500 have a standard deviation of around six games, so a similar calculation applies to them.įor instance, suppose you have a. So far this is just an intellectual exercise because, of course, not every team is a. Also, there should be only one team above 93 wins (we had two), with the next best at 89. On average, the difference between the best team and the worst team will be about 24 games, not 27. 500 teams - but, by chance alone, the Mets finished ahead of Arizona by 27 games!Īs it turns out, this season is a little more extreme than usual. In this simulated, randomized season, the Mets in the East and Diamondbacks in the West were both really. This is a lot easier to picture if you see a real set of standings, so Table 1 shows a typical result of a coin-tossing season for a hypothetical National League where every team is. Furthermore, of those five teams, you’d expect one of them (actually, about 0.8 of a team) to finish more than 2 SDs away from the mean - that is, with more than 93 wins, or more than 93 losses. If all 16 teams in the National League were exactly average, you’d nonetheless expect five of them to wind up with more than 87 wins or with fewer than 75 wins. More interesting are the one-third of the seasons that fall outside that range. Two-thirds of the outcomes should be within that range, so if you were to run 300 coin-seasons, or 300 cloned-Yankee seasons, you should get 200 of them winding up between 75 and 87 wins. The SD can be thought of as a “typical” difference due to luck - so with an SD of six games, a typical record of a coin tossed 162 times is 87–75, or 75–87.
500 team can be described by a normal (bell-shaped) curve, with an average of 81 wins and a standard deviation (SD) of about six wins.
(That’s hard to do with real players, but easy in a simulation game like APBA.) Again, on average, each team should win 81 games against each other, but, again, the records could vary significantly from 81–81, and the difference would be due to luck.Īs it turns out, the range and frequency of possible records of a. Suppose you were able to clone a copy of the New York Yankees, and play the cloned team against the real one.
Just by random chance alone, your team might go 85–77, or 80–82, or even 69–93. But for any individual season, the record may vary significantly from 81-81. You’d expect, on average, to get 81 wins and 81 losses. Each time it lands heads, that’s a win, and when it lands tails, that’s a loss. Suppose you take a coin and flip it 162 times to simulate a season. 34 (2005).Ī team’s season record is massively influenced by luck. This article was published in SABR’s Baseball Research Journal, Vol.
0 kommentar(er)
