PDA

View Full Version : Free Marcel projections through the end of 2008



RedsManRick
07-14-2008, 12:30 PM
Sal Baxamusa over at THT has released two spreadsheets that can be used to run Marcel projections through the end of 2008. For those of you who don't know, Marcel is an extremely basic projection system which merely regresses player performance to the mean. Remarkably, it's only marginally less accurate than the significantly more complex systems like PECOTA.

http://www.hardballtimes.com/main/article/is-this-guy-for-real/

For example, I've run both Edinson Volquez and Bronson Arroyo -- two guys who both should do some regressing in the 2nd half.



Volquez IP ERA K BB WHIP
2008 to Date 117.7 2.29 126 56 1.24
Proj Balance 89 3.75 80 39 1.36
2008 Totals 207 2.92 206 39 1.29

Arroyo IP ERA K BB WHIP
2008 to Date 104 5.97 93 41 1.66
Proj Balance 79 4.28 63 27 1.49
2008 Totals 183 5.24 156 68 1.59

While past performance is the best predictor of future results, it's important to remember that "past performance" means a lot more than just what a guy has done this year. It's what he's done in the past and what we general expect from all players.

Big Klu
07-14-2008, 12:55 PM
http://upload.wikimedia.org/wikipedia/commons/8/8b/Marcel_Marceau.gif



Sorry...I couldn't resist. :D

Rojo
07-15-2008, 09:30 PM
http://upload.wikimedia.org/wikipedia/commons/8/8b/Marcel_Marceau.gif



Sorry...I couldn't resist. :D

" "

jojo
07-16-2008, 08:43 AM
Actually, the Marcel "doing the projections" is a monkey....

Basically Marcels looks at the last three years of data for a player (weighting the most recent year the heaviest) and then regresses the player's performance to the mean that data suggests.

A projection for Volquez is a bit problematic for obvious reasons....

Spring~Fields
07-16-2008, 09:30 AM
Sal Baxamusa over at THT has released two spreadsheets that can be used to run Marcel projections through the end of 2008. For those of you who don't know, Marcel is an extremely basic projection system which merely regresses player performance to the mean. Remarkably, it's only marginally less accurate than the significantly more complex systems like PECOTA.

http://www.hardballtimes.com/main/article/is-this-guy-for-real/

For example, I've run both Edinson Volquez and Bronson Arroyo -- two guys who both should do some regressing in the 2nd half.



Volquez IP ERA K BB WHIP
2008 to Date 117.7 2.29 126 56 1.24
Proj Balance 89 3.75 80 39 1.36
2008 Totals 207 2.92 206 39 1.29

Arroyo IP ERA K BB WHIP
2008 to Date 104 5.97 93 41 1.66
Proj Balance 79 4.28 63 27 1.49
2008 Totals 183 5.24 156 68 1.59

While past performance is the best predictor of future results, it's important to remember that "past performance" means a lot more than just what a guy has done this year. It's what he's done in the past and what we general expect from all players.

Will it work even for players like Griffey, Patterson, Bako and the cleanup hitting Phillips against right handed pitchers ?

cumberlandreds
07-16-2008, 09:41 AM
http://upload.wikimedia.org/wikipedia/commons/8/8b/Marcel_Marceau.gif



Sorry...I couldn't resist. :D

I thought he was being held prisoner when I saw the title of this thread. :D