Originally Posted by
ScreaminPain
Yes, I use Base Runs for my model and I agree it is the preferable for predictive purposes.
Ok, here is what I have so far...attempting to use Total Bases (TB's) as a predictive value. I used the "Last 3 games" value for each pitcher in a matchup, subtracting the TB's from the Innings pitched (IP). I then used an arbitrary differential of + or - 9 to separate one from another. I then filtered only teams whose last 7 gm batting avg, was higher than their season avg. I then filtered the bullpen ranking, stating the selected team must have a ranking higher than their opponent. So, to capsulize:
1. Pitchers TB's to IP ratio must be better than opponent by at least 9
2. Bullpen ranking must be higher than opponents ranking
3. Last 7 BA must be higher than overall season Avg.
What you end up with is a snapshot of a better starting pitcher, better team bullpen and a team thats hitting better, at present, than their overall average.
In August there have been only 6 plays, using the above stated criteria....resulting in a record of 6-2, with an average lay of -146. Promising, but not earth shaking, especially with so few plays.
I'm currently going thru July's games using the same set of data and filters. We'll see how that comes out.