PDA

View Full Version : Quantifying RBIs Sabremetrically



Archer1979
12-03-06, 08:20 AM
This is from fellow forumer Groovitude who can't start new threads yet. After you're done pointing fingers and laughing at him for his newbieness, read the following, it's a good point of discussion:

Per Groovitude:

The results of the American League Most Valuable Player award has spurred quite a bit of backlash from many fans of baseball, far beyond the mere confines of Yankee fandom. Many people point to the Baseball Writers' Association of America looking foolishly at merely HR and RBI totals and base their decision solely off of that, when they feel there are much more important statistics to look at. Furthermore, some view RBIs as an entirely outdated institution altogether -- Rotoworld, in their short debunking of Morneau's win, stated that he was "leading the league in no significant categories" (November 21st, http://www.rotoworld.com/content/pla...t=MLB&id=3602) (http://www.rotoworld.com/content/playerpages/player_main.aspx?sport=MLB&id=3602)), despite that he was the AL RBI leader.

I, however, feel that a player's ability to drive in runs is a perfectly valid way to judge their worth. Solely looking at it as a matter of totals, however, does not give us a full picture of the player's ability to do so. For instance, Jeter's ability to drive in as many runs as Morneau is hindered by his normal spot in the lineup -- driving in lots of runs from the two hole is significantly harder than it is from the five or six hole. It is also arguable that Jeter's ability is not hindered significantly, as his supporting cast is generally stronger. How can we compare these two -- or any players, for that matter -- in a way that essentially ignores these factors?

As a matter of comparison, I will be using Justin Morneau and Derek Jeter in 2006. The stats I am using are from ESPN's site -- Morneau's stats from http://sports.espn.go.com/mlb/player...ting&year=2006 (http://sports.espn.go.com/mlb/players/splits?statsId=7063&type=batting&year=2006) and Jeter's from http://sports.espn.go.com/mlb/player...ting&year=2006 (http://sports.espn.go.com/mlb/players/splits?statsId=5406&type=batting&year=2006).

To try to gain a better insight as to what their RBI numbers mean, we should look at their total RBI opportunities. To do so, we should treat each at-bat with no one on as one possible RBI (a solo homerun), each at-bat with one man on as two possible RBIs, and so on and so forth. Looking at Jeter's chart, we see that he has 355 at-bats with one RBI opportunity, 189 at-bats with two RBI opportunities (113 with a man on first, 48 with a man on second, and 28 with a man on third), and so forth. In total, he has a total of 981 RBI opportunities. Formulaically,

RBI Opportunities = (AB with bases clear) + 2*(AB with one man on) + 3*(AB with two men on) + 4*(AB with bases loaded)

To get a ratio to compare, we simply need to divide total RBIs by total RBI opportunities, giving us an RBI to RBI opportunity ratio (which, for simplicity, I will refer to as RBIA, or RBI Average). Comparing Jeter and Morneau, Morneau still stands above Jeter considerably with an RBIA of .133 to .099. By multiplying the RBIA with a set number of RBI opportunities. Morneau and Jeter actually had 981 RBI opportunities apiece; thereby, it is perfectly comparable in this regard to compare Morneau's 130 RBIs to Jeter's 97.

It occured to me, however, that it might be valid to look at how many of their teammates they were able to bring home. This effectively changes the situation that we are measuring. For one, it measures a player's ability to hit in more "clutch" situations, certainly important when naming the Most Valuable Player. Secondly -- and this may be inadvertent Jeter favoritism -- it takes some of the power factor out of the equation, as we are no longer counting one RBI per home run, taking the players' ability to drive himself in -- only his teammates.

To measure this, we must first calculate Teammates Batted In (TBI). This is simple enough; simply subtract home runs from RBIs. Jeter's 97 RBI and 14 HR leaves him with 83 TBI.

Teammates Batted In = RBI - HR

In order to calculate the total number of TBI opportunities, we revise the RBI opportunities formula, disregarding at-bats with no one on, and subtracting one from each constant. For instance, rather than two RBI opportunities when there is a man on-base, there is one TBI opportunity; rather than three RBI opportunities when there are two men on, there are two TBI opportunities, and so forth. Jeter, for instance, has 358 TBI -- a considerably different number than his 981 RBI opportunities.

TBI Opportunities = (AB with one man on) + 2*(AB with two men on) + 3*(AB with bases loaded)

This changes the comparison between Morneau and Jeter considerably. While their RBI opportunity counts were the same, their TBI opportunity counts are not. Jeter's increased at-bat count from batting higher in the order added to his single RBI opportunities considerably; Morneau's batting later in the order allowed more men on base ahead of him, also increasing his RBI opportunties. With Jeter's added at-bats (and additional single RBI opportunities) eliminated, we see that Morneau had more than thirty more men on to possibly drive in -- Morneau's TBI opportunities stood at 389, while Jeter's stood at 358. In order to compare, we need to divide their TBI by their TBI opportunties -- the ratio leaving us with what I will call the Teammates Batted In Average, or TBIA for short.

Jeter and Morneau's TBIAs are much more comparable than their RBIAs are. Jeter stands at .232, while Morneau stands at .247. If you multiply Jeter's TBIA by Morneau's TBI opportunities, you get 90.25 TBI -- the amount of TBI Jeter would have earned if he had hit comparably with Morneau's chances. Comparitively, Morneau had 96 TBI.

This is by no means a complete look at an objective look at RBIs. I feel as though a number of factors have not been addressed. Some, I believe, are not addressable, like the speed of one's teammates on base -- for instance, having Jorge Posada on second base and having Jose Reyes on second base are two entirely different situations. Others, however, may be addressable; for instance, TBIA does not value batting in one man on first and one man on third the same way, nor does it value batting in one man on two separate occasions differently than two men at the same time. How this should be addressed, or how to value these situations are, I believe, out of my range of expertise, and I put it forth to other statistic-hungry baseball fans to suggest additions, changes, and other formulas to help take a fresh look at what many unfortunately see as a totally outdated stat.<!-- / message --><!-- Sig Was Here --><!-- edit note -->

groovitude
12-03-06, 09:14 AM
Thanks for posting, Archer. Please note that I misspoke early in the piece -- Morneau was not the AL RBI leader, he was second.

JDPNYY
12-03-06, 10:53 AM
Nice work Archer.

Soriambi
12-03-06, 11:47 AM
Yeah, nice stuff, Tim.


Seriously, nice work. I came up with a formula like this several months ago trying to see how far above average each guy is RBI wise given their opportunities but it didn't control for a bunch of things that I thought it should have. As for your work, groovy, the one suggestion I would make is to distinguish between what bases they're on. A guy with one guy on base has a different opportunity if the guy is on 3rd base rather than 1st base. I think it would be a lot more valid if it included that in some way.

JL25and3
12-03-06, 01:40 PM
TBIA does not value batting in one man on first and one man on third the same way, nor does it value batting in one man on two separate occasions differently than two men at the same time.

I think you mean that it does value them the same way - an RBI opportunity is an RBI opportunity, whether it's a man on first with two out or a man on third with one out.

I also think that removing home runs from the equation is a mistake. Even if one buys the concept and importance of clutchitudity, this doesn't improve the measure's sensitivity to it. In the late innings of a close game, with one man on base, isn't a home run inherently more valuable than an RBI base hit?

jughead
12-03-06, 02:32 PM
I think you mean that it does value them the same way - an RBI opportunity is an RBI opportunity, whether it's a man on first with two out or a man on third with one out.

I also think that removing home runs from the equation is a mistake. Even if one buys the concept and importance of clutchitudity, this doesn't improve the measure's sensitivity to it. In the late innings of a close game, with one man on base, isn't a home run inherently more valuable than an RBI base hit?
It is, but the point of TBI seems to be that it doesn't matter how the guy on base got driven in - whether the player hit a double or a HR to score him doesn't matter. That's why he said it takes the power factor out of the equation.

Archer1979
12-03-06, 02:37 PM
I think you mean that it does value them the same way - an RBI opportunity is an RBI opportunity, whether it's a man on first with two out or a man on third with one out.

I also think that removing home runs from the equation is a mistake. Even if one buys the concept and importance of clutchitudity, this doesn't improve the measure's sensitivity to it. In the late innings of a close game, with one man on base, isn't a home run inherently more valuable than an RBI base hit?

:lol:

Don't talk to me. Talk to groovitude.

JL25and3
12-03-06, 03:59 PM
It is, but the point of TBI seems to be that it doesn't matter how the guy on base got driven in - whether the player hit a double or a HR to score him doesn't matter. That's why he said it takes the power factor out of the equation.

Okay, but why is scoring a runner from first base inherently better than scoring a runner - the batter - from home? Why discount those RBIs?

groovitude
12-03-06, 05:29 PM
I think you mean that it does value them the same way - an RBI opportunity is an RBI opportunity, whether it's a man on first with two out or a man on third with one out.


As for your work, groovy, the one suggestion I would make is to distinguish between what bases they're on. A guy with one guy on base has a different opportunity if the guy is on 3rd base rather than 1st base. I think it would be a lot more valid if it included that in some way.

That's correct, JL25and3; I meant to say that they don't value them differently, when I believe that a totally accurate measurement of these stats should, as Soriambi put forth. I'll see what I can do about this; I am, however, completely open to outside help, suggestions, or someone else coming up with a formula completely on their own.


I also think that removing home runs from the equation is a mistake. Even if one buys the concept and importance of clutchitudity, this doesn't improve the measure's sensitivity to it. In the late innings of a close game, with one man on base, isn't a home run inherently more valuable than an RBI base hit?

A home run is always an RBI opportunity, but is not always a TBI opportunity. Therefore, we exclude it from the TBIA and allow for the possibility of it to be factored in later. For instance, if we were to compare Jeter and Morneau again, we might start with their calculated TBIAs and multiply them by one of their TBI opportunity counts. I've already calculated this in the previous document -- Morneau has 96 TBI, and Jeter has 90.25. To reinclude home runs (and the RBIs associated with them), we should calculate their HR/AB ratios, and then multiply that by the AB of the player whose TBI opportunity count we used.


HR/AB: Jeter .022; Morneau .057

(HR/AB)*(592): Jeter 13; Morneau 34
(Morneau had 592 AB)

Add these numbers to the TBIA's projections, and they will be even in terms of ABs and TBI opportunities. This is to say, if Jeter performed to the same percentages as he did this season from where he was in each of Justin Morneau's at-bats, with however many men on as Morneau did, he would've earned this RBI total.

Adjusted RBI Totals: Jeter 103; Morneau 130.

While this is not a tremendous difference -- Jeter only has six more RBIs in Morneau's shoes than he did from his own spot -- he would've been tied for #18 in the American League for RBIs (coming in ahead of Manny Ramirez) rather than coming in at #22, and he would've broken the 100 barrier. While those six RBIs don't make a whole lot of mathematical difference, they sure do make an impressionable one.

As an argument for this system rather than including HR from the get-go (using the RBIA system), I'll compare Justin Morneau and Mark Grudzielanek, using Grudzielanek's RBI and TBI opportunity counts and adjusting Morneau's stats to compare.

RBI: Morneau 130 / Grudzielanek 52
RBI Opp: Morneau 981 / Grudzielanek 857
RBIA: Morneau .133 / Grudzielanek .061
Adjusted RBI (by RBIA): Morneau 114 / Grudzielanek 52

TBI: Morneau 96 / Grudzielanek 45
TBI Opp: Morneau 389 / Grudzielanek 309
TBIA: Morneau .247 / Grudzielanek .146
HR/AB: Morneau .057 / Grudzielanek .013
Adjusted RBI (by TBIA): Morneau 107 / Grudzielanek 52
(Grudzielanek had 548 AB)

I don't know about you, but I have a much easier time believing that Morneau would hit 107 RBI rather than 114 RBI from the two hole on the Kansas City Royals. I feel that adjusting RBI by TBIA is more accurate rather than by RBIA, and thereby giving cause to leaving out HR from TBIA.


Yeah, nice stuff, Tim.

My name's Erik. ;)

Soriambi
12-03-06, 07:01 PM
My name's Erik. ;)

I was playing off of John's joke that Archer did the work. (Archer's name is Tim.) ;)

Archer1979
12-03-06, 08:18 PM
I was playing off of John's joke that Archer did the work. (Archer's name is Tim.) ;)

That's actually my middle name. My first name is, ironically enough, Groovitude.

jughead
12-04-06, 09:31 AM
Okay, but why is scoring a runner from first base inherently better than scoring a runner - the batter - from home? Why discount those RBIs?Because the point is to compare how well the player did at driving in runs when there were runners on base, not hitting HRs.

Dave Visbeck
12-04-06, 10:37 AM
This is from fellow forumer Groovitude who can't start new threads yet. After you're done pointing fingers and laughing at him for his newbieness, read the following, it's a good point of discussion:

Per Groovitude:

The results of the American League Most Valuable Player award has spurred quite a bit of backlash from many fans of baseball, far beyond the mere confines of Yankee fandom. Many people point to the Baseball Writers' Association of America looking foolishly at merely HR and RBI totals and base their decision solely off of that, when they feel there are much more important statistics to look at. Furthermore, some view RBIs as an entirely outdated institution altogether -- Rotoworld, in their short debunking of Morneau's win, stated that he was "leading the league in no significant categories" (November 21st, http://www.rotoworld.com/content/pla...t=MLB&id=3602) (http://www.rotoworld.com/content/playerpages/player_main.aspx?sport=MLB&id=3602)), despite that he was the AL RBI leader.

I, however, feel that a player's ability to drive in runs is a perfectly valid way to judge their worth. Solely looking at it as a matter of totals, however, does not give us a full picture of the player's ability to do so. For instance, Jeter's ability to drive in as many runs as Morneau is hindered by his normal spot in the lineup -- driving in lots of runs from the two hole is significantly harder than it is from the five or six hole. It is also arguable that Jeter's ability is not hindered significantly, as his supporting cast is generally stronger. How can we compare these two -- or any players, for that matter -- in a way that essentially ignores these factors?

As a matter of comparison, I will be using Justin Morneau and Derek Jeter in 2006. The stats I am using are from ESPN's site -- Morneau's stats from http://sports.espn.go.com/mlb/player...ting&year=2006 (http://sports.espn.go.com/mlb/players/splits?statsId=7063&type=batting&year=2006) and Jeter's from http://sports.espn.go.com/mlb/player...ting&year=2006 (http://sports.espn.go.com/mlb/players/splits?statsId=5406&type=batting&year=2006).

To try to gain a better insight as to what their RBI numbers mean, we should look at their total RBI opportunities. To do so, we should treat each at-bat with no one on as one possible RBI (a solo homerun), each at-bat with one man on as two possible RBIs, and so on and so forth. Looking at Jeter's chart, we see that he has 355 at-bats with one RBI opportunity, 189 at-bats with two RBI opportunities (113 with a man on first, 48 with a man on second, and 28 with a man on third), and so forth. In total, he has a total of 981 RBI opportunities. Formulaically,

RBI Opportunities = (AB with bases clear) + 2*(AB with one man on) + 3*(AB with two men on) + 4*(AB with bases loaded)

To get a ratio to compare, we simply need to divide total RBIs by total RBI opportunities, giving us an RBI to RBI opportunity ratio (which, for simplicity, I will refer to as RBIA, or RBI Average). Comparing Jeter and Morneau, Morneau still stands above Jeter considerably with an RBIA of .133 to .099. By multiplying the RBIA with a set number of RBI opportunities. Morneau and Jeter actually had 981 RBI opportunities apiece; thereby, it is perfectly comparable in this regard to compare Morneau's 130 RBIs to Jeter's 97.

It occured to me, however, that it might be valid to look at how many of their teammates they were able to bring home. This effectively changes the situation that we are measuring. For one, it measures a player's ability to hit in more "clutch" situations, certainly important when naming the Most Valuable Player. Secondly -- and this may be inadvertent Jeter favoritism -- it takes some of the power factor out of the equation, as we are no longer counting one RBI per home run, taking the players' ability to drive himself in -- only his teammates.

To measure this, we must first calculate Teammates Batted In (TBI). This is simple enough; simply subtract home runs from RBIs. Jeter's 97 RBI and 14 HR leaves him with 83 TBI.

Teammates Batted In = RBI - HR

In order to calculate the total number of TBI opportunities, we revise the RBI opportunities formula, disregarding at-bats with no one on, and subtracting one from each constant. For instance, rather than two RBI opportunities when there is a man on-base, there is one TBI opportunity; rather than three RBI opportunities when there are two men on, there are two TBI opportunities, and so forth. Jeter, for instance, has 358 TBI -- a considerably different number than his 981 RBI opportunities.

TBI Opportunities = (AB with one man on) + 2*(AB with two men on) + 3*(AB with bases loaded)

This changes the comparison between Morneau and Jeter considerably. While their RBI opportunity counts were the same, their TBI opportunity counts are not. Jeter's increased at-bat count from batting higher in the order added to his single RBI opportunities considerably; Morneau's batting later in the order allowed more men on base ahead of him, also increasing his RBI opportunties. With Jeter's added at-bats (and additional single RBI opportunities) eliminated, we see that Morneau had more than thirty more men on to possibly drive in -- Morneau's TBI opportunities stood at 389, while Jeter's stood at 358. In order to compare, we need to divide their TBI by their TBI opportunties -- the ratio leaving us with what I will call the Teammates Batted In Average, or TBIA for short.

Jeter and Morneau's TBIAs are much more comparable than their RBIAs are. Jeter stands at .232, while Morneau stands at .247. If you multiply Jeter's TBIA by Morneau's TBI opportunities, you get 90.25 TBI -- the amount of TBI Jeter would have earned if he had hit comparably with Morneau's chances. Comparitively, Morneau had 96 TBI.

This is by no means a complete look at an objective look at RBIs. I feel as though a number of factors have not been addressed. Some, I believe, are not addressable, like the speed of one's teammates on base -- for instance, having Jorge Posada on second base and having Jose Reyes on second base are two entirely different situations. Others, however, may be addressable; for instance, TBIA does not value batting in one man on first and one man on third the same way, nor does it value batting in one man on two separate occasions differently than two men at the same time. How this should be addressed, or how to value these situations are, I believe, out of my range of expertise, and I put it forth to other statistic-hungry baseball fans to suggest additions, changes, and other formulas to help take a fresh look at what many unfortunately see as a totally outdated stat.<!-- / message --><!-- Sig Was Here --><!-- edit note -->

Thanks Archer! I'd like to talk to this Groovitude person. :D I've always done a bunch of research on this subject in my own way. Not sure if the formula would be the same as above. And as far as Jeter is concerned ... people are using where Jeter is batting in a line-up is kind of goofy. Don't people realize that there were quite a few games the guy played when he wasn't batting in that spot? What about those occasions? I know Jeter did not do well at all driving in runs when he batted third in the line-up. I don't know how many people realize that.

An RBI is important in the rounded picture of things. Everyone does have those opportunities when they come up for each batter. What a player does with those chances makes the man.

Also ... the slowness of players in front of a batter can hinder things for some. A very good example would be having Ortiz on base in front of Manny. It takes a lot to get the guy around to home in that at-bat Manny will have. Now considering that fact ... Manny was outstanding this past year. According to RBI's vs. LOB in an at-bat, Manny was indeed terrific in what he did in my formula. Ortiz was great also in the RBI's vs. LOB chances he had. If you take into consideration that what David Ortiz did with those chances ... vs. the shift and all ... along with the fact his batting average was somewhat lower because of such shift on him ... Ortiz was amazing.

Total RBI's vs. LOB percentages exceeded his batting average ... which meant he was clutch with those opportunities. His percentage was way higher than his actual batting average - which is saying a lot. Ortiz has done the same throughout his career actually.

I can say I also know how poorly driving in runs in post season has been for Derek Jeter for his entire career. Although he has batted at a relatively high average ... those times he batted with men on base - and what he did with those people on base has been terrible. Derek didn't bring hardy anyone home with those chances he had as one might expect or believe. He has left so many people on base over the years. If you look at his total chances for an RBI vs. how many people he still left on base after his hit can suggest many things - and not all positive about his clutchness of hits - given the situation. Walk-off hits are a factor for clutch situations ... as they are RBI's just waiting to happen for a win. What everyone does with those chances to be clutch in possible game-ending at-bat situations ... is just a small but very important factor also.

groovitude
12-04-06, 01:22 PM
... people are using where Jeter is batting in a line-up is kind of goofy. Don't people realize that there were quite a few games the guy played when he wasn't batting in that spot?
Jeter had 623 total AB, 541 of them batting in the two hole. Only about 13% of his AB were not, and all of them were from the three hole. Comparitively, Morneau never batted higher in the order than the five hole, and considering Mauer and Cuddyer were batting in front of him for the extent of the season, it makes a considerable difference in given RBI opportunities. It's not an entirely "goofy" argument.


the slowness of players in front of a batter can hinder things for some.
Since there's no way of addressing that within the stat, an analyst would have to take that into consideration. Your use of Manny as an example is excellent and is a prime example of this necessity of alternately looking at the situation. What would his RBI count look like if he had Jose Reyes batting ahead of him, for instance?


I've always done a bunch of research on this subject in my own way. Not sure if the formula would be the same as above.
What would your formula be? Does your RBI/LOB ratio come into this? I'm very interested in having other people's input on a way to objectively quantify RBIs. (I'm currently in college and dorming with a hardcore Twins fan, so our debates before the revealing of the AL MVP sort of spurred this interest on -- when it came to the RBI count, my feeling was that it was absurd to try to compare the two when Morneau batted behind Mauer and Cuddyer, and Jeter hit primarily from the two hole.)

JL25and3
12-05-06, 06:20 AM
A home run is always an RBI opportunity, but is not always a TBI opportunity. Therefore, we exclude it from the TBIA and allow for the possibility of it to be factored in later.

I understand that. I just don't get what value TBI opportunities has over RBI opportunities. A run is a run, however you bring it home, and power is a crucial part of that. I get that you're measuring something different, but conceptually, I don't get why.


I don't know about you, but I have a much easier time believing that Morneau would hit 107 RBI rather than 114 RBI from the two hole on the Kansas City Royals. I feel that adjusting RBI by TBIA is more accurate rather than by RBIA, and thereby giving cause to leaving out HR from TBIA.

I'm not sure it's appropriate to adjust a formula because the result doesn't "look" riht, especially in a hypothetical situation. In this case, you've adjusted it by counting how many RBIs Morneau would have had without the home runs. But wouldn't he still have hit home runs? Don't they still count as RBIs?


Because the point is to compare how well the player did at driving in runs when there were runners on base, not hitting HRs.

Again, I get this, I just don't get why. I understand the value of looking at RBIs as a function of opportunities. But if RBIs are to be made a useful measure -- which was the original point - I think power is a basic part of that equation. Driving in a runner isn't that different from driving in oneself.

OneRedSeat
12-05-06, 11:35 AM
BP has RBI Opportunities as one of its reports on the statisitics page. You need to be a subscriber to view the report, but it is quite useful. It breaks down how many runners on each base the player had during ABs for the season, and it calculates the RBI/RBIOpp rate. It uses TRBI (total runners batted in) and TBR (total base runners). This way it gives you the non-HR RBI rate.

I think it's good to keep HR out of the equation (and good to neglect the batter as an RBI opportunity). Every RC metric includes Total Bases in its computation (as well as all other non-hit/non-out results and some incorporate the running game). So, with RC metrics we get a close approximation of the context independent value provided by each plate appearance. RBI rates add context to the comparison of value. So, like usual, you need to look at more than one statistic in order to figure out what value the player provided.