Rating drift

A team in February is not the same team in May. Rating drift is the amount a team's rating is allowed to change between today and the end of the season when PEAR simulates the rest of a season.

• In each simulated season, every team gets its own random change in rating, which arrives gradually over the remaining games.
• The possible change is large when most of the season is left and small near the end.
• Its size was set from how much real ratings have moved over the rest of past seasons.

• Drift barely changes a team's expected record. It widens the range of outcomes around it.
• That makes preseason and early-season odds less certain, which matches how seasons actually go.

• simulate-wins uses it for D1 baseball, as do the projected records and résumé odds on the site.
• Game lines and win probabilities for single games (games, matchup) do not: they use today's ratings.

• It applies to D1 baseball only. Other divisions and softball are simulated with fixed ratings.
• It is the same for every team. It does not know that a young team may improve more than an old one.