With one half of MLB’s regular season in the books, we have a good sample size of games to evaluate the strength of the various teams so far. While many analysts will simply look at winning percentage to evaluate how a team has performed, my Expected Winning Percentage metric incorporates a few other variables in order to better capture team quality:

  1. Accounting for close game luck: Using Bill James’s pythagorean expectation formula, one can reliably estimate how many games a team *should* have won based on the number of runs they scored and allowed. Essentially, if a team scores more runs than they allow (ie: have a positive run differential) they should be a winning team over time. While teams can sometimes deviate from their pythagorean expectation winning percentage, this is usually a result of luck rather than skill; mediocre teams that do well in one-run games, for example, can sometimes post good Win-Loss records early in the season, but these teams usually regress later in the season when their luck runs out.
  2. Using expected runs scored and expected runs allowed to account for Cluster Luck: While adjusting for close game luck is important, it’s not the only type of luck that can affect a team’s winning percentage teams; Cluster Luck is also important. Cluster Luck is simply how lucky a team is at sequencing its hits in an order that produces more runs (or for pitchers, sequencing its hits allowed in an order that produces fewer runs). As an example, say that Team A and Team B each have three walks, one home run, and three outs in an inning. The home run for Team A, however, happens before the three walks, so Team A only scores 1 run, whereas the home run for Team B happens after the three walks, scoring 4 runs. Even though both teams had the same inputs, Team B scored 3 runs more than Team A because of cluster luck. In order to account for cluster luck, my model estimates how many runs a team should have scored and allowed based on that team’s underlying inputs: singles, doubles, triples, home runs, walks, errors, stolen bases, caught stealing, double plays, outfield assists, etc.
  3. Adjusting for strength of schedule: This is the step that makes my ranking system unique among publicly available rankings. Steps 1 and 2 are also featured in FanGraphs’ BaseRuns Standings, for example. Strength of schedule, however, is often ignored when evaluating MLB teams, presumably because the opponent quality does not differ as dramatically as it does in college sports, for example. Still, the MLB schedule is quite unbalanced, with teams playing divisional opponents 19 times each compared to just 6 or 7 games against intra-league opponents and just 20 total games against the 15 inter-league teams. As a result, when certain divisions are unusually weak or strong (which is definitely the case this year), methods not adjusting for schedule strength will underrate teams playing in strong divisions and overrate teams playing in weak divisions.

Expected Winning Percentage Rankings

Team Winning Percentage Luck BaseRuns Strength of Schedule Overall Difference Team Expected Winning Percentage
Dodgers 0.615 -0.028 0.643 -0.001 -0.027 Dodgers 0.642
Giants 0.64 -0.006 0.646 -0.006 0.000 Giants 0.640
Astros 0.604 -0.030 0.634 -0.003 -0.027 Astros 0.631
Rays 0.589 0.005 0.584 0.007 -0.002 Rays 0.591
Blue Jays 0.517 -0.043 0.560 0.015 -0.058 Blue Jays 0.575
White Sox 0.607 0.002 0.605 -0.048 0.050 White Sox 0.557
Padres 0.57 0.015 0.555 0.000 0.015 Padres 0.555
Yankees 0.517 -0.027 0.544 0.006 -0.033 Yankees 0.550
Marlins 0.438 -0.097 0.535 0.013 -0.110 Marlins 0.548
Athletics 0.565 0.026 0.539 0.004 0.022 Athletics 0.543
Braves 0.494 -0.032 0.526 0.005 -0.037 Braves 0.531
Red Sox 0.604 0.086 0.518 0.003 0.083 Red Sox 0.521
Nationals 0.472 -0.031 0.503 0.018 -0.049 Nationals 0.521
Mets 0.54 0.011 0.529 -0.008 0.019 Mets 0.521
Brewers 0.576 0.031 0.545 -0.029 0.060 Brewers 0.516
Angels 0.506 0.007 0.499 0.009 -0.002 Angels 0.508
Phillies 0.5 0.022 0.478 0.029 -0.007 Phillies 0.507
Reds 0.533 0.040 0.493 -0.015 0.055 Reds 0.478
Rockies 0.44 -0.010 0.450 0.006 -0.016 Rockies 0.456
Cardinals 0.489 0.019 0.470 -0.014 0.033 Cardinals 0.456
Mariners 0.527 0.079 0.448 0.007 0.072 Mariners 0.455
Rangers 0.389 -0.040 0.429 0.018 -0.058 Rangers 0.447
Cubs 0.489 0.045 0.444 0.002 0.043 Cubs 0.446
Twins 0.438 -0.030 0.468 -0.036 0.006 Twins 0.432
Cleveland 0.517 0.054 0.463 -0.034 0.088 Cleveland 0.429
Tigers 0.44 0.008 0.432 -0.022 0.030 Tigers 0.410
Pirates 0.378 -0.027 0.405 -0.006 -0.021 Pirates 0.399
Orioles 0.315 -0.052 0.367 0.023 -0.075 Orioles 0.390
Royals 0.404 0.013 0.391 -0.018 0.031 Royals 0.373
Diamondbacks 0.283 -0.052 0.335 0.034 -0.086 Diamondbacks 0.369

A few patterns stand out when comparing expected winning percentage with each team’s actual winning percentage. First, it’s clear that the Central divisions in both leagues – but particularly the AL Central – are especially weak. Incredibly, just two out of the ten teams playing in the AL and NL Central have expected winning percentages above .500, and six out of the ten rank among the eight worst teams in the majors. These weak divisions have worked to the benefit of the AL Central leading White Sox and NL Central leading Brewers. Accounting for their weak schedules drop the White Sox from 3rd in winning percentage to 6th in expected winning percentage and the Brewers from 7th to 15th.

In terms of close game and sequencing luck, the Marlins stand out as particularly unlucky; their BaseRuns winning percentage is nearly 100 points higher than their actual winning percentage. This is mainly a result of an atrocious record in close games; they have actually scored 17 runs more than their opponents despite being 11 games under .500. Furthermore, they have also played a difficult schedule; while the NL East doesn’t have any elite teams, all five teams have expected winning percentage’s above .500. Indeed, despite ranking just 25th in actual winning percentage, after adjusting for their poor luck and difficult schedule the Marlins rank 9th in expected winning percentage, suggesting that they may be in line for some second half improvement.

The Red Sox hot start, meanwhile, may be somewhat of an illusion; despite ranking 4th in the majors in winning percentage, their expected winning percentage ranks just 14th. The Red Sox have benefited from both types of luck discussed above: close game luck and sequencing luck have netted them 4 wins each, for a total over-performance of 8 wins.

With the exception of a few outliers (the Red Sox, White Sox, and Marlins, most notably), the Expected Winning Percentage rankings look similar to the regular standings. The Dodgers, Giants, and Astros stand out as a clear top three, with the Rays and Blue Jays following behind. From there, twelve teams (the White Sox, Padres, Yankees, Marlins, Athletics, Braves, Red Sox, Nationals, Mets, Brewers, Angels, and Phillies) all are both above .500 and within 50 points of each other in expected winning percentage, quality teams that nonetheless need to play better in the second half. The remaining 13 teams are noticeably worse, and would be wise to sell at the July 31st trading deadline rather than try to make a run for the playoffs.

Note: All data includes games through Sunday, July 11th. Games played after the All Star Break are not included in this analysis.