Saves Stat Pool

[zEli173]

I've been wondering about the best way to set the stats pool for saves, and I'd like to throw my thoughts out for discussion. Pardon me if this has been covered before.

When we calculate value based on replacement level we want to find the share of the accumulated stat pool that each player will earn. We understand that many strikeouts, runs, rbi's, etc. will be compiled by unrostered players. Therefore, for most categories we properly accept that the size stat pool is roughly equal to just the contributions of the players above replacement level.

But that strikes me as perhaps wrong when it comes to saves because very few are compiled by pitchers who are unrostered. Unlike other categories, a very high percentage of saves in the majors wind up being accumulated in a mixed fantasy league since nearly any free agent pitcher who starts earning saves is quickly rostered (unless the league severly restricts transactions). I can only guess at the numbers -- but maybe 80%, maybe 85%, maybe 90% of all saves in the majors are accumulated in an average mixed league. Whatever the correct percentage, it figures to be much higher than the percentage of other categories that are captured by a fantasy league.

This is maybe not so much an issue in single universe (or other deep leagues) where most middle relievers who are projected to get some saves are above replacement level and so their numbers are included in the stat pool calculation. But in a shallow league those MR's are projected to have negative value and as a result the stat pool is short a number of saves that are nevertheless likely to be accumulated in that league, or any other league. As a result closers are given credit for contributing to an artificially small pool of saves and their projected values are correspondingly inflated.

Consider whether instead the stat pool for saves should be fixed at a historical percentage of saves captured in a fantasy league of a given size.

[Todd Zola]

We use a slightly different means of calculating replacement level which is described in the essays accompanying the tools. That said, I do a slight internal alteration to replacement wins and saves based on the shallower pool. Choosing the MIX option automatically kicks in this on the CVRC.

Down the line, I an going to rewrite the code to allow the user to use their own replacement level, much like you can do to switch the baseline for ratio stats.