Bullpen Usage Can Skew Overall Run Differential

I'm in a '70s mystery league and have a record of 62-49 after 111 games. The expected won-loss record with my +27 run differential shows I should only be 58-53. But here's the thing: the run differential is skewed because I've lost five games by more than 10 runs and a total of 12 games by at least five runs. The reason for this is I set my pitching preferences to have my two worst relievers mop up when I get down big early and avoid wasting my four best relievers in a blowout. In real life I'm not sure I could let two guys continue to get shelled, but it saves my best relievers to be available and fresh for the next few games. My attitude is a big loss is just a loss, so be ready to fight another day at maximum strength. I'd be interested to hear if others work their bullpens in a similar way.

Yes. I know a lot of guys have their 2nd or 3rd best pitcher in mopup, never giving up on a game. I consider 5 runs down a loss, and like you, want to be loaded for the next game.

In theory real MLB teams do the same thing. So theoretically the formula should account for that. Like many saber stats though always have to be careful about placing more value on the saber stat than on real life events.

Also, the formula for expected wins is actually a moving target. If you read the history of the formula it has changed a couple times over the years. One of the factors changing is the exponent. The original formula was just s simple square of the RS, etc while the last one I read about used 1.83 as the exponent. I read in one of the Baseball Prospectus articles one reason for the change was the discovery that as scoring overall increased the correlation of expected wins with actual wins diminished. Lowering the exponent brought it more in line. The natural conclusion is that the correct formula for any given league is going to depend somewhat on the ballpark makeup and salary cap among other things. More hitter parks at higher caps may need a lower exponent due to the higher scoring. More pitching parks at lower caps may result in lower overall runs scored league wide and thus might warrant an increas of the exponent. The formula used by the site however is fixed and thus to some extent inaccurate for all leagues.