The first test of the Official Goatrider Predict-a-tron!
First off, just a little dedication - Ed in the Burg, this post is for you.
This isn't QUITE ready for prime time yet, so what you're seeing here is a stripped down presentation of my single-game prediction machine, the Predict-a-Tron. (By "machine" I of course mean "spreadsheet.") The internals aren't quite ready to publish, either. (Sure, it all WORKS, but it's ugly as sin and not well documented.) So consider this a trial run. (And at some point, yes, I will stop calling it the Predict-A-Tron. It's almost 1 in the morning and I'm punchdrunk off the spreadsheet fumes.)
First what I did was I modeled each team's expected runs scored per game. I took the Hardball Times Marcels projections, which were just updated today. I guestimated a likely Cubs lineup - Lord only knows what order that Lou will use tomorrow - and used the Pirates lineup from today. Then I took the projections and rated them out to one game's worth.
Essentially, I figured out who was batting where in the lineup - the leadoff hitter is expected to take 12.2% of his team's plate appearances per game, while the number eight hitter takes 10.2% of his team's PAs. Using that, I calculated each team's OBP with a weighted average, and used that to figure out how many PAs each team would consume per game. Then each hitter's stats were prorated out to that number of PAs.
After that, I summed everything up and calculated team Runs Scored using BaseRuns. Given those lineups, the Cubs are expected to score 5.37 runs per game against an average pitcher, and the Pirates 4.74 runs per game
Of course, we're not dealing with average pitchers, are we? I took each pitcher's stats from this year and fed them into a custom version of BaseRuns I developed to predict future RA. (I have a deep and abiding hatred for distinguishing between earned and unearned runs. )
Jason Marquis sports a mediocre projected 4.95 RA. (Average RA in the NL this year is 4.50.) He's still better than Zach Duke, of the wonderous projected 5.09 RA.
From there, I calculated each team's expected win percentage against the average team, using Pythagorean win expectation - .536 for the Cubs, .468 for the Pirates. I added in home field advantage for the Pirates and used the log5 method, and came up with a 52.8% chance of a Cubs victory
Future refinements are possible - I hope to have a version that incorporates platoon splits at some point. If I have the time and energy, I'll try to start feeding this info to Kurt for the series preview.This really works better for a short series than a single game - after all, a 52.8% chance of winning isn't worth betting on once you figure in the vig unless you're getting absurd odds. Consider this a toy - a sophisticated one, but a toy nonetheless.