As I talked about last week, my newest obsession is the Pitch Count estimator I built in Microsoft Excel, as well as the Game Score calculator I built into the same file. My goal is to have people post the pitch counts of games when they find them, so we can try to figure out if this method has validity, and how accurate it is. It was then pointed out to me that milb gives pitch counts for all AAA games. While we don’t have many pitching prospects in AAA, we do have JA Happ, so I decided to plug all of his starts into the formula and get the estimates, and then compare the estimates to his actual pitch counts, which I hoped would give me an idea of how accurate the model is. So, lets get to it…

First, a few notes. As I mentioned in the previous post, the estimator formula is the one that tangotiger created. It factors in IP, BB, H and K, and has been back tested through time at the major league level. He talks about how it might be effected by high strikeout/walk pitchers here. A brief summary

Now, if you are only interested at the point where the current pitchers are, then the previously published basic pitch count estimator (3.3 x BIP + 4.8 x K + 5.5 x BB) will do the job. The basic pitch count estimator matches the model when the BIP rate is between 50% and 80%.

BIP = Ball In Play. The average across time is 3.3 pitches per BIP, the other multipliers involve strikeouts and walks. Now that you know the formula, lets get to the results. Since I have every recorded game this season for JA Happ, he’s the first test subject. I’ve looked at the formula a number of times, and I think it’s correct. If its not, well this whole thing is rendered useless, but I’m pretty sure its accurate. Anyway, here is Happ’s chart

The chart is self explanatory. If it doesn’t fit into your screen properly or you need a larger view, click here.

Happ has made 16 starts this season, throwing a total of 1,618 pitches. The estimator has him at 1,568 pitches, or 50 less than he’s actually thrown. I’ve chosen to round UP in every case, so 89.3 = 90. Averaged across 16 starts, thats an error level of about 3 pitches per start. In 10 of his starts, the estimator was off by 3 pitches or less. There are 5 starts with a difference of more than 12 pitches, most of them underestimating his count, while one of them (May 26th), greatly overestimated his pitch count. The games with large spreads are the ones I’m most interested in. I plan to do a bit more work into this to try and figure out exactly where the discrepancies are happening. It could be that Happ doesn’t fit the standard bill (which is where the 3.3 pitches per BIP comes from), and that he’s more of an outlier. So far, I have two recorded pitch counts for Tyson Brummett

6/19/08: Estimated 125, Actual 122 (+3)
6/24/08: Estimated 110, Actual 104 (+6)

So we have one close, one not. Obviously the data is going to be inconclusive until we get more info and have a much larger sample. Nevertheless, this is really interesting to me, and something worth looking at going forward. As I said before, the more games we get official counts from, the better. So if you come across any pitch counts for any of our starting prospects at AA or lower, please post them on the Pitch Count Log page at the top of the site. Thanks.