It’s no secret that an error has been detected which caused Rome Generals catcher Cody Richards (
http://www.worldbaseballassociation.com/fssl/reports/news/html/players/player_3076.html), who has an avoid Ks rating of 11, to strike out 202 times in 125 games last season. Manually editing his rating to 212 (one less than the maximum behind-the-scenes rating of 213, where he was last season) seems to have fixed the problem, as Richards struck out just 4 times in the first sim of the 1933 regular season. But could this OOTP glitch have really cost the Generals the title?
It would be easy (well, probably not easy, but easier) to say that based on Richards’ negative WAR and the WAR he is on pace to reach this season, the Generals should have won x more games. I don’t know what x is, though, because I took a different approach. Abstractions in WAR are all well and good in theory, but if Rome is claiming to have lost the 1932 title unfairly, I decided to look at the actual games in 1932 and see how Richards having his real ratings would have affected the league.
Predicting the actual outcomes of games is notoriously difficult – if some player hadn’t struck out, there’d be one fewer out, but who can say what the next batter would have done? To solve this issue, I used a projection system. Richards’ current ratings are 9/9/3/4/11. There are no other players with those exact ratings, but the closest comparisons I found were Ottawa 2B Kareem Spence (9/9/4/5/11) and Rome LF Angelo Adorno (9/9/3/5/11). As a side note, Rome 3B Elias Salazar (9/9/3/5/11) is also a good comparison, but he did not play any games in the majors in 1932. Spence hit .297 and Adorno hit .292 in the 1932 season, and while Spence’s contact rating didn’t reach 9 until July, Adorno’s reached 9 in March, before the season started. I thus made the assumption that Richards would have hit somewhere between .290 and .300 if his rating had been correct. For anyone curious, at the time of this writing, this season, Richards is hitting .286, Spence is hitting .328, Salazar is hitting .289, and Adorno has not played. I therefore believe the assumption of approximately .300 is warranted. As further proof of this, Richards had 468 at-bats in 1932 and struck out 202 times – if 30% of those Ks were instead hits, that would be an extra 60.6 hits to add to his total of 68, giving him 128.6 hits in 468 at-bats – an average of .274, a slight discrepancy that could be explained by the fact that a lowered Ks rating could also lead to a lower contact rating.
With this assumption in mind, I looked at every one of Rome’s 1932 losses and examined every one of Richards’ at-bats based on a few factors – most importantly, whether there were runners in scoring position or on first base, and how many outs there were. I made the simplifying assumption that any time Richards singled with a runner on 2nd or 3rd, that would lead to a run. So, every time Richards struck out with a RISP, I counted it as 0.3 runs, and if he struck out with runners on both 2nd and 3rd, it would count as 0.6 runs (30% chance of scoring 2 runs). Furthermore, in 1932, the Generals had 1566 hits in 5730 at-bats. After subtracting Richards’ 68 for 468, that gave a team-minus-Richards batting average of approximately .2847. However, with a runner on first, it became necessary to differentiate between singles and extra-base hits. In 1932, 28 of Richards’ 68 hits were doubles or triples, or 41.2%. Comparisons with Adorno’s and Spence’s rates indicated that this rate was probably slightly higher than would be expected from a full season – Spence’s rate was 41.5%, but his speed and baserunning ratings are much better than Richards’s, while Adorno’s rate was 32.8%, but his speed and baserunning ratings are somewhat worse than Richards’s. Thus, I made the assumption that approximately 35% of Richards’s hits would be extra bases. Therefore, with a runner on first, Richards had a 30% chance of getting a hit, which had a 35% chance of going extra bases – leading to 0.105 runs on average. That hit had a 65% chance of being a single, which would require another hit to score a run – leading to 0.0555 runs (30% chance of a hit * 65% chance of a single * 28.47% chance of the next Rome hitter to get a hit). Thus, the total expected runs from a Richards strikeout with a man on first was 0.1605 runs. For mixed situations, I could simply add the numbers - a K with a runner on first and second or first and third was worth 0.4605 runs, and with bases loaded, 0.7605 runs. Finally, every time Richards got a hit, he became a baserunner himself, with the potential to score a run. Rome's XBH% in 1932, minus Richards's stats, was 31.8%. Thus, he could score a run in one of two ways - one extra base hit, or one single followed by one hit of any kind. Therefore, for every at bat, there was an additional (30% chance of a hit * ((28.47% chance of a Rome hit * 31.8% chance of an XBH) plus (28.47% chance of a hit * (68.2% chance of a single * 28.47% chance of another hit afterwards)) = 0.0437 runs on average, which had to be accounted for, simply by adding it to all of the other expected runs (since it was present no matter the situation). However, at this point, I had to take into account the number of outs - after all, if he gets a hit with 0 outs, the Generals had 3 chances to score him. This took me a long time to fully account for, and I won't bore you all with the details (if you haven't just skipped over this part!) but suffice it to say, I made a formula to determine the actual expected runs based on the number and location of baserunners
and the number of outs. I can share the Google Sheets if anyone's curious, but the final numbers ranged from 0.04 expected runs (bases empty, 2 outs) to 1 expected run with bases loaded and 0 outs (this was actually .996 expected runs, but I rounded my final numbers to the nearest hundredth to make the in-game application more practical.)
Well, with all of that math out of the way, I finally got around to actually analyzing all of Rome's losses to discover how many games this overflow error truly cost them.
Rome's April 15th game against Manila saw Richards strike out three times - once at the top of the inning (0.17 expected runs), and once with the bases loaded and one out (0.93 expected runs), for a game total of 1.10 expected runs. (The third strikeout was with nobody on base and nobody out, but the third strike was dropped and he reached first, which I counted as a single, not a K, for expected runs purposes.) Rome lost this game by just 1 run.
In a June 6th game against Buenos Aires, Richards struck out 5 times - once with 0 outs and a runner on 1st and 3rd (0.7 expected runs), once with 2 outs and nobody on base (0.04 expected runs), twice at the beginning of the inning (0.34 total expected runs), once with 0 outs and a runner on 1st (0.4 expected runs), for a game total of 1.48 expected runs - in a game Rome lost by just 1 run.
On June 11th in a game against Mexico City, Richards had 3 strikeouts - two with nobody on base and one out (0.26 total expected runs), one with one out and runners on 2nd and 3rd (0.73 expected runs), for a total of 0.99 expected runs, in a 1 run loss. This doesn't really qualify as an unfair loss (it should hav ended in a tie), so when revising the standings, I called this a tie - there's no way to know who would win in extra innings.
On July 2nd, Rome lost to Santo Domingo by one run. In that game, Richards struck out 3 times - once with a man on third and two outs (0.34 expected runs), once with a runner on first and no outs (0.4 expected runs), and once with a runner on second and two outs (0.34 expected runs), for a total of 1.08 expected runs.
In a September 13th game against Melbourne, Richards struck out 4 times in Rome's 1-run loss - once at the top of the inning (0.17 ER), once with one out and a runner on 2nd (0.43 ER), once with one out and nobody on base (0.13 ER), and once with two outs and a runner on 2nd (0.34 ER), for a total of 1.07 expected runs.
Overall, I found 4 games that Rome lost but would have been expected to win if this error hadn't occurred, plus one extra that had to be called a tie. Rome therefore has a valid argument that they should have been 4th with a record of 85.5-70.5, overtaking Santo Domingo (who would have had one more loss than they did),. In fact, if we simply assume Rome would have won the "tie" against Mexico City, this would put them in a three-way tie for second, with Barcelona and Melbourne.
Is Rome justified in declaring themselves the 1932 FSSL champions? Probably not. However, there are several factors at play here, and others may not agree with my methodology. I tried to be generous to Rome in my assumptions (such as the assumption that a single will always score a runner on second, which is not always true and likely would decrease some of the expected run totals), but it's possible I should have been even more generous (assuming a 40% XBH rate for Richards rather than a 35% rate, for example, could have affected some run totals as well). My overall takeaway from this is that Rome would not have been the clear best team in the league, but could have easily been in the conversation and probably would have had an outside shot at the title.
Guests: 3
Hidden: 0
Users: 0