The most common use of a recharge round is to allow the undefeated winner of the upper bracket a second chance if it loses the initial bracket unification match. But a recharge round can be used in other ways also.
The 2017 Big 10 baseball championship being held in Bloomington over the next five days has a bracket unlike any other I’d encountered. In it, there are two possible recharge matches, neither of which eliminates the possibility that a team will can fail to win the overall championship even when it has only one loss.
This bracket, B1G baseball 2017, has a number of interesting features, which can be appreciated by comparing it to a standard 8 bracket (A.B.|.C.X): B1GbaseballAlt.
I do not claim that the simulation recorded on these analyzed brackets is a fair prediction of the likely result of the tournament. Apart from setting luck to 3 (to reflect the fact that there is a great deal of chance variation in the result of a single baseball game), I’ve made no effort to incorporate any of the subtle factors that SABRmetricians have found useful in predicting baseball results, instead relying on my very simple general match model.
The tourney itself is played by eight teams over five days on a single baseball field. Usually, what makes a tourney sensitive to the number of rounds played, it is because several matches can be played at the same time, at least in early rounds. But a short baseball tourney is round-sensitive for a different reason. Any difference in the number of rounds played could significantly affect the balance of play because the team that plays more rounds would tire its pitching staff more.
This would suggest the use of a shifted bracket, but there’s no sensible shift available for an eight-team tourney. But the unusual bracket in use for the tourney does alter the spacing of the games in a way that seems likely to help equalize the strain on pitchers. The standard bracket would have the upper bracket champion sit out two full rounds waiting for the lower bracket to resolve itself into a single challenger. But in the tourney as played, no team has to sit out more than one round at a time.
The first step in understanding how this is accomplished it to avoid the official bracket drawn by the conference, which can be found here. That bizarre drawing is an object lesson in how not to present a bracket. Refer instead to my drawing of the structure, linked to above.
The basic idea is to unify the upper and lower brackets one round earlier than usual. As with the standard bracket, this causes a possible problem in that the two players in these games (the ones played at 10 am and 1:30 pm on Saturday) match one team that’s so far undefeated against a team with one loss. So, in the same way that a recharge round comes into play if the lower bracket team wins the “X” match in a standard bracket, there can be recharge rounds to determine either or both of the teams that move onto the final.
The analyzed brackets show that the performance of the two brackets is almost identical, with one exception. Fairness (C) is equal in the second decimal place. The overall win chance for any team differs by 0.1%, at most. The one factor that is different is the incidence of repeat pairings. The bracket that will be used generates, on average, one more repeat pairing than the standard bracket does. This, in a tourney consisting of, at most, 15 games, is a huge difference.
This one-game difference is directly related to the recharge rounds. A recharge round is, by definition, a repeat pairing, but each of the two happens, on average, only about half the time, so the two together raise the average number of repeats by one game. Is this a matter of concern? Apparently, the powers that be in the Big 10 Conference think not, and I can see why they would feel that way. While in tournaments of other kinds, repeat pairings are considered blemishes that should be minimized, in baseball there’s a long tradition of playing consecutive games against the same opponent. It’s well-enough established that it even has a name: the “double-header”. And there are interesting strategic factors in playing a double-header, especially where one team need to win both games, while the other will be happy with a split. I think the two possible Saturday double-headers in the B1G Baseball tourney might well be considered a feature rather than a bug.
There is no possible recharge for the Sunday championship game. It would be impractical, even if one were wanted. It’s possible that both of the teams in the final will be undefeated, and so if the tourney wanted to hew strictly to the double-elimination format, it might be necessary to play two recharge rounds, not just one.
Without a recharge round, there’s one possible outcome that is a bit unpalatable. It can happen that two teams go undefeated against all of the other teams in the tourney, and have one win and one loss in two games against each other. If this happens, the champion is the team that won in the final, but it may well seem unfair, as the teams records are essentially identical. This result, happily, is less likely to happen than it would be if a standard bracket were used. The final match is a repeat in 68% with the standard bracket, but only 27% of the time with the bracket that will be used.
All in all, the B1G Baseball bracket seems to be well suited to this tourney, and the Big 10 should be congratulated on the innovation (though they really should learn how to draw it). In future posts, I’ll explore its suitability for other sorts of tourney, but I’m inclined to think it will show well – perhaps even well enough to be admitted to the pantheon of recommended formats on my printable brackets page.
2 thoughts on “Recharges and B1G Baseball”
I like the pure, slightly different full double elimination format that the NCAA used from 1950 to 1987 with the 8 team college world series. The only reason that they changed the format was because CBS wanted to carry the final only and CBS wanted it to be a single game. If the Big 10 is going to have an 8 team tournament, I would like to see them revive that format.