Relaford is estimated as 142, Guerrero is estimated as 216, Relaford is an outlier since he isn't remotely close to a 16P/per IP pitcher, which is one of my assumptions, he's more a 13P/per IP level, judging from his recent game log, that extra 20% accounts for your additional 20% of IP usage over the estimated, there's probably another 10% or more of accuracy in there somewhere, if you tweak the adjusting values rather than using ball park figures, but its good enough for the purpose i use it for in the draft.
Also it assumes DUR is linear, when it clearly isn't, that doesn't matter so much for higher DUR guys, but the low DUR guys, aren't as accurate, because 0 DUR guys recover, and the formula assumes that they're done for the season after one max game.
Oh you meant the other post.