When to cut underperforming studs? (1 Viewer)

[RommelDAK]

I wrote this a couple of years ago when Chris Hope, a concensus pre-season #1 db, was putting up atrocious numbers. I wondered now many weeks of data represented a reliable sample for saying, "Clean out your locker, son!" There are, of course, many other factors to consider in addition to raw performances, but this tries to put the latter into persective:

http://www.econ.tcu.edu/harvey/dak_ffl/forecasting_idp.pdf

For those who don't want to read the whole thing, it basically says that past week 5 you really don't learn much more. Here is the summary at the end of the paper:

LESSONS:

1) According to Table Five, past IDP scores as a predictor of future ones are most accurate for

defensive backs, then linebackers, then linemen. That surprised me a bit as I expected linebackers

to be best since tackles (their main source of points) tend to be among the most stable stats. It’s

not a huge difference, but defensive back scores are clearly more accurate (bad news for Chris

Hope owners!).

2) Generally speaking, there’s quite a bit to be gained by waiting until week four to make your

decision on a player (I’m thinking in particular about what to do with an underperforming stud).

But, past that, the forecast errors tend to level off. There’s definitely dang little in week 6 that

you didn’t already know in week 5 if we are just talking about the statistical trends.

3) Don’t underestimate the size of those errors in Table Five. Based on the 2004-2006 numbers,

being short by even the smallest percentage shown is more than enough to knock a player

completely out of the top ten. In other words, an error of 23% (the smallest on the table) is a big

deal if you are trying to win your league! By my scoring system, Chris Hope was the #1 db in

2005 with 9.4 points/game. If he earned that through week six and that induced you to start him

every week, bear in mind that a fall of 23% would leave him with 7.2 pts/game, or Will Dempslike

numbers–not terrible, but 13th overall rather than 1st. If your best db is #13 in the league, I

hope that position isn’t your strength!

Thought this might be useful to the Jon Beason owners!