THE BOOK cover
The Unwritten Book is Finally Written!
An in-depth analysis of: The sacrifice bunt, batter/pitcher matchups, the intentional base on balls, optimizing a batting lineup, hot and cold streaks, clutch performance, platooning strategies, and much more.
Read Excerpts & Customer Reviews
If you are a media member and would like a review copy of The Book, please contact Kevin Cuddihy of Potomac Books.

Buy The Book from Amazon

MOST RECENT ARTICLES
MAIL : You ask | We say

Advanced


THE BOOK--Playing The Percentages In Baseball

<< Back to main

Sunday, November 02, 2008

The Book - High School Edition

By Tangotiger, 10:14 AM

This is NOT any kind of announcement.  But, someone posted a mailbag question that I think deserves its own thread, and allow for discussion among everyone.  A tangential question could also be: what would The Book be at the turn of the 20th century?  Here is the original question in its entirety:


Guys,

I’m a high school baseball coach in California and I have a question I’d love for all three of you to tackle. Even though it might only make a difference of a few runs per season, I’m passionate about trying to manage in the most logical way possible, and I read your blog regularly for insights.

Here’s my question:  In a low-quality high school environment, if you assumed that none of the hitters in the game had the ability to hit home runs, all pitchers are throwing 70-82 mph with average-at-best control and mediocre off-speed pitches, and virtually all runners reached base via singles or walks (at a 1:1 ratio or something near it)...how would you manage?

Would there be any “rules” you’d establish for your players?  Do the sabermetric principles that you discuss on the blog that apply to MLB change slightly when discussing high school? For example, does the “break even” point for stealing bases or taking the extra base change in this environment?  How would you approach or (de?)emphasize bunting at the high school level, assuming no elite players on your team and low-level competition, too?

My initial thoughts are more on the defensive end: I’d want to play aggressively shallow to take away bloop singles and give OFs a chance to prevent runners from taking the extra base.  Should my hitters take more pitches, due to the wildness of the pitchers and their complete inability to hit for power?  I’d love any ideas you can give. Basic topic for reply is “What you do if you coached high school baseball”.

Thank you guys, I am grateful.

#1    MGL      (see all posts) 2008/11/02 (Sun) @ 11:26

The initial, short, and incomplete answer, of course, is that everything changes.  Run scoring in MLB (and most high-level baseball) centers around the basic MLB model we use for most of our analysis.  When the components of run scoring change so dramatically (as you say, few, if any HR’s, lost of walks, lots of ROE’s I assume, probably more GB’s, then the models that determine optimal strategies have to drastically change.  Well, it’s not really the models, but the inputs I guess.  I also assume that in high school, you have a large spread of talent in all areas, for example, some runner cannot get thrown out on an attempted steal and others could not make it more than 25 or 30% of the time.  Anyway, I am about to head to the airport, but I’m sure there will be lots of good input in this thread.  Good question!  For those high school (or college or Little or Pony League, etc.) coaches out there, hopefully if you read enough on sabermetrics, you can use the methodologies that sabermetricians use to determine optimal or at least roughly optimal strategies, and by using your own inputs and making logical inferences, etc., you can answer these questions yourself, rather than us saying, “Yeah, you need to this, this, and that.”


#2    brent      (see all posts) 2008/11/02 (Sun) @ 20:42

My first question would be what the run environment is like? How many runs are scored a game?


#3          (see all posts) 2008/11/02 (Sun) @ 23:06

Right #2 brent… I think the HS coach would need to know things like what the league OBP is.  Knowing that the ratio of singles to walks is 1:1 is not too important, I don’t think.  For example, in a no-power .350 OBP environment, the break-even point on SBs is probably pretty low.  But if it’s a .750 OBP league, stealing is likely a waste of time (and outs) at even a high success rate.


#4    weskelton      (see all posts) 2008/11/03 (Mon) @ 10:45

Echoing the sentiments of brent and Mike, I think it would be very useful to understand the environment in which you play.  Assuming you have the data available, I would plug values into Tango’s Run Modeler at http://www.tangotiger.net/markov.html

This will provide you with custom Run Expectancy charts, Run Values for various events and more.

You’ll probably need to convert your errors into singles due to sub-par fielding.  My recollection is that this was found to be a necessary step when using the tool to evaluate the Israel Baseball League last year.


#5    Tangotiger      (see all posts) 2008/11/03 (Mon) @ 11:42

Right, you want to use the Markov calculator, as the starting point for the analysis.

Let’s come up with an environment for discussion purposes: change the 2B, 3B, HR values to 0, and change the H, BB to 12 and K to 6.  That gives you 10.0 runs per game.

Here’s what the LWTS values generates, compared to what BaseRuns expected (and Runs Created, which we should ignore completely):
Event Markov BaseRuns Runs Created
Walk 0.654 0.349 0.124
Single 0.778 0.680 0.619
Double 1.087 1.231 1.114
Triple 1.373 1.764 1.609
Homerun 1.714 1.821 2.105
Out, sans K -0.678
Strikeout -0.715

With no doubles, triples, or HR in the league, it becomes:
BB +.65
1B +.78
Out -.69

Here’s the RE chart:
Bases 0 outs 1 out 2 outs
xxx 1.109 0.535 0.159
1xx 1.706 0.927 0.324
x2x 1.816 1.050 0.446
xx3 2.002 1.266 0.493
12x 2.498 1.531 0.658
1x3 2.617 1.687 0.689
x23 2.727 1.809 0.811
123 3.447 2.369 1.147

The one to focus on, in all these RE charts, is man on 2B, 1 out and man on 1B, 0 outs.  Normally, the are off by .20-.30 runs or so, meaning that while on average you would not want to sac bunt, you can conceive of conditions where you are better off with man on 2B 1 out, than man on 1B 0 outs.

In this particular case however, the gap is .65 runs.  You’d have to have a very extreme condition for the sac bunt play to even be attempted. 

The next one to focus on is man on 1b 0 outs, man on 2b 0 outs, no men on 1 out: 1.706, 1.816, .535.  That’s the SB situation with 0 outs: add +.11 runs against losing 1.17 runs.  Basically, you have to be safe 10 times for each time you are thrown out, just to breakeven.  With 1 out, you have to be safe 86% of the time.

You notice that with 0,1, or 2 outs, you always gain around +.12 runs on the SB.  But, on the CS, you lose less with 2 outs.  So, the ONLY time you really should try to steal is with 2 outs.  With 2 outs, the breakeven point is 73%.  So, LOTS of steals should be attempted on 2 outs, if you have a runner on 1B.  The reason should be clear enough: there are no extrabase hits, and so, you always need TWO safe plays to score the runner.  And with 2 outs, you have no net.


#6    MGL      (see all posts) 2008/11/03 (Mon) @ 15:28

Please keep in minds as an amateur coach, especially at the lower levels, is that you have to also decide what your “utilities” are.  It is doubtful that it is solely to operate such that you maximize your chances of winning every game. There are probably all kinds of other, perhaps more important, considerations, not the least of which are the kids having fun, developing skills that might be important later in their careers, fostering respect for one another, and teamwork, thinking positive, setting goals, etc.

You also have a chance to share some of the technical things we are discussing with the kids so that they can make better tactical choices in their baseball careers and for them to be more critical, logical thinkers in life in general.

And don’t forget that a big part of the tactical decisions-making process in baseball - maybe the biggest - is not SB, bunts, IBB’s and the like.  It is the batter/pitcher matchup.  Helping the kids to make better decisions on the mound and at the plate, given the context of the game, is invaluable.  Introduce them to game theory.  Explain how they can hit better if they put themselves in the mind of the pitcher, and vice versa.  Explain to the pitchers how it is important to randomize their pitches.  Explain how the approach of the batter and pitches can change dramatically depending on the outs, runners, inning, score, and the following batters.  Etc.


#7    Jeff      (see all posts) 2008/11/03 (Mon) @ 15:29

Tango - Can you ask the coach what section and league(s) he plays against in California.  The website, MaxPreps (make sure to sort by OBP to get all the stats) has the stats to get the actual numbers for the players (looks a little lower than your assumptions).  I know that H.S. baseball definitely has different levels of competition and before getting the numbers, I would like to do at least choose a comparable set of data.


#8    Tangotiger      (see all posts) 2008/11/03 (Mon) @ 15:37

Unfortunately, he provided no email address, and so, I will leave it to him to contribute to this thread…


#9    Eric      (see all posts) 2008/11/03 (Mon) @ 23:01

Not the original emailer, but I would say focus on the CCS.


#10    Jeff      (see all posts) 2008/11/04 (Tue) @ 20:09

Finding HS data is rough (the previously mentioned MaxPreps doesn’t have SO data), but I did find some initial data.  I might be able to use the MaxPreps data and just set SO to 0 since with this data, they have the same run loss (-.51). Comments?

The data I got was the season stats from the 8 finalists for the Kentucky State championships (http://www.khsaa.org/baseball/)from 2005 to 2009.  By far not a great sample, but all I could find.  Here are the results:

AB R H RBI 2B 3B HR SO BB
32655 8930 11045 7439 2262 347 524 5130 4517

8.053 : Runs Scored per Game

Run Expectancy Matrix
Bases 0 outs 1 out 2 outs
xxx 0.895 0.464 0.160
1xx 1.450 0.836 0.332
x2x 1.591 0.981 0.461
xx3 1.785 1.195 0.500
12x 2.193 1.395 0.649
1x3 2.347 1.576 0.680
x23 2.488 1.721 0.809
123 3.118 2.186 1.068

Marginal Run Values by Event
Event Markov BaseRuns Runs Created
Walk 0.513 0.409 0.244
Single 0.649 0.595 0.662
Double 0.944 0.914 1.081
Triple 1.205 1.232 1.500
Homerun 1.551 1.478 1.918
Out, sans K -0.501
Strikeout -0.501


Page 1 of 1 pages


Name (required)
E-Mail (optional)
Website (optional)

<< Back to main


Latest...

COMMENTS

Jan 09 16:41
Sabermetric Moves of the 2009 Pre-Season

Jan 09 19:56
Modeling Baseball Player Ability with a Nested Dirichlet Distribution

Jan 09 18:08
Line Drives

Jan 09 18:04
Challenging Nate Silver (and all other forecasters)

Jan 09 17:31
Cheers

Jan 09 17:14
Teaching sabermetrics at school

Jan 09 16:51
The first Hardball Times Annual available for download!

Jan 09 14:44
Vote for the Worst Player in MLB

Jan 09 12:29
Clint Eastwood is Archie Bunker

Jan 09 12:16
Mailbags on Parade