THE BOOK--Playing The Percentages In Baseball

Unless I completely misread his article, he does not have the same hitter in each spot.  He has a hitter that has had his rates increased by the same amount, over his baseline hitter in each spot.

His baseline leadoff hitter hits 283/375/392; his baseline cleanup hitter hits 274/365/513.  The cleanup hitter is a markedly superior batter.

The great leadoff hitter bats 381/481/625; the great cleanup hitter bats 372/470/751.

Anyway, that shouldn’t be a big deal because he’s looking at the difference between the two.  But I think his 10 run conclusion is completely unjustified.  He reaches it by looking at (change in team runs)/(PA for lineup slot).  PA are not the proper denominator for a theoretical team runs above average figure.  I would argue that the correct denominator in this case is the player’s PA times the team out rate (this is what David Smyth and I used to call O+).

The leadoff batter’s rate by the aformentioned stat is .1505.  The cleanup hitter’s is .1687.  So the cleanup hitter is 12% more effective.  A standard ERP/Out calculation based on the rates shows the cleanup hitter as 12.3% more effective.

Summary of all that rambling: I think Hanrahan’s study probably is more in line with your more conservative figure than he knows.

Right, he is not putting in the same hitter, but adding the same amout of extra goodness.  My bad for not explaining.

The key point is that we need to resolve to the extra runs at the team level, which is why I said it should come out at +74 or so, however you wish to value the extra goodness.

If the standard linear weights says that the extra goodness is worth +.11 runs per PA, then it would be unfair to count the number of PA higher for the leadoff hitter than the #5 hitter.  The end-result is that that extra goodness gives your team +74 runs whether it was added to the leadoff hitter or the #5 hitter.

That is takes 72 more PA for that +74 runs of extra goodness to materialize by putting it in the #1 spot than the #5 spot is really irrelevant.

I agree that the 10% figure is off. 

He was also totally off base in comparing Raines and Dawson.  If he wants to chop off 2% of Raines’ +500 runs or whatever he has, that’s fine (he was not 100% a leadoff hitter).

It seems that Hanrahan is using the same run estimator formula to evaluate players no matter where they bat. But I think that a batter’s run value, although it can be based on OBP and SLG, is also affected by their lineup slot. Here is something I wrote on this. It was specifically about leadoff men and their value does change, although not alot, if you factor in where they bat

http://www.beyondtheboxscore.com/story/2006/9/18/234859/360

As noted, I published the linear weights by lineup slot in The Book.  Table 51 I think.

Non-owners can find it using “Look Inside” on Amazon.

I would think that hitting leadoff would not reduce the value of OBP and SLG equally.  In fact, OBP should be more valuable (even apart from the extra PA):  1) the team’s best hitters follow you in the lineup, and 2) at least once a game you have the opportunity to get on base with zero outs.  Each point of SLG, conversely, must be much less valuable than average.  By failing to separate the two, I think Hanrahan likely overstates his case by a lot.  Great leadoff hitters do not have .600+ SLG. So while it’s true that such talent would be wasted in the leadoff slot, that’s not terribly relevant to evaluating players like Raines and Henderson. Conversely, the OPB of a cleanup hitter will be worth a bit less, as he will be stranded more often.

test (have tried twice to post, and gotten nowhere)

Good points, all. A few comments to your comments on my research:

As Cy and Guy mentioned, yes different spots would have varying effectiveness (OBP/SLG), but I intentionally did NOT do this, because the point was to show that using standard (LW/RC/BsR) metrics (which is what most everyone uses when comparing players), leadoff hitters are overrated.

Tom T, I wish I had seen your table 51 in the Book before I wrote my paper up. I’m familiar with much of your work but had not purchased that one. Your numbers do seem to confirm my conclusion, though, don’t they?: Singles are worth 7-8% less for #1 hitters than for 3-4-5 guys, home runs worth MUCH less, outs 10% less. Walks are the only thing that is valued similarly. So why, if I can ask the group, would the 10% figure be way off? I agree it’s only 5% compared to the 2nd, and 6th-9th hitters (per PA!), but closer to 10% compared to the cleanup batter. While the leadoff man gets up more than the #4 man, its not 10% more, hence the #4 is more ‘important’. If indeed the figure IS much less than 10%, then we ought to conclude the best hitters should bat 1 and 2!

On Raines/Dawson, as I stated in the paper, Raines is still obviously the better player, but I think it is by less than typical sabermetric analysis. If you take away 6-7% of the value of Timmy’s “above averageness”, instead of being 600ish runs better than average, he may only be 560, which is closer to Dawson’s 300ish above average.

One of the good objections mentioned is that I raised the ‘goodness’ in such a way that the uber-impproved man would not really bat leadoff (they don’t hit 40 home runs).

I ran new simulations, and instead of increasing 20W, 20S, and 20HR per 550 PA, I added 10W and 50S.

lineup with super #1 hitter scored an extra 50.6 runs per year.

lineup with super #4 hitter scored an extra 52.9 runs per year.

That is 4% more runs.... even tho the #4 man gets up 8% less often.

Tom H. - Its great of you to respond on this site.  I emailed you yesterday about the Star v1.2 lineup simulation that you used in your study.  I was trying to locate a copy to replicate your study.  I have looked at the same question using linear weights with a different methodology and come to the opposite conclusion that you did.  Linear weights seems to slightly undervalue leadoff hitters.  I wanted the simulation that you used to see if there was some way to resolve these different conclusions.

Even if leadoff hitters create 5% fewer runs than a cleanup hitter with comparable “goodness” (in terms of rate stats), that leaves the question of how much weight we give that in evaluating the value of these players.  I’m inclined to give it very little.  To the extent there’s a “leadoff penalty” in terms of reduced production, the team must pay that penalty to some extent no matter who hits there.  If Raines and Rickey had hit second, their teams would have put someone else in the leadoff position whose production would suffer.  The fact is the best thing for the team is to put a high-OBP player there (leveraging the high # of PAs), while “wasting” as little power as possible.  R&R obviously both did that extremely well.

Moreover, their ability to get on base is one reason for the success of #3 and cleanup hitters—hitting with men on base gives a hitter an advantage (especially LH hitters, I think).  Some of Dawson’s success can be attributed to Raines being on base—but that is captured in Dawson’s rate stats, not Raines’. 

As we’ve discussed with relievers, the kind of leverage enjoyed by middle-of-the-order hitters is to some extent a team asset—it’s created by the team as a whole, not the player who uses it.  So it’s tricky to decide how much credit the individual player should get.  Maybe chaining is the right answer here.  But I’m not sure it’s worth the trouble.

Guy, I think chaining is a good way to look at it. I’d suggest lineup simulators as a practical means of evaluating any ‘chaining’ effect someone might theoretically derive.

Peter, did you get my Star simulator attachment?

No, I haven’t received an email from you.  When did you send it?

Tom H. - I don’t know whether you have had time to try again but, I still haven’t received an email from you.  You might try sending a simple email without the attachment first and see if that makes it through OK.  Thanks in advance for trying to get the information to me.