Odds of Winning the Division

We’re now in the home stretch of the my baseball simulation league season, and I thought it might be fun to try to run some numbers to see what the odds are for each team to win their division.

Most sites that calculate these sorts of probabilities (FanGraphs, Baseball Reference, Baseball Prospectus, and FiveThirtyEight) do so by running thousands of simulations (maybe using Diamond Mind Baseball) to determine how often a given team wins the division. I tried to come up with a different method that just uses the raw numbers from the standings to spit out some odds.

I’m looking at how good your team is, how many games back you are, and how many games are left in the season. No, it doesn’t take into account remaining home/road games or strength of schedule or how many games left against teams in your division or anything like that. I’ll leave all the micro-split analysis to others. Micro-split analysis is pretty much all you hear these days from the sportscasters, and most of it is bullshit.

Here are the data points:

a) team record (PCT)

b) Pythagorean Winning Percentage (PY PCT, which uses run ratio, not differential, because a team that wins a game 1-0 is better than a team that wins a game 10-9)

c) number of games behind in your division (GB)

d) number of games left in the season (GL)

And here’s how it works:

Talent = (PCT + PY PCT) / 2
[Uses both record and run ratio to better approximate how good a team is.]

Urgency = MIN(GB / GL, 1)
[I calculate it this way to avoid any division by zero errors and to hard-set a value of 1 for any teams that have been mathematically eliminated.]

Chance = Talent * (1 – Urgency) ^ 10
[The exponent of 10 is arbitrary, but seems to produce probabilities that are similar to the ones created for MLB teams on FanGraphs. The result is a raw number that has little meaning until it is normalized by the last step.]

Win Division % = Chance / SUM(Chance for all teams in the division)
[Normalizing the numbers so that all the numbers in a given division neatly add up to 100%.]

The results, updated through 8/15/2019:

Odds change each day as the end of the season draws nearer. Anyone who’s already been mathematically eliminated will show 0%, and those who are on the verge will be way down there.

Also, these numbers might make it easier to objectively decide when to close up shop for the season and start rebuilding for the future. This guy recommends that at the trade deadline, “If you have more than a 15% playoff odds, then BUY. Else SELL.”

Might expand this later to try to figure out what each team’s odds are of being a wild card, getting to the playoffs, or winning the championship.