devil 2, cardinals 0
these are the games that try SABRgeeks’ souls — 10 hits, five walks, but only one run. by rights the cardinals should have tallied four times yesterday, the braves but once; and whether we credit smoltz for pitching with grit or blame the st louis hitters for swinging impatiently, something seems to have gotten lost in the translation between statistics and outcomes. the cards outhit the braves and outpitched them; therefore they won the game, right?
well no, they didn’t; nor is it necessarily true that anything got lost in the translation. run-scoring models are not deterministic; they only project probabilities. some days 10 hits and 5 walks gets you 1 run, and some days it gets you 7 — four is the fat part of the bell curve, and it’s all supposed to even out over time.
but does it? how often does the scoreboard get it "wrong"? if, in every game, both teams scored the number of runs they "should," how many outcomes would be reversed? for the cardinals and braves, it has happened twice in two days: the cardinals "should" have won saturday 4-3 and yesterday 4-1. on friday night the teams got it right: each side hit its expected run output on the nose, and perfect justice reigned on the scoreboard as it does in heaven.
thankfully heaven doesn’t control all the outcomes; that would make for an incredibly boring sport. but how often does the devil overturn the outcome that heaven ordains? one game in 10? one in 3?
only one way to find out . . . . i’m a-gonna track, season-long, the cardinals’ won-loss record per heaven. it’ll take me a few days to go back through all the april boxes and get up to date, but once i get current i’ll keep it so for the rest of the year — despite having many, many more important things i ought to be spending my time on. (an example of the devil’s messin’ w/ my outcomes. . . . ) i’ll use three run-scoring models to determine how heaven intended each game to turn out:
1. bill james’ runs created tech-1
2. tangotiger’s linear weights ratio
3. a brand-new model introduced at lookout landing that correlates runs scored to OPS: call it "OPS runs"
those of you who want the formulas should click on the links directly above; i’m not going to get into them here. for each game, i’ll plug the raw data — hits, walks, total bases, steals, etc etc — into all three formulas to derive expected run outputs for both the cardinals and their opponent. if all three formulas agree that the devil did his business and tipped the win to the "wrong" team, i’ll so note — but if only one or two of the formulas conflicts with reality, reality trumps. at the end of the year we’ll see how often it happened and whether the cards came out ahead or behind on the deal.
as for now, we can say the devil owes st louis a couple of games — the two he stole from them saturday and sunday. and with that s.o.b. afoot, what SABRgeek’s soul wouldn’t suffer some pangs?
well no, they didn’t; nor is it necessarily true that anything got lost in the translation. run-scoring models are not deterministic; they only project probabilities. some days 10 hits and 5 walks gets you 1 run, and some days it gets you 7 — four is the fat part of the bell curve, and it’s all supposed to even out over time.
but does it? how often does the scoreboard get it "wrong"? if, in every game, both teams scored the number of runs they "should," how many outcomes would be reversed? for the cardinals and braves, it has happened twice in two days: the cardinals "should" have won saturday 4-3 and yesterday 4-1. on friday night the teams got it right: each side hit its expected run output on the nose, and perfect justice reigned on the scoreboard as it does in heaven.
thankfully heaven doesn’t control all the outcomes; that would make for an incredibly boring sport. but how often does the devil overturn the outcome that heaven ordains? one game in 10? one in 3?
only one way to find out . . . . i’m a-gonna track, season-long, the cardinals’ won-loss record per heaven. it’ll take me a few days to go back through all the april boxes and get up to date, but once i get current i’ll keep it so for the rest of the year — despite having many, many more important things i ought to be spending my time on. (an example of the devil’s messin’ w/ my outcomes. . . . ) i’ll use three run-scoring models to determine how heaven intended each game to turn out:
1. bill james’ runs created tech-1
2. tangotiger’s linear weights ratio
3. a brand-new model introduced at lookout landing that correlates runs scored to OPS: call it "OPS runs"
those of you who want the formulas should click on the links directly above; i’m not going to get into them here. for each game, i’ll plug the raw data — hits, walks, total bases, steals, etc etc — into all three formulas to derive expected run outputs for both the cardinals and their opponent. if all three formulas agree that the devil did his business and tipped the win to the "wrong" team, i’ll so note — but if only one or two of the formulas conflicts with reality, reality trumps. at the end of the year we’ll see how often it happened and whether the cards came out ahead or behind on the deal.
as for now, we can say the devil owes st louis a couple of games — the two he stole from them saturday and sunday. and with that s.o.b. afoot, what SABRgeek’s soul wouldn’t suffer some pangs?
posted by L Boros at 9:22 AM
<< Home