THE BOOK--Playing The Percentages In Baseball

That’s really cool.

Where does the 3.5 in step # 2 come from?

Lev: It’s the SB + CS.  If SB=2.5 and CS=1, then the percentage is SB/(SB+CS), so 2.5/(2.5+1).

I’ve mentioned this before, but we often forget that one of the not insignificant benefits of the SB attempt is staying out of the DP, and getting the extra base on a hit more often (and of course a little negative from the extra DP’s on line drives and occasionally a short fly ball).

Can anyone come up with an estimate of the value of the above and how much it changes the BE point?

Also, I often have to remind people that you CANNOT use a player’s or team’s SB/CS numbers.  The Sb % ends up being too high because there are a significant number of pick-offs that are not included in those numbers.

You need to use SB, CS, pick-offs, balks, pick-off errors, and stolen base throwing and catching errors, in order to analyze the stolen base attempt for a particular player, team, or pitcher (or league, or whatever).  For example, a runner sometimes gets credited with a CS or a PO (picked-off) even if he is safe but someone makes an error and if not for the error, he would have been out.

Also, stolen base attempts, or at least a “green light” (you can’t really tell a runner to steal - he has to get a good jump, etc.), should be somewhat a mechanical thing for teams.  They should simply be plugging 4 things into a “computer” or formula and coming up with a steal or no steal.  Those four things are the speed/basestealing ability of the runner, the pitcher’s time to home plate, the catcher’s time from reception to the base, and the overall situation (inning, score, run environment, etc.).  Obviously the count is important as well, for 2 reasons:  One, it changes the BE point, I think, and two, the chances of an offspeed pitch or one out of the strike zone, changes the times.

Even if a team did some kind of a crude calculation (they could use “rules of thumb") based on these 4 things (the 4th “thing” is actually a bunch of things), they would be much better off than doing what they do now, which is just going by “instinct,” as usual, and often running when they are below the BE point or not running when they are above it.  Again, this is one of those things, among many, that a human being can’t figure out without “help,” unless the analysis/result is obvious.

In the last sentence about playing small ball against a great pitcher, Tango is basically repeating the axiom that outs are more valuable in a higher run scoring environment (therefore, it’s more acceptable to “give away” outs in lower run scoring environments).

So this is my question: Why? I’m not questioning the math or the answer, but there is some kind of logical issue where my brain just can’t quite comprehend why, when you’re expected to score fewer runs, it’s BETTER to do things that could cause more outs. Can somebody explain this with something other than “because the math says so”? I’m not opposed to accepting it based on just the math, but I have a feeling it’ll be one of those answers with which I’m never satisfied.

The most basic way to put it is this way:
the run and win value of the SB remains constant, regardless of run environment (within reason).  Somewhere around +.02 wins per SB.

the run and win value of the CS gets larger, the higher the run environment (the out value is directly correlated to the run environment); if you expect few runs, then the cost of the out becomes more negligible; you do not want to make an out, if the next 3 batters are Ortiz, Manny, and Pujols.

So, if in one environment you gain +.02 wins on the SB, but lose .03 on the CS (Pedro) and in another environmenet you gain +.02 on the SB, but lose .06 on the CS (Coors-induced Ortiz et al), then you can see how you better goshdarn be stealing at a high rate in front of a high run environment.

Another way to put it, which ties in to Tango’s Ortiz, Manny, and Pujols example, is if all you are going to get out of your batters are a few singles, some ground outs, and an occasional sac fly (I am exaggerating of course), then you might as well try moving up on the bases any way you can (so that batted ball outs can drive home the runner), and if you make an out on the bases, what is the difference if the next guy is probably going to make out anyway.  That is the “logic,” using exaggeration to illustrate the point.

But isn’t the most important aspect of small ball vs Pedro that the HR retains most of its value. So sure, try a steal or H&R against a stud pitcher, but maybe the best way to beat him is to have HR-heavy offense. And you can’t call that small ball.

I forgot to mention that tango’s shortcut BE calculation is pretty cool. Sometimes, simpler is better (hey MGL )

We’re talking about two different things, I think.  One is what has more or less relative value in a low run environment and the other is what has more or less relative value against a certain type of pitcher.  I would think that a HR would have more relative value against a pitcher who allows lots of baserunners and less relative value versus a pitcher who doesn’t, like most stud pitchers, no?  Although it depends on what makes him a stud pitcher. If it is because he does not allow a lot of HR’s mainly, then a HR against him is going to be worth more (than a pitcher who is a stud because he does not allow lots of baserunners), although obviously you are not going to hit a lot of HR’s against him.

Regardless, I don’t think that you can get much of an edge against any type of pitcher by “loading up” (if that were possible) with any particular types of hitters.  Maybe a little, but not much.

Use this to know how much value someone has relative to someone else against different run environments:

http://www.tangotiger.net/customlwts.html

DS is certainly right that the HR doesn’t change much with the run environment.  The biggest changes seem to be with the single, walk, and out (at a glance).

So I guess it pays to load up on high OBP, no power guys (relatively speaking) in low run environments.  Certainly, you have to change the formula you use to evaluate player worth depending on the run environment.  Another good reason why a formula like OPS is so crappy for evaluating players.  At least with linear weights, you can choose your own weights.

Tango, would it make that much of a difference how the run environment was different?  For example, if the run environment were high simply because the HR rate were high (like in old Coors Field), would the lwt values be around same as if the run environment were high because much fewer outs were made, but not necessarily more HR?  And other combinations?  IOW, is it (the lwt values) more sensitive to simply the change in runs per out or more sensitive to the individual component rates?

Ask and ye shall receive:

http://tangotiger.net/markov.html

Put in whatever numbers you want, and you get the LWTS values at the bottom of the next page.

Tango,

re: http://www.tangotiger.net/customlwts.html

Are the signs incorrect for PickoffError, and OtherAdvance?

Hint for Markov: ONLY change the AB value to change the run envionment IF you want to keep the H/HR/BB proportions exactly the same.

You MIGHT also want to change the K value to keep the K/out proportion the same.

***

Kodiak/15:
Signs are correct.  You can see more data here:
http://www.tangotiger.net/bsrexpl.html

So much good stuff I forgot existed!  Why are the signs negative again for those events?

There was only 30 events total of pickofferror in the Retro years (which has subsequently been deprecated). So, most ended with an out, but some didn’t.

Other advance were probably unclassified outs on base.

***

That Markov program should be on everyone’s favorite. It explains everything, plus the source code is right there for anyone to use.

How does a pickoff error end in an out?  It is possible of course, but it would be rare.  Unless retrosheet made a mistake in the classification.

Yeah, there were only 30 at the time I downloaded the files, and now there are zero.  So, I would just ignore that altogether.

here’s my answer to JD
forget the math… straight logic…
in a low scoring environment, outs are more probable… more outs per run scored… therefore, its easy to give more value to risky actions…
If your team averages 10 runs/game, stealing 3rd makes little sense; why risk such a behavior, when your team is likely to drive you in
However, if your team averages 2 runs/game, you have no confidence that you’ll score from 2nd on a base hit… you’d hate it if the one time in the game you get two hits in one inning, your lead runner gets nailed by a strong outfielder’s throw…
So despite the high risk, you attempt to steal 3rd, knowing that if you can get to 3rd with less than 2 outs, there’s so many more ways to score (past ball, dropped 3rd strike, sac fly, rbi groundout, etc.)
Also, low scoring teams leave lots of men on base, and since having men on base serves no purpose EXCEPT leading to runners scoring, there is no harm in losing runners that are likely to be stranded in the end.

Thanks to Rob Neyer for sending over his crowd here.

It should be remembered that the breakeven point changes ALOT based on the score/inning.  It can be anywhere from 60% to 90%, depending.

I have a good portion of that in The Book, but maybe I should try to come up with a rule of thumb that includes inning and score too.

On the subject of small ball vs. good pitchers:

Is it also correct to say that you should play small ball when YOU have a great pitcher pitching?  [Is this captured in the run environment, or does that just refer to the offensive team’s RunEnv?]

If you have Santana pitching for you, isn’t it better to play for an increased chance of 1 run than to take a chance on multiple runs since Santana will give up so few himself?  A blowout is an improperly leveraged use of Santana.

No, I don’t think it works that way.  The win value of the out, relative to the base, is predominantly driven by your own hitting environment, not your pitching environment.

I could be wrong, and the answer is as easy as running some tests.