here earlier) on the likelihood that runners on base would score, in Detroit and el..."/> here earlier) on the likelihood that runners on base would score, in Detroit and el..."/>

Does Speed or Power Score Runs?


After doing a little work (posted here earlier) on the likelihood that runners on base would score, in Detroit and elsewhere, I decided the dig a little deeper. The question is this: what increases the odds that a runner will score?

My data set consists of lineup positions (each of which is a sort of composite player) for each team in the big leagues last year – adding up to 270 observations. For each ‘player’ I had the number of times he reached base (without circling the bases with a home run) and the percentage of those times that he scored. I also had the percentage of those times on base due to a double or a triple, the percentage of times on base when the runner stole a base and the batting averages and ISO numbers for the batters 1, 2 and 3 slots behind him. In theory, the percentage of times on base in which the runner scored should be a function of all of those things: starting base, speed and the potency of the bats attempting to drive him in.  So all that was left was to stick all that info in some statistical software and click a button to run a linear regression to estimate how important each factor was.

My initial assumptions would be that ‘initial base’ (in other words, doubles and triples) would be more important than steals or speed and that the batter immediately following would be the most important one. As it turns out, I was completely wrong. All of those things DO matter, but not in the expected proportions.

I’ll put my results in a table below, but it will take a bit of explanation:

Again, the stat I’m trying to predict is the player’s plate rate (assuming that he doesn’t get caught stealing, which I tried to control for) and the average here was in the neighborhood of 28% (or 0.28). If we see that the impact of the batting average of the guy one spot after (AVG_1) is 0.241 that would mean that a batting average 100 points higher (comparing, say, Miguel Cabrera and Ryan Raburn) would cause the runners plate rate to jump from 28% to 30.41%.

As it turns out – unless the data is misleading me here – the batter immediately following isn’t even the most important one.  That’s the guy two down in the order.  And while the batter three down is less important than the batter one down, the difference isn’t all that large.  One thing that does stick out is that while ISO is pretty important for the batters one and two down (since it may take a home run to drive in a batter from first all by yourself) it doesn’t matter much for the guy three down or presumably anyone coming after.  So batting average matters more than power (and Dombrowski has built this team for average rather than walks and homers) and it takes lineup depth to get anything done.  I don’t think that’s a weakness for Detroit, though OBP in slots 1 and 2 could be.

While those three batters all count quite a great deal in driving in runs, it doesn’t look like they matter as much as the guys legs do.  The percentage of the time that the guy started on second base has a positive impact on his chances to score, but not by very much.  With an impact like the one estimated, a batter could double 100% of the time and see his plate rate go up from about 28% to about 30%.  That’s not much, certainly a bit less than I would have expected.  On the other hand, the impact of triples and the impact of steals are quite large.

Is that because of the extra base?  Certainly, a runner on third can score on a sacrifice and often does – that must count for some of the difference between 2B percent and 3B percent – but can it explain such a huge gap?  I should mention, of course, that though these results predict that a player who only tripled would score about 100% of the time nobody actually comes close to that.  In fact, I don’t think anybody actually tripled in more than 10% of those times on base.  Nonetheless, the difference between a guy who never tripled and a guy who did so in 10% of his times on base is a plate rate jump from 27% to 34.8%.  Which brings me to stolen bases:  the vast majority of the time that a player steals a base, he’s stealing second.  That extra base might mean a great deal – as far as winning the game – if one run would make all the difference.  But… there shouldn’t be much inherent difference between getting to second by stealing it and simply starting there on a double, but the data says otherwise.  The only possible reason is that while simply being on second might not matter all that much, the runners speed matters a great deal – and both stolen bases and triples are good proxies for speed.

One thing the Tigers do not have (though they can slug) is great team speed, so perhaps this doesn’t strike an optimistic note for the coming season…