Perception, reality and expectation can be at odds over the course of 162 games during the MLB season. Strength-of-schedule differences are a lot harder to see in baseball for a variety of reasons and are things most bettors don’t consider. The imbalanced schedule (which is going away next season) has teams playing 76 of their 162 games against division opponents. A team in a bad division can really take advantage of that, given that intradivision games account for nearly 47% of the schedule.
There are three alternate standings metrics that bettors and analysts use to evaluate a team’s performance to date and project into the future. They are Pythagorean Win-Loss, BaseRuns and 3rd Order Win%. Isolating differences between actual record and these alternate records can shed some light on teams that may be worth betting on and those worth betting against.
A team’s Pythagorean Win-Loss Record (Pyth W-L) is based entirely on run differential. The comparison between Pyth W-L (also sometimes called Expected W-L) and actual record is usually an indicator of performance in one-run games. It will also be an indicator of a team that has gotten blown out a lot.
For example, going into Monday’s games, the Pirates were 16-24 overall but had a Pyth W-L of 11-29. The reason? They’ve lost 10 games by 5+ runs and have been outscored 115-37 in those games. They have a 21-0 loss and also just had an 18-4 loss over the weekend, which will greatly skew run differential.
Consider a team like the Marlins. Miami had a record of 18-22 entering Monday’s action but had a positive run differential, leading to a 22-18 Pyth W-L record. The reason? The Marlins were 6-13 in one-run games. A lack of success in one-run games will lead a team to have a much lower Pyth W-L than actual record.
Similarly, the Brewers were 9-3 in one-run games, so their 26-15 actual record looked more like 24-17 by Pyth W-L. Most teams fall within + /- 3 games in one-run games record, but there will be significant outliers. The 2021 Mariners had a 90-72 record but a 76-86 Pyth W-L record. They were 33-19 in one-run games.
Some teams with stronger bullpens tend to do better in one-run games than others, but everybody usually levels off.
BaseRuns is my favorite alternate standings metric. The concept behind BaseRuns makes sense, but can be tough to fully embrace. Think of these six outcomes — 1B, 1B, HR, K, K, K. Depending on how those outcomes are sequenced, a team could score 0, 1, 2 or 3 runs. HR, 1B, K, K, 1B, K would be one run. 1B, HR, K, 1B, K, K is two runs. If all three strikeouts come first, that would be zero runs in the inning.
BaseRuns creates a context-neutral environment that removes the sequencing of outcomes. It takes a team’s entire set of outcomes, throws them into a hopper, mixes them up and then creates expected runs scored per game (R/G) and expected runs allowed per game (RA/G).
Then, using the expected R/G and expected RA/G, the Pythagorean Win-Loss formula is applied. The rationale is to remove luck and sequencing factors. A home run is a home run, but a home run with the bases loaded came at a better time than a solo home run. The “Clutch” factor is not a predictive trait. Timing and sequencing can really alter a team’s results in a positive or negative way.
Entering Monday’s games, the Padres were + 4 in BaseRuns record as a 27-14 team with an expected record of 23-18. In actuality, the Padres were + 29 in run differential with 4.51 R/G and 3/8 RA/G, but BaseRuns had the Padres down for 4.22 R/G and 3.74 RA/G. That would give the Padres a + 20 run differential, and the record is lowered as a result.
The Padres went into Monday’s game batting .257/.347/.423 with men in scoring position but .209/.289/.327 with the bases empty. They are much more productive with RISP than with nobody on base. The discrepancy between BaseRuns and actual record generally says a lot about the timing of a team’s hits, which can make an offense (or a pitching staff) look a lot better or a lot worse than it is.
3rd Order Win%
This is the one that most data-driven baseball bettors will turn to when looking for positive or negative regression on a team level. 3rd Order Win% takes things a step further by accounting for strength of schedule. The record is adjusted for quality of opponents and a lot of “underlying metrics.” The formula uses “Adjusted Equivalent Runs Scored and Allowed,” which is adjusted for team pitching and defense.
Take the Yankees for example. New York entered Monday’s action at 29-12 but had a 3rd Order Win record of 24.1-16.9. The Yankees had scored 195 runs and allowed 128 runs, but Adjusted Equivalent Runs took their run differential from + 67 down to + 33 because of the lack of quality competition. Like BaseRuns, Pythagorean Win-Loss is then used to determine the expected record based on the adjusted run differential.
The Phillies are a team that seems rather unlucky. They were -4.7 games in 3rd Order Win% record, playing more like a 23.7-17.3 team based on their performance and the quality of competition. Instead, they were 19-22, despite a + 11 run differential. Being 3-7 in one-run games doesn’t help, but that is a team with a bad bullpen. Philadelphia was 9-0 when scoring eight or more runs and just 10-22 when scoring seven or fewer entering Monday’s 7-3 win.
The difference between actual record and the alternate standings metrics can point us toward teams that have signs of positive and negative regression. Applying additional layers of context is still extremely important, but you can use this to isolate teams to bet on or against as the season goes along.