THE BOOK--Playing The Percentages In Baseball

Kind of surprising to me just how bad the previous year’s stats did

Thanks, Tango.

Pitchers soon.  Teaser: ZiPS seems to know it’s pitching.

What is being done in the case where PECOTA (or CHONE, or ZiPS) forecasts a player when Marcels does not?

For hitters such players were just dropped from the sample but for pitchers I’m going to look at a wider group.  Marcel gives a default (league average) prediction for everyone who doesn’t have prior major league stats.

I think you will see an improvement in PECOTA’s results if you do the same (predict the league average) for any player not forecasted by Marcels; at least, that’s how it appears when I ran a similar test (using a slightly different sample of players).

I would presume that CHONE and ZiPS show similar improvement under the same conditions, but I haven’t finished ID mapping them yet and so I don’t have those results yet.

That makes sense.

yeah, the ID mapping is a pain in the neck.  Colin, I’d be happy to send you the 512 pitchers I have lined up (with pecota, marcel, zips, oliver, cairo, steamer, chone, fantistics and sporting news) but it might not have all the players/stats you’re looking for.  I’m going to try to get everyone to send me their 2010 projections *with* MLBAMIDs to make this really quick and easy for next year.

You should keep rookies in the test and assume a league average projection for Marcel.  In some cases Marcel will still win (Wieters).

Actually, Marcel may not do too bad in a lot of cases.  Almost all minor leaguers should be projected as below average players.  If they weren’t, then they wouldn’t be minor leaguers, would they?  But the ones who are allowed to play in the majors will be a lot closer to average, so the way this test selectively samples helps Marcel out.

The real value of a projection system that uses minor league data is to tell you beforehand which players deserve a chance to play in the majors.

Rally, so long as you’re around, do you have a copy of the ‘09 CHONE projections with IDs included? The copy of the ‘09 hitter projections I have lacks IDs.

What is being done in the case where PECOTA (or CHONE, or ZiPS) forecasts a player when Marcels does not?

Marcel actually forecasts MORE players than any system out there.  As per the FAQ:

FAQ: “But, what about a player who’s never played MLB? Where’s his forecast?” That’s simple. His forecast is the league mean over 200 PA, 60 IP (starter) or 25 IP (reliever). If you want to know what the league mean is, just take the average of anyone forecast with a reliability of 0.00. So, Marcel’s official forecast for anyone coming over from Japan is that.

I really wish Marcel was accurate for me, I’d be worth 6-8 million bucks to a team.  Not a criticism at all, just my preference for being on a ballfield instead of a cubicle.

Colin, I’ll see what I have.  I used my own ID system last year since I was matching things by player name.  Went to MLBAM this year.

I do have them.  Not with the projection spreadsheet but I do have a map.  Just match up by name, all my player names should be unique (ex. Drew Anderson and Drew Anderson1).

Just send me an email and I’ll shoot them back to you, or post your email here.  I know you’ve sent me stuff in the past, but I think that was before I got this computer.

You (or anyone else, for that matter) can reach me at:

pontifexexmachina at hotmail dot com

Pitchers soon.  Teaser: ZiPS seems to know it’s pitching.

I heartily approve of this conclusion!

Do any of you guys have a good spreadsheet with MLBAM IDs?  I could try to map them - I usually just go with the DMB IDs since I store projections within DMB.

Dan, see link in Section 3:

http://www.tangotiger.net/forecast/rules.html