Running the Pythagorean win-loss numbers on Sugar Land, Somerset and York; a closer look at “luck”
By Paul Braverman
At 30-23 through June 14, we can all agree that the Revolution are putting together a pretty solid first half, and if they aren’t able to catch Sugar Land/Somerset over the final 17 games of the half, they’ll start the second half with more than a decent shot to make the playoffs as either the second half winner or in a wild card scenario.
However, as gaudy as the Skeeters and Patriots records are, one can’t help but wonder if the Revs record would be similar, if they were getting more breaks in 1-run games. (This only further exacerbates the disparity between the Freedom and Liberty Divisions this season.) Of the first 23 losses this season for York, 15 are by 1 run. That’s a staggering number, as 65% of their losses are by a single run. This of course has me pulling out Bill James’s Pythagorean winning percentage metric again, to see how “unlucky” the Revolution are based on their runs scored (319) and runs allowed (250) as of June 14. I’ve written about this before here, but if it’s new to you, here is James’s Pythagorean winning percentage equation:
(Runs Scored)^1.83 + (Runs Allowed)^1.83
The results are simply stunning (if you’re a stat nerd). At 30-23, York’s actual winning percentage is .566. But when you consider the eye-popping amount of 1-run losses and their run differential of plus-69, their Pythagorean winning percentage is .699, suggesting that the Revolution should be 37-16 through 53 games rather than 30-23. That pushes the boundary of statistical possibility on how “unlucky” one team can be based on runs, with York being a full seven games “unlucky” in the actual win-loss column. The Revolution are basically the most “unlucky” team in professional baseball based on run differential, and about as “unlucky” as one team can be based on this metric.
To put it in greater perspective, York’s .699 Pythagorean winning percentage is greater than 36-16 Sugar Land’s (.692) and 33-17 Somerset’s (.660) actual winning percentages. This tells us based on runs this should be a three-team race, though throughout the first half it’s been just a two team race. (York is 6.5 games back of first place Sugar Land on June 14.) But to confidently say it should be a three-team race, we must compute the Pythagorean winning percentage for both Sugar Land (256 runs for, 145 against +111) and Somerset (300 runs for, 214 against +86). Methinks running the numbers will tell us both teams are about where they should be in actual wins and losses based on runs, as you can see right away neither allows very many runs with the two best pitching staffs in the league. Sugar Land’s team ERA is an incredible 2.83 to lead the league, allowing a full 63 runs fewer than Somerset’s 2nd ranked team ERA at 4.00. York is fourth in the league at 4.17. I’m making this assumption before running the numbers. The results:
Sugar Land’s Pythagorean winning percentage: .638 suggesting a 33-19 record, meaning at 36-16 they’re “lucky” by three games. While not as huge a statistical anomaly as York’s “unluckiness,” three games is a significant margin to be lucky by.
Somerset’s Pythagorean winning percentage: .584 suggesting a 29-21 record, meaning at 33-17 they’re “lucky” by four games. Even “luckier” than Sugar Land.
So, not only can York make the argument that based on runs they should be better off in the standings, when you combine York’s “unluckiness” with Sugar Land and Somerset’s “luckiness,” you can really dwell on what might have been in this first half if just a few of those 15 1-run losses had gone the other way. Based on Pythagorean winning percentage, run differentials suggest the Freedom Division standings should look like this:
Sugar Land 33-19
Ouch…perhaps there have been some missed opportunities late in games this season based on what the Revolution has scored and allowed. The great disparity in the Skeeters and Patriots actual records and their Pythagorean records leaves little doubt that’s true. My assumption that their Pythagorean numbers would shake out in a way that would prove their records are right where they should be was quite incorrect. And therein lies the value of Bill James and sabermetrics in the first place, it takes what we see on the surface and shines a whole new light on it. As James said during his cameo on The Simpsons, “I’ve made baseball as fun as doing your taxes!”
The disclaimer: There’s a reason every time I wrote “lucky” or “unlucky” or any variation of “luck,” I put it in quotation marks. While “the numbers never lie,” they don’t always tell the truth either. An oxymoron perhaps, but while Bill James and the numbers magnify what we see on the surface by looking deeper into standings and stat sheets helping us to better understand just what kind of season a team is having beyond numbers that don’t tell the whole story, sabermetrics do not tell us about the un-quantifiable intangibles of teams, specifically how players react in late game situations in tight games. Please don’t read this column as a complaint from a homer that the Revolution “should be” better off than they are. In a statistical sense that may be true over a long period of time with an acceptable sample size of runs scored and allowed. However, each game is it’s own competition, and whether you win/lose by 1 or 10 a win’s a win and a loss is a loss. While the Pythagorean numbers of his team might make Mark Mason cringe, he’d be the first one to admit that the Revolution has let myriad opportunities slip away late in close games during the first half. Run differential or not, those are missed opportunities. On the same token, I’m not suggesting Sugar Land and Somerset have anything to apologize for either, having been “lucky” thus far based on runs. They’ve converted close games into wins more often than the Revolution and certainly deserve the impressive win-loss records they have thus far.