Setting Guidelines For an Under Performance Metric

Most fantasy owners expect drafted players to under and over perform some amount. When a player overperforms, the owner looks and feels great because they “knew” a breakout was coming. Owners hope they didn’t pick too many players on the other end of the spectrum. The underperformers are the ones who drag down a team and owners find as escape goats for a bad season. I’m going to start laying the groundwork to determine a hitter’s disappointment chances.

The first major step in finding a disappointing hitter is to define what is a disappointment. After owning too many fantasy teams over the years, I’ve had my share of disappointments (e.g. Brandon Webb in a 2009 first round) and feel they are just part of the game. This ambivalence doesn’t mean I shouldn’t know the breakout chances. Even small changes in the odds can make a major difference after rostering 20 or more players.

Two types of disappointments exist. The hitters who underperform production or those who miss a bunch of time. And many times, both happen because of injury. With my high disappointment tolerance, I took to Twitter to set the initial let down parameters.

I’ll start with playing time. For it, I asked the following:

If a hitter is projected for 600 PA, gets hurt, & misses time. Minimum PAs he can miss until his season is considered a disappointment?

— Jeff Zimmerman (@jeffwzimmerman) October 26, 2017

Weighing out the averages, my followers believe having just 378 PA (63% of total) is a disappointment. Basically, if the player misses over two months, it’s a letdown.

As for finding the playing time drop, I could have gone a couple different ways on production, so again, off to Twitter. I first asked which stat should I use and wOBA came out the big winner.

I'm still working at some breakout/breakdown metric. If given one metric to measure a player's offensive output, what would you prefer?

— Jeff Zimmerman (@jeffwzimmerman) October 25, 2017

I would have probably used an SGP like measurement but wOBA is probably the best method. The next step was to find out what is an acceptable drop. For the drop, I put out two tweets (the first one had too small of a range I thought) but with just OBP since it is on the same scale as wOBA.

If a hitter is projected for a .350 OBP, how far would it need to drop to seem like a disappointment?

— Jeff Zimmerman (@jeffwzimmerman) October 27, 2017

Try 2, first ? outcome range was too wide

If a hitter is projected for a .350 OBP, how far would it need to drop to be a disappointment?

— Jeff Zimmerman (@jeffwzimmerman) October 27, 2017

The first one ended up with a weighted average of .309 and the second one was at .317. Just for convenience, I went with a 35 point drop (.315).

With both parameters set, I used the 2017 Steamer preseason projections and queried the players meeting at least one of the two drop-off measures. Here’s the list (300 PA minimum projection).

On the surface, the results seem fine. My first question is:

Which hitters should or should not be included in the list?

The biggest misses I see involves projected speedsters, like Odubel Herrera, who didn’t steal as many bases as projected. Anyone else? I am just trying to adjust the baseline.

One other change I could see is:

Should the wOBA and PA changes be combined in some way or remain separate?

If a player misses his plate appearance projection by 200 and wOBA by .030, he won’t get flagged but maybe he would if they were combined. I’m still thinking of some value to use here. Or should they be two separate categories?

After I find a reasonable framework, I can start utilizing previous projections to get a simple disappointment chance for each player. Again, let me know if any suggestions exist for improving the metric as it’s harder to go back and fix it later.