
18 CHC 1.25
19 BOS 0.85
20 ARI 0.67
21 CIN 0.62
22 SFG 0.42
23 SEA 0.41
24 KCR 0.37
25 DET 0.22
26 PIT 0.18
27 BAL 0.17
28 TEX 0.04
29 MIA 0.03
30 COL 0.03


Finally, we have the most likely matchups for each round of the post season:

NL Wild Card

Matchup Probability
1 NYM-SDP 0.0691
2 ATL-SDP 0.0671
3 SDP-WSN 0.0628
4 PHI-SDP 0.0436
5 ARI-SDP 0.0412
6 MIL-SDP 0.0353
7 LAD-NYM 0.0347
8 LAD-WSN 0.0338
9 SDP-STL 0.0334
10 ATL-LAD 0.0330


AL Wild Card

Matchup Probability
1 TBR-TOR 0.0556
2 NYY-TBR 0.0458
3 NYY-TOR 0.0373
4 LAA-TOR 0.0372
5 HOU-TOR 0.0318
6 LAA-NYY 0.0313
7 LAA-TBR 0.0306
8 HOU-NYY 0.0285
9 HOU-TBR 0.0279
10 MIN-TOR 0.0279


NLDS

Matchup Probability
1 LAD-NYM 0.11605
2 ATL-LAD 0.10245
3 LAD-WSN 0.07820
4 LAD-PHI 0.04805
5 MIL-SDP 0.04705
6 NYM-SDP 0.04145
7 ATL-SDP 0.03745
8 SDP-STL 0.03565
9 SDP-WSN 0.03210
10 MIL-STL 0.02780


ALDS

Matchup Probability
1 HOU-NYY 0.10990
2 HOU-TOR 0.08150
3 LAA-NYY 0.06435
4 HOU-TBR 0.04625
5 LAA-TOR 0.04240
6 CHW-MIN 0.04175
7 NYY-OAK 0.03575
8 MIN-TOR 0.03005
9 CHW-TOR 0.02800
10 OAK-TOR 0.02705


NLCS

Matchup Probability
1 LAD-SDP 0.1302
2 LAD-MIL 0.0927
3 LAD-STL 0.0698
4 CHC-LAD 0.0415
5 LAD-NYM 0.0346
6 ATL-LAD 0.0322
7 MIL-SDP 0.0299
8 CIN-LAD 0.0283
9 MIL-NYM 0.0278
10 ATL-MIL 0.0262


ALCS

Matchup Probability
1 NYY-TOR 0.0678
2 MIN-NYY 0.0593
3 CHW-NYY 0.0510
4 HOU-MIN 0.0480
5 CHW-HOU 0.0451
6 NYY-TBR 0.0445
7 MIN-TOR 0.0434
8 CHW-TOR 0.0401
9 CLE-NYY 0.0322
10 TBR-TOR 0.0305


World Series

Matchup Probability
1 LAD-NYY 0.0747
2 LAD-TOR 0.0567
3 HOU-LAD 0.0511
4 LAD-MIN 0.0340
5 LAD-TBR 0.0334
6 CHW-LAD 0.0286
7 LAA-LAD 0.0272
8 NYY-SDP 0.0270
9 SDP-TOR 0.0247
10 NYM-NYY 0.0215


And the distribution of outcomes for each team:

Results Distribution

There we have it. The 2021 MLB season As Told by Ginger Satchel. After the season I'll write up an autopsy of the model to see how we do, and if time permits I'll post updated projections as we go. If you want to learn more about the changes made to the model since it was first introduced, I've listed them below.

Model Updates

There haven't been many changes to the inner workings of the model, but those that I've made have been important to the final results:

  1. I noticed the random noise being added to team talents was only added to the base talent level once, rather than in each simulated season. The latest iteration of the model fixes that.
  2. I restricted the WAR projections we use to measure team talent to just players on the 40-man roster. Not every player on every 40-man is included in the FanGraphs' projections, but those who are missing are typically minor leaguers who aren't likely to see much playing time. I gave them a projected WAR of zero.
  3. I fixed a bug in the playoff simulations that flipped the probabilities of each team winning any given series. This is why in my first post the Dodgers only had a 2.07% chance of winning the World Series while they jump up to 21.61% in this one.

All of the other updates were really just related to what outputs get saved when you run the model to make the analysis easier. I haven't made these changes public yet because there are still a few more I want to make before unleashing Satchel onto the world. I'll post a write up and documentation once I'm done with those.

Footnotes

1All of the data used is current as of March 28, 2021

2I look forward to being heckled by Mets fans if they end up running away with the division.

3I define a team's schedule strength as the average talent of their opponents for the entire season.

Last Updated: March 29, 2021

Tweet

Follow @andersonfrailey