Two-Start Rankings: What’s the Cutoff?

In last week’s “Who is Being Dropped & Why”, I noted the following on Jeremy Hellickson

At home, he faced the Mets and Cubs and they lit him up. I went back over the previous two weeks in the NFBC Main Event to see if owners targeted him for this two-step. In the Main Event, owners typically look two weeks ahead on the two-start pitchers to get a jump on the crowd.

Week: Add Count, Avg Bid
8: 7, $14
9: 5, $5

First, I found it a little funny that the owners hoped to get a jump on the bidding paid more. Second, almost all of the Hellickson shares added were then dropped (12 of 13).

Going back, I looked at his ranking from three free two-start rankings and he was between 27 and 33. He wasn’t the lowest on the list but near to it. There are usually around 35 pitchers making two-starts in a given week. Were owners hoping for a turnaround from a pitcher who had a 5.00+ ERA with matching ERA estimators?

I decided to find out what should have been the right move when using weekly two-start rankings. I asked one of my podcast partners, Rob Silver, to provide the weekly values for all pitcher in which he uses in Baseball Prospectus column, “The Weekly Game”. His values are created from 15-team NFBC Main Event leagues.

I took the average ranking from the three websites and matched them up with Silver’s weekly results. One immediately observation was the number of pitchers who didn’t hit the two-starts via injury or bad weather. There was a heavy regression to $0. Also, I know some weeks will have more or fewer legit options and since I only found time to match three weeks of data making the final results are a little suspect and could change with more data. I was just hoping to create a starting point in the discussion.

With all the conditions out of the way, here is a plot of the data and table of showing the average return at various cutoffs.

The variance in expected versus actual outcome is crazy. It’s about a $60 range in values putting the r-squared is only at 0.17. Even with the variation, there is definitely an observable a trend.

The best fit line crosses the x-axis near 22 so that is why the division on the table was set there. It seems like a ranking of 20 or better is a viable 2-start option for a positive return.

Getting back to the original question on Hellickson, the average return for him was supposed to be around -$9.9 using the above analysis information. He returned -$31.

While there still may be some use using high strikeout pitchers to accumulate counting stats during the two-start weeks, owners should focus on just the top-20 or so pitchers each week to get any positive return.