Wednesday, May 13, 2009

Hitting Streaks and Other Stupid Things

I've saved this little rant since I didn't want to rain on the fun, and Zimmerman clearly deserves the national attention that he's getting, but hitting streaks are stupid.

OK, they're not totally stupid. It's kind of a curiosity when a guy can get hits in many games in a row. But the way baseball fans and (mostly) media act like hit streaks are some grand achievements attributable to great skill and not (mostly) randomness is just weird.

Plus, if you are going to focus on streaks like this, why wouldn't you hype "on base" streaks? The guy reached base twice today and scored a run. Why is that regarded as a fruitless day? Joe DiMaggio reached base in 74 straight games, including his record hit-streak. The Baseball Reference "streak finder" goes back to 1954, and in that time Orlando Cabrera has the record with 63 games in 2006. But generally the record is neither recognized as significant nor conclusively known.

Aside from that, I have to say it was pretty dumb when Bruce Bochy intentionally walked Zimmerman in the 7th inning. Runners are on second and third, one out, and the Nationals are leading 2-0. At this moment, the Giants' win expectancy is 13.4%. Bochy walks Zimmerman to get to Josh Willingham and Elijah Dukes. By doing this, the Giants' win expectancy actually falls to 12.9%. The Nationals run expectancy for the inning started at 1.42 with one out and men on second and third. With the bases juiced, that number rose to 1.58.

There are situations where it would make sense to intentionally walk the bases loaded, but that requires a much more significant drop-off in the effectiveness of the batter you're walking and the batters you're choosing to face.

In The Book, Tom Tango, Andy Dolphin and Mitchel Lichtman calculate us the "break even points" for this difference between a hitter and the guys following him and when the IBB starts to be advantageous for the pitching team from a win expectancy point of view. Down 2 in the top of the seventh with runners on second and third, the ratio of the hitters' wOBA and the weighted average "teammate wOBA" of the three men following him must be at least 1.20.

Zimmerman was to be followed by Josh Willingham, Elijah Dukes, and Austin Kearns. Zimmerman is currently sitting on a .430 wOBA, which is extraordinarily high, almost equal to what Manny Ramirez did in 2008, and it's not reasonable to expect him to maintain that going forward. ZIPs just released their updated projections for every player in the league posted at Fangraphs, and they expect him to post a very good but not MVP-caliber .370 wOBA over the rest of the season, which seems reasonable (think Prince Fielder in 2008). Based on that, you'd need the hitters following him to average less than about a .308 wOBA (think Brian Schneider's '08). Willingham, Dukes, and Kearns are a lot better than that. It's not even close.

OK, but you say, Zimmerman's hot. If he's a .430 wOBA hitter so far this season, we should treat him as such. This is kind of silly, playing into a kind of gambler's fallacy that just because the ball has landed on red three in a row, it's sure to land on red again. We know that Zimmerman isn't Manny Ramirez, regardless of what the small sample size says.

But whatever, let's consider this. To make the IBB make sense for a .430 wOBA hitter, you'd need to the weighted average of the three guys following him to be less than .357. The ZIPs "rest of the season" projections for the Willingham/Dukes/Kearns trio gives you a teammate wOBA of exactly .357. So it's basically a wash, and you ruined everyone's fun.

Now, let's do the same calculation using their actual 2009 wOBAs. The Willingham/Dukes/Kearns trio this season combines for a weighted teammate wOBA of .346. The drop-off is mainly because Willingham and Dukes are both off to somewhat slow starts; Hammer in particular is 15 points below is career wOBA. Still that gives us a Zimm-to-teammate wOBA ratio of 1.24 (and remember, we needed a ratio of 1.20 to justify the walk).

So is that it? Have we cracked the Bochy code and proved that he did the right thing after all? Well, only if you completely ignore sample size, assume that Zimm will continue to vastly out-hit his historic performance levels, and Willingham and Dukes will significantly underperform theirs.

And I guess that's what we've learned here: Bruce Bochy (as well as Manny Acta, Dave Jageler, and Zimmerman himself, since they all said the move was a no-brainer) doesn't understand sample size.

15 comments:

Deacon Drake said...

Bad managers always try to make the move that will make them look the smartest if it succeeds, not necessarily the move that increases their chances of relieving the jam.

In this case, he loaded the bases and brought in an extreme ground ball pitcher to force the double-play. Not a bad strategy if the batter is still Zimmerman, who finds double plays like crumbs under the sofa cushion.

However, he walked Zimmerman to get to Willingham, a prolific fly ball hitter. Willingham didn't do anything, but the numbers caught up and Dukes got the 2 out hit.

In that situation, Bochy had few options, and apparently adding that extra runner, already down 2-0, wasn't much of a concern to him.

Steven said...

FYI I just added a bunch more detail to the analysis. This is what I meant to write last night before I fell asleep.

flippin said...

The cost is one percentage point. The benefit? Manager gets to do something that people notice...

Steven said...

That's basically the Frank Robinson school of bad managing, and if you keep up that act over the course of a full season, you can cost your team 2-3 games.

Mike said...

I think Zim and Acta were being gracious. If they had answered the question with a long diatribe about why it might not have been the smart move, someone would have gone back to Bochy with that looking to stir up controversy. I have no idea what they really think about the move, but I think it's a big leap to assume that they don't understand sample sizes from a standard, baseball response designed to avoid showing up Bochy.

Steven said...

yeah, that's probably fair.

CoverageisLacking said...

None of your statistical analysis seems to take into account the possibility of inducing a groundout with one out, to either get a DP or a force at home. Remember that Bochy didn't walk him until the force was off. And presumably your numbers change if Bochy pitches to Zimm with the infield in, which you apparently didn't consider.

Steven said...

CIL--the increased likelihood of a DP and/or force at home are both factored into the win expectancy and run expectancy numbers.

The fallacy you're buying into is that this increased likelihood outweighs the damage done to the hitting team by putting a man on base.

The only way that it becomes advantageous to the hitting team to put the man on is if there's a really major drop-off in the hitting ability of the batters coming to the plate.

Basil said...

In that situation, Bochy had few options, and apparently adding that extra runner, already down 2-0, wasn't much of a concern to him.That's essentially it, I suppose. Perhaps Bochy thought that his only shot of winning, on this particular day anyway, was to hold the Nats to 2 and pray for the best. Basically hail mary mode. Just hope that Willingham hits the ball on the ground.

Granted, that seems a little bit a silly if you're only down 2 right now, you've got 9 outs left, and you're looking at getting to a terrible bullpen at some point. Maybe it looks a little less silly if you do the same in the 8th or especially the 9th.

Steven said...

It's just so much worse for the pitcher though to have the bases loaded. A walk scores a run. An HBP scores a run. You can't really throw inside or try to make the hitter chase. It's not just the difference between a 4-run and 5-run lead.

CoverageisLacking said...

Steven, I understand your point, but I think the fallacy is assuming that you can take statistics that are derived from actual games where actual managers are making actual tactical decisions (assuming you are using past empirical data) and use them to dictate what the "right" decision is in a particular instance. The variables that are baked into those run expectancy and win expectancy numbers are way too many for you to be able to say that a particular decision is "dumb" because it is not consistent with them.

Also, how much is the run expectancy number you've cited for bases loaded situations inflated by more instances of big innings? What is the relative chance of getting out of the inning giving up 0 or 1 run? Those questions would be relevant also if you are going to look at this from a solely statistical standpoint. You do know that can't actually give up .58 runs in an inning, or even .16.

Also, you seem to recognize that you need to look at the particular circumstances of the situation by discussing the wOBAs for the upcoming batters. But you haven't addressed how Zimm's wOBA would be affected by playing the infield in, as I raised in my previous post. Maybe you would take issue with Bochy for playing the infield in also, but that is another decision, and another variable.

My bottom line point is that it is not so easy as just looking at 2 numbers spit out by a computer to dictate what the right decision is in a particular instance. A lot more than that goes into it.

Anonymous said...

For what it's worth, the record for reaching base in consecutive games is 84 by Ted Williams, July 1-September 27, 1949.

Steven said...

citation? That's cool. I've read in several places that the record is "unknown."

Anonymous said...

I found it at a bunch of places, wikipedia mostly, but here's another one: http://www.redsoxdiehard.com/stats/record.html

DeezNats said...

You call the hitting streak stupid, but as a beleagured Nats fan, I'm not about to shit all over pretty much the only positive national publicity we are likely to get this year.