| Simulations for Wednesday, May 30th. Current prediction record is 78-50. | ||||
| Teams Playing | Pitching Matchup | Favored Team | Expected Chance of Winning | Outcome |
| * Dodgers vs Nationals | D.Lowe vs M.Bacsik | Dodgers | 57.85 | Dodgers |
| Padres vs Pirates | C.Young vs P.Maholm | Padres | 56.75 | Padres |
| Giants vs Mets | B.Zito vs T.Glavine | Mets | 56.85 | Giants |
| Cardinals vs Rockies | T.Wellemeyer vs J.Hirsh | Rockies | 63.80 | Cardinals |
| DBacks vs Phillies | R.Johnson vs J.Moyer | Phillies | 55.85 | DBacks |
| * - Sim was reran with Bacsik's ZIPS projections due to small sample size of 2007 innings pitched | ||||
Tuesday, May 29, 2007
Simulations for Wednesday, May 30th
Subscribe to:
Post Comments (Atom)
4 comments:
Can you explain how your software would be giving the Nationals (21-31) an expected win over the Dodgers (30-21) today? Kind of makes it hard for me to trust it. Apparently the Nats pitcher Bacsik just got his first ML win since 2004 last week. Maybe there's a small sample problem here? Could it need some tweaking maybe?
Yes, pitching stats have been moved from ZIPS to 2007 pitching data, so there is a small sample size for the Nationals pitcher. I may institute a minimum number of innings pitched rule for starting pitchers if this becomes a problem. So far it hasn't. Let's see how the game pans out. If a pitcher doesn't meet the minimum IPs then I may use his ZIPS projections instead. But I want to be consistent. For example, Lincecum is much better than his ZIPS projection, but for a while had a small sample size of innings. Thanks for the post.
vr, Xeifrank
Thanks for looking into it, xf. It was the the high probability (about 58% or so) that looked wrong. Much better this way. How many innings are you now using as the determinant minimum for using current year stats vs. ZIPS?
Berk, I'm probably going to go with (0.5 * TeamGamesPlayed) for starters innings pitched and about 0.25 for relief pitchers. I'm holding off on converting hitters from ZIPS to 2007 data until I find a good way to automate it.
vr, Xeifrank
Post a Comment