Monday, December 20, 2010

Rox Talk - Pitches Thrown

A Pitch Thrown
In past posts, I have shown how many pitches thrown in an inning typically lead to a run. Data collected from 2008 - 2010 seasons (Rox pitchers and Rox opponent pitchers) indicate that usually 23 pitches lead to a run. Using a best fit curve of the data (run rate calculated as a ratio of total runs scored in an inning per instances of that pitch count), it would indicate in the model that as the pitches go up, the amount of runs increase.


Although following this logic out, eventually you would be scoring runs without throwing a pitch!Either way it is interesting to see what the break point is but then how quickly runs compared to the number of pitches and how it adds up after the initial run is scored.

One interesting thing to consider when looking at this data is comparing pitchers' runs allowed with pitches thrown in an inning and comparing to what would be expected. For instance, three years of data indicate that if a pitcher throws 13 pitches in an inning then the runs expected for this would be 0.121. Add up the season totals and the table below shows what the 2010 Rox starters did.

Of the eight starters, only two starters had less runs then their expected runs. Jimenez, I can believe, because he throws a lot of pitches (and was saved by a lot of double plays...). Hammel is the troubling one because the data would suggest that he gives up a lot of easy runs which is one way of saying he throws very few pitches but gives up a bunch of runs on low pitch counts. Which is better? Hard to say will have to look back at previous years...

Independent Innings?
One question that comes to mind is if pitching (or offense) is extremely efficient or inefficient in one inning does it effect the next inning? Note the graph below shows pitches in inning x (y-axis) followed by pitches thrown in inning x+1 (x-axis). This is independent of pitcher, so some of the data below could be one pitcher followed by a different pitcher in the following inning. Also it should be noted that 70% of all inning x was between 9 - 20 pitches and in inning x+1 70% was between 8 - 19 pitches. And thus this overlap (meaning inning x had a pitch between 9 - 20 followed by an inning x+1 of between 8 - 19) accounts for 49% of the back to back innings.
Some other observations are that if you pitch inning x between 20 - 68 then 20% of pitches in inning x+1 was less than 20 pitches. If you pitch inning x between 3 - 20 then 17% of pitches in inning x+1 were greater than 20 pitches. Pitching inning x greater than 20 followed by inning x+1 greater than 20 amount to about 8%.

Just one other point and that would be that in 3622 of 7708 instances inning x was less than inning x+1 meaning that generally less pitches are thrown in inning X+1 (53% vs 47%). My guess is that with additional data and data that includes other teams that this number would generally get closer to 50/50 and that generally inning x and inning x+1 are independent. Just for kicks in 2010, Jimenez had 92 instances (out of 194) where he threw more pitches after inning x. It would seem in 2010 he seemed to throw less in inning x+1 then inning x.

Monday, December 13, 2010

Rox Talk - Year in Graphs

The Year in Graphs
Graph 1 shows Rox 2010 winning percentage versus historical (1993 - 2009) winning percentage and pythagorean winning percentage through the different months of the season. What a great start, too bad the rest of the year was average!

Graph 2 shows projected wins versus previous years (2007 - 2009). It was so hopeful midway through the year - then the post All Star hangover, then a late in year surge followed by the awful collapse.

Graph 3 shows the average runs scored per game both home and away. Who loves the humidor? What is with scoring on the road?

Wimisical data for the next three graphs. Home run rates for temperature, wind speed, and wind direction (2007 - 2010, purple points indicate 2010 grey squares is average). Interesting how wind speed doesn't do much for home run rates.