RISP (A Thought Experiment for the Intrepid)

[GullyFoyle]

The Dunn and Marty thread got me thinking, and I thought I'd break this issue out a little because the original thread was getting a little unwieldy. Feel free to ignore, delete or move back into it.

This thought experiment starts with two players:


Player A: (avg/obp/slg)
Low / High / High
strengths: avoids outs, hits for power
weakness: trades some hits for walks

Player B:
High/ Low / High
strengths: more hits, hits for power
weakness: more likely to get out



Now, I think most everyone today would prefer Player A in general. Because most of the time you don't care how he gets on base, just that he gets on base. But with men on the dynamic shifts and plays a little to Player B's strengths and Player A's weaknesses (but not all the time).

Examples:

A) Bases Loaded: Player A preferred - just avoid the damn out, bonus - can hit with power
B) Man on 1st: Player A preferred - a hit wont necessarily score the run, so move him over, avoid the out

C) Man on 2nd:
D) Man on 3rd:
E) Man on 1st/2nd:
F) Man on 2nd/3rd:
G) Man on 1st/3rd:

Now, here are the dilemmas... I would still prefer Player A, but I can understand why Player B is getting more attention. You have to look at what batters are coming up next in the lineup. If Player A is followed by someone with a high AVG or a high OBP, then you prefer him – he is most likely to avoid the out and increases the chance of extra runs being scored (with the possible exception of close and late games). If the next guy has a high chance of making an out then you might prefer Player B, but the real problem is with the next hitter not with Player A.

So if the debate is between keeping Player A or Player B, then it should be realized that in most plate appearances A is preferred, and in the minority of plate appearances that play to Player B's strengths (those with RISP), A is still a very strong candidate (if not out right preferred) based on the quality of hitters in the lineup after him.


OK, Thanks for reading... feel free to poke holes in my thinking, experiment, etc.... :)

[Highlifeman21]

For me, it's very simple.

Player A has a strength of avoiding outs, and I'm not sold that it's a weakness to trade hits for walks.

Player B has a glaring weakness: more likely to get out. That weakness will magnify itself in the spectrum of a large sample size. He can get all the hits for all the power he wants, but it won't eliminate his weakness.

Numerous seemingly educated people on this board have accurately simplified hitting and scoring runs to the principle of not making outs.

Regardless of the situation (men on, where are they, outs, count, etc), Player A still has an overwhelmingly large competitive advantage over Player B.

[TRF]

Player A Adam Dunn.
Player B Alfonso Soriano, or if you really want to be current and use stats from this year, Austin Kearns.

I'd take Dunn.

[TOBTTReds]

and I'm not sold that it's a weakness to trade hits for walks.


It might not be a weakness, but you still want a hit over a walk.

As far as player B is concerned, correct me if I'm wrong, it would be tough to have a high avg and slg, but have a low OBP. High SLG almost always equals high OBP, especially if you have a good avg....just throwin that out there. I'm sure there are some cases out there, just not sure.

[GullyFoyle]

Regardless of the situation (men on, where are they, outs, count, etc), Player A still has an overwhelmingly large competitive advantage over Player B.

In general I agree, but playing devil's advocate a little, I would say that in VERY specific situations its not clear cut...

2 outs, bottom of the ninth, down a run, man on third...

Here Player A' has a better chance to not make an out... but he still has a chance to be out. If you take that gamble and just "win" a base on balls (as opposed a base on balls), then you have to gamble again...

(this may be simplified too much, but communicates the idea)

So with Player A the outcomes in this situation are:
AVG (say .250) -> 25% chance to get a hit
OBP (say .450) minus AVG -> 20% chance of passing the at bat on to next hitter
Plate Appearance minus OBP -> 55% chance to make an out

25% to tie game
20% to go to next batter
55% to lose

With Player B:
AVG (say .350) -> 35% chance to get a hit
OBP (say .400) minus AVG -> 5% chance of passing the at bat on to next hitter
Plate Appearance minus OBP -> 60% chance of making out

35% to tie game
5% to go to next batter
60% to lose

Now if the next batter sucks (say a .225 AVG / .250 OBP) then he has a 75% of making an out himself, so that 20% chance to pass the next at bat on becomes another 15% of an out, and a 5% of hit (or passing it on again)making the combined chances of Player A and sucky guy being:

Player A with sucky guy:
~30% to tie
70% to lose

what is lost here of course is that there is also an increased chance of scoring more than run, and usually that chance is preferred... but in this scenario (again very specific) you prefer Player B I think.

(any problems with my math?)

[TRF]

I don't see how you ever prefer the guy that makes more outs. And what .350 hitter ever only OBP'd .400?

Why am I sure woy has the answer to that? :)

[GullyFoyle]

I don't see how you ever prefer the guy that makes more outs. And what .350 hitter ever only OBP'd .400?

Why am I sure woy has the answer to that? :)

Yeah, those were quick number inputs... .350 was a little much... I'm laughing at them now

Though....

2006 David Wright .353 / .379 / .667 :)

[Handofdeath]

It might not be a weakness, but you still want a hit over a walk.

As far as player B is concerned, correct me if I'm wrong, it would be tough to have a high avg and slg, but have a low OBP. High SLG almost always equals high OBP, especially if you have a good avg....just throwin that out there. I'm sure there are some cases out there, just not sure.


I don't buy the "a walk is the same as a single". If there is a runner on second and you hit a single then that runnner scores or at worst is at third. If you get a walk then that runner is still at second base.

[RedsManRick]

I don't see how you ever prefer the guy that makes more outs. And what .350 hitter ever only OBP'd .400?

Why am I sure woy has the answer to that? :)

In 2001 Ichiro hit .350 with an OBP of .381.

[Highlifeman21]

It might not be a weakness, but you still want a hit over a walk.

As far as player B is concerned, correct me if I'm wrong, it would be tough to have a high avg and slg, but have a low OBP. High SLG almost always equals high OBP, especially if you have a good avg....just throwin that out there. I'm sure there are some cases out there, just not sure.


High SLG or AVG doesn't have to equal high OBP at all.


Career Totals

Player AVG OBP SLG
A. Soriano .280 .321 .500
I. Suzuki .330 .376 .439
M. Loretta .300 .365 .408
M. Young .298 .342 .452
C. Guillen .283 .349 .427
J. Damon .290 .353 .432
R. White .287 .340 .469

There's a small collection of players who with their respective career BA, have relatively low career OBP, which means 1 blatant thing: they don't take enough walks.


Career Totals

Player AVG OBP SLG
J. Francoeur .283 .314 .512
T. Clark .268 .343 .494
D. Lee .276 .364 .502
D. Ortiz .283 .368 .537
K. Griffey Jr .293 .377 .561
A. Jones .268 .343 .505
A. Ramirez .276 .328 .480
R. Sanders .267 .344 .491
R. Sexson .270 .352 .529

With this group, I looked at guys that have career SLG near .500, and only KGJ had a decent OBP (thus why I included him in this group). I think you would agree, aside from maybe Francoeur, that these are "power" hitters, but over the course of their career, they also have not walked enough to increase their career OBP.

And what about this guy?


Career Totals

Player AVG OBP SLG
A. Dunn .249 .384 .524

High SLG does equal high OBP in this case, but at the age of 26, we as Reds fans can only hope that career AVG of .249 only goes up


Edited: TRF pointed out that I flip-flopped Dunn's Career OBP and SLG numbers

[TRF]

I'm assuming that .384 SLG you have for AD is a typo right?

[TRF]

If your OBP is 40 points higher than your BA, you better be hitting .340 or higher.

.280 BA for Soriano and a .321 OBP. blech. A .350 OBP to me is respectable, in fact it's pretty decent.

Ichiro hitting .330 and having an OBP of .376 means to me his OBP is hit driven, and prone to slumps. His slumps hurt the team because he doesn't walk when he isn't hitting. We know this because he doesn't walk much when he is hitting.

Dunn is always getting on base. And when he goes into a hitting slump, he's still getting on base.

[Handofdeath]

This feels too much like math class.

[Cyclone792]

Run Expectancy Matrix, courtesy of TangoTiger (http://www.tangotiger.net/RE9902.html):


Run Expectancy 1999-2002

Outs Situation
Base Situation 0 outs 1 out 2 outs

Empty 0.555 0.297 0.117
1st 0.953 0.573 0.251
2nd 1.189 0.725 0.344
3rd 1.482 0.983 0.387
1st_2nd 1.573 0.971 0.466
1st_3rd 1.904 1.243 0.538
2nd_3rd 2.052 1.467 0.634
Loaded 2.417 1.650 0.815

It's pretty simple for me to determine what type of hitter I want up in any given situation. That hitter would be the hitter that has the highest probability of creating an event with the best possible sum, in any given plate appearance, of Actual Runs Scored + Subsequent Run Expectancy for the rest of the inning thereafter. In other words, I prefer the hitter who does a better job of avoiding outs and acquiring bases because that's the hitter with the highest probability of creating the highest run producing event combined with the most run producing events period.

The only time where I may prefer a hitter who is worse in avoiding outs/acquiring bases, but has one specific skillset geared towards producing one "scratch" run is at the end of a game (i.e. last inning) when a runner is on third base and that one individual baserunner on third can win or tie the game simply by crossing the plate at that precise moment.

The key point to that scenario, however, is that it is an exceptionally rare situation. In fact, if I had to fathom a guess as to how rare, that scenario probably pops up in 0.01% of all plate appearances, if even that. Essentially, those situations are so rare that they're not even worth worrying about.

[Yachtzee]

I don't buy the "a walk is the same as a single". If there is a runner on second and you hit a single then that runnner scores or at worst is at third. If you get a walk then that runner is still at second base.

Sure, we all would like a single in that situation. But a walk is still a positive outcome that helps the team (+1 base, 0 outs). With a walk, the guy on second is still in a position to score for the next guy. The guy who took a walk instead of wildly hacking in an attempt to put the ball in play saved his team an out.

[RedsManRick]

Run Expectancy Matrix, courtesy of TangoTiger (http://www.tangotiger.net/RE9902.html):


Run Expectancy 1999-2002

Outs Situation
Base Situation 0 outs 1 out 2 outs

Empty 0.555 0.297 0.117
1st 0.953 0.573 0.251
2nd 1.189 0.725 0.344
3rd 1.482 0.983 0.387
1st_2nd 1.573 0.971 0.466
1st_3rd 1.904 1.243 0.538
2nd_3rd 2.052 1.467 0.634
Loaded 2.417 1.650 0.815

It's pretty simple for me to determine what type of hitter I want up in any given situation. That hitter would be the hitter that has the highest probability of creating an event with the best possible sum, in any given plate appearance, of Actual Runs Scored + Subsequent Run Expectancy for the rest of the inning thereafter. In other words, I prefer the hitter who does a better job of avoiding outs and acquiring bases because that's the hitter with the highest probability of creating the highest run producing event combined with the most run producing events period.

The only time where I may prefer a hitter who is worse in avoiding outs/acquiring bases, but has one specific skillset geared towards producing one "scratch" run is at the end of a game (i.e. last inning) when a runner is on third base and that one individual baserunner on third can win or tie the game simply by crossing the plate at that precise moment.

The key point to that scenario, however, is that it is an exceptionally rare situation. In fact, if I had to fathom a guess as to how rare, that scenario probably pops up in 0.01% of all plate appearances, if even that. Essentially, those situations are so rare that they're not even worth worrying about.


I agree with your general point Cyclone, but one problem I have with the use of the expected run matrix, particularly in this is example, is that in a given context it fails to account for ability of the current batter and the ability of the subsuquent batters.

Anyways, to flush out your point Cyclone...

Dunn (.250/.380/.550) has 25% chance of scoring the runner and a 15% of merely extending the inning while getting on base. If he extends the inning in any way, the next batter (.270/.340/.500) comes up and has a 27% chance of scoring the run. We could continue the train, but let's stop there before the possibilities get out of control. Because a walk requires the next hitter to get a hit in order to score the run, we must look at his BA and multiply it by Dunns walk % to see the odds that Dunn's walk leads to a scored run. (.27*.15 = 4%). As we extend this sequence of possibilities, the percent chance that the run scores in a given AB following Dunn approaches 0% quickly. Let's add another 1% to account for these. The total chance the Run scores on or after Dunn's AB is 30%. (.344>.3 because the right tail of the potential run score distribution raises the average even though the number of occurances is extremely small).

In Soriano's case (.300/.330/.550), he has a 30% chance of scoring the runner himself a 3% chance of extending the innnig via a walk without scoring the run. 5% times the following batter's BA gives us ~1%. This gives us a 31% chance to score the runner. However, the tail of the distribution is flatter and shorter because of the lower percentage of times the inning extends past him.

So there you go; with two outs, the go ahead run in scoring position, first base empty, and a good but not great hitter behind the current batter, the runner on base stands a ~1% greater chance of scoring with a HIGH/LOW/HIGH guy at the plate than with a LOW/HIGH/HIGH guy. However, your ability to score more than 1 more is still greater with Dunn than with Soriano, so unless you only want to score 1, you're better off with Dunn.

[Cyclone792]

I agree with your general point Cyclone, but one problem I have with the use of the expected run matrix, particularly in this is example, is that in a given context it fails to account for ability of the current batter and the ability of the subsuquent batters.

Anyways, to flush out your point Cyclone...

Dunn (.250/.380/.550) has 25% chance of scoring the runner and a 15% of merely extending the inning while getting on base. If he extends the inning in any way, the next batter (.270/.340/.500) comes up and has a 27% chance of scoring the run. We could continue the train, but let's stop there before the possibilities get out of control. Because a walk requires the next hitter to get a hit in order to score the run, we must look at his BA and multiply it by Dunns walk % to see the odds that Dunn's walk leads to a scored run. (.27*.15 = 4%). As we extend this sequence of possibilities, the percent chance that the run scores in a given AB following Dunn approaches 0% quickly. Let's add another 1% to account for these. The total chance the Run scores on or after Dunn's AB is 30%. (.344>.3 because the right tail of the potential run score distribution raises the average even though the number of occurances is extremely small).

In Soriano's case (.300/.330/.550), he has a 30% chance of scoring the runner himself a 3% chance of extending the innnig via a walk without scoring the run. 5% times the following batter's BA gives us ~1%. This gives us a 31% chance to score the runner. However, the tail of the distribution is flatter and shorter because of the lower percentage of times the inning extends past him.

So there you go; with two outs, the go ahead run in scoring position, first base empty, and a good but not great hitter behind the current batter, the runner on base stands a ~1% greater chance of scoring with a HIGH/LOW/HIGH guy at the plate than with a LOW/HIGH/HIGH guy. However, your ability to score more than 1 more is still greater with Dunn than with Soriano, so unless you only want to score 1, you're better off with Dunn.

Yep, the run expectancy paradigm will shift slightly with different team offenses as the overall run expectancy is pretty much for an average offense. Specifically, though, with an offense like the Reds and the overall potent lineup Dunn is in, the shift should be more in the favor of wanting him to take the walk and add the extra baserunner since the subsequent hitter in our offense should be slightly better than the average offense. How much the paradigm shifts for different offenses, high scoring and low scoring, is something I'm not too sure of, but also something I'd very much love to find some data on.

And like Earl Weaver, I'm always looking to score more than one run, heck I'm looking to score as many runs as I possibly can :) That's why unless there's a situation very late in the game where one run is of vital importance, I'm preferring to play the odds to the advantage of scoring as many runs as I possibly can, and I'll always prefer that strategy over the very slight advantage of being able to scratch across merely one run if it comes at the cost of losing a potentially big inning.

[Cyclone792]

Yep, the run expectancy paradigm will shift slightly with different team offenses as the overall run expectancy is pretty much for an average offense. Specifically, though, with an offense like the Reds and the overall potent lineup Dunn is in, the shift should be more in the favor of wanting him to take the walk and add the extra baserunner since the subsequent hitter in our offense should be slightly better than the average offense. How much the paradigm shifts for different offenses, high scoring and low scoring, is something I'm not too sure of, but also something I'd very much love to find some data on.

I must be going crazy since I'm now quoting myself, however ... what I found was highly convenient to our discussion here ...

The very question of how the run expectancy matrix shifts from batter to batter was brought up on SOSH and TangoTiger chimed in. Here is what Tango says regarding run expectancy and how it changes from batter to batter:

http://sonsofsamhorn.net/index.php?showtopic=5082


Easily done. Suppose that Ortiz is about a 0.10 runs / PA better than an average hitter. Therefore, on average, you would add +.10 runs to every value in the run expectancy chart. .555 becomes .655, etc. That's what the +.10 means, after all.

Of course, he doesn't add +.10 equally to each of the 24 base out states. Another shortcut you can use is +.07 for bases empty and +.14 with men on base. For those who follow my Leverage Index (LI), the LI with men on base is about twice that with bases empty. And of course, you'd have different numbers by outs.

In the end, all you have to do is this. Go here:
http://www.tangotiger.net/RE9902event.html

Take the base/out state you are interested in. Apply the player's expected stats for that base/out state. You'll end up with his Linear Weights by the 24-base/out state. Add that figure to your RE matrix for that particular base/out state. Done.

This assumes that the following batter is average.

If you want it complicated, you simply write a Markov chain program, put in your inputs, and you get your RE chart.

In the book, I have an RE chart for a pitcher or team that allows 3.2 Runs per game. I used that program to do the batting order analysis, where I put different batters in each slot.

[GullyFoyle]

Dunn (.250/.380/.550) So there you go; with two outs, the go ahead run in scoring position, first base empty, and a good but not great hitter behind the current batter, the runner on base stands a ~1% greater chance of scoring with a HIGH/LOW/HIGH guy at the plate than with a LOW/HIGH/HIGH guy. However, your ability to score more than 1 more is still greater with Dunn than with Soriano, so unless you only want to score 1, you're better off with Dunn.

Thanks for illustrating my point a little better... when I was considering this I thought there must be an instance when it was at least debatable... turns out it to be a very extreme case but thats fine... I'm a big believer in finding the extremes of an argument to better situate the typical cases.

Thanks for the info.

[GullyFoyle]

I must be going crazy since I'm now quoting myself, however ... what I found was highly convenient to our discussion here ...

The very question of how the run expectancy matrix shifts from batter to batter was brought up on SOSH and TangoTiger chimed in. Here is what Tango says regarding run expectancy and how it changes from batter to batter:

http://sonsofsamhorn.net/index.php?showtopic=5082

Thanks Cyclone... I had vaguely remembered seeing this over at SoSH (where I often lurk) but I'm glade to have the info again... going to store it away as a PDF for future reference.

BTW, its good to see someone over there keeping an eye on Wily Mo and other Reds past / future interests.

[2001MUgrad]

I can follow Gully,

Good post. I like the fact that you went further than just that AB and went down to at least the next guy in the batting order. One thing that I didn't notice taken into account and maybe it was, is how many outs there where. If there were less than 2 outs, how often does Player B produce an out either by Sac Fly, ground ball to the right side, or sac bunt that scores a run. Also, the guy that walks, how many of those times that he walks does he actually score Vs. the guy that gets an out, but produces the run.

Way too much math. What ever happened to just watching the game?

[Yachtzee]

Way too much math. What ever happened to just watching the game?

That's still the best part. :)

Baseball has both a narrative element and a statistical element. We can debate the stats all day and disagree on who's better in such and such situation, yet still enjoy the story that each game presents as it unfolds.

[2001MUgrad]

That's still the best part. :)

Baseball has both a narrative element and a statistical element. We can debate the stats all day and disagree on who's better in such and such situation, yet still enjoy the story that each game presents as it unfolds.

You'd think :beerme: