BWAA MVP: RBI and PA w/ Runners On; and some Astros RBI data too
With the recent discussions on the inaneness with which BWAA MVP's are often chosen, I figured it'd be worth it to drive home a point with some statistics. Ryan Howard received a ludicrous twelve first place votes, mainly because he had a staggering 146 RBI. But RBI isn't a good indicator of a hitters ability, it's just an indicator that he had guys who got on base in front of him and because he succeeded in putting the ball in play sometimes, drove them in. Want to call B/S? Hold off for a few paragraphs.
For Howard, 2008 provided him 353 PA with ROB and total of 483 ROB in those PA, meaning -- if we exclude the RBI Howard collected by driving in himself via HR, a stat known as Others Batted In (OBI) -- Howard converted around 20% of his potential OBI in to OBI, or he had 20% OBI% (OBI% click for definition). That's actually a hefty amount, and good for first in the NL among batters with a minimum of 500 PA.
But was it the fact that Howard lead NL regulars in OBI% that helped him accumulate RBI, or was it that was almost as high on the list for PA with ROB or just tROB in general? To figure it out, I ran correlations for PA with ROB, tROB, and OBI% on OBI for MLB in 2008. The data set came from Baseball Prospectus Sortable statistics (as has just about all of that stats I've thrown out so far).
These were the results:
While OBI% does seem to exert some influence on OBI, and therefore RBI, clearly having more PA with ROB and, actually slightly more so, having more ROB drives OBI and RBI production -- almost entirely so. To rephrase that: RBI has almost everything to do with opportunity -- rather than skill.
Ipso facto: RBI should never be used as justification for someone being an MVP. Saying that is essentially saying that the guys in front of Howard (usually Rollins and Utley) were the most valuable.
Playing around with some the Astros data from 2008, our top 10 OBI producers, sorted by OBI% looked like this:
Pretty disparate results. Miguel Tejada was dismal at converting RBI opportunities, while Mark Lorretta was clearly an incredibly valuable PH.
As a team, here's what our correlation coefficients looked like:
Spot on, just about, with the MLB-wide data. No skill: just opportunity.
Comments
Interesting
This is interesting, but can you run similar correlation checks for everything to RBI? Sure, anyone would expect runners on base and plate appearances with runners on base to drive OBI. Without them, your OBI would be 0 and the correlations wouldn’t exist.
What I think is more interesting is that the Astros OBI% had a higher correlation to OBI than the MLB average. Where does that stack up among all teams? Do the teams that have a higher correlation of OBI% on OBI make it to the playoffs?
by JEH629 on
Nov 21, 2008 2:04 PM CST
reply
actions
0 recs
RE: OBI% and playoff chances
That was kind of the point of the exercise. OBI% doesn’t have a significant correlation with RBI or OBI, but with Runners on Base. So the question your asking is best asked by seeing which teams had the best OBP and whether they made the play offs:
Red Sox: 1st in MLB in OBP
Cubs: 2nd in MLB in OBP
Rays: 10th in MLB in OBP
Dodgers: 14th in MLB in OBP
Phillies: 16th in MLB in OBP
Angles: 18th in MLB in OBP
Brewers: 21st in MLB in OBP
Astros: 23rd in MLB in OBP
The Crawfishboxes
A good friend of mine used to say, "This is a very simple game. You throw the ball, you catch the ball, you hit the ball. Sometimes you win, sometimes you lose, sometimes it rains." Think about that for a while.
by DyingQuail on
Nov 21, 2008 3:57 PM CST
up
reply
actions
0 recs
I can't do it for all of MLB
Because I don’t possess the database skills necessary to merge each player’s RBI with the data set from BPro (which doesn’t include RBI).
I went through and manually entered the Astros RBI and reran the correlations"
PA w/ ROB on RBI: .9623
ROB on RBI: .9656
OBI% on RBI: .4030
The caveat is that comparing OBI% to RBI isn’t an exact apples to apples comparison. The trend, however, holds.
The Crawfishboxes
A good friend of mine used to say, "This is a very simple game. You throw the ball, you catch the ball, you hit the ball. Sometimes you win, sometimes you lose, sometimes it rains." Think about that for a while.
by DyingQuail on
Nov 21, 2008 3:43 PM CST
reply
actions
0 recs
Nice effort...but I don't think the correllation comparisons prove anything.
And I preface that criticism by saying that I agree with your overall point, i.e., RBIs are not a great measure for MVP. RBIs are influenced by too many things outside the batter’s control, like teammates’ performance and batting order position.
But all of the variables you compare above, ROB, PA w/ROB, and OBI, have too much overlap and cross correllation to tell us much.in terms of simple correllation. I think you have to perform multiple regression in order to identify proportionate causation. In fact, it might be interesting to see how OBI correllation compares to correlations for other stats like OBP, BA, SLG, etc., when you control for runners on base.
As for the individual Astros’ numbers, it raises some interesting questions. Most significantly, why did the Astros put the batter with the worst OBI% in a position to bat with the second most runners in scoring position? These numbers suggest that batting Tejada third in the order was a significant mistake. Perhaps the Astros would have gained more runs by batting Berkman in the 3d spot and Lee in the 4th spot.
by clack on
Nov 22, 2008 9:02 AM CST
reply
actions
0 recs
Definitely
Your points about the cross correlations are true, but my pitiful spreadsheet program (Numbers) doesn’t have the capability to perform them.
An interesting trend I did note after playing with the data, was that if I limited the data set to players with more than 500 PA (getting rid of pitchers, PH, and non-regulars) the correlations for PA with ROB and tROB dropped to .83, while OBI% increased to .81!! I know that the heteroskedacity issues are still present (I believe this is heteroskedacity, but it’s been two years since I took stats), but that made me reevaluate the claim. Dealing with players who get decent sample sizes of PA with ROB, OBI% exerted a good deal of influence on total OBI.
I’m going to get Office on my comp over Christmas break and I’ll try to redo the analysis at that time.
The Crawfishboxes
A good friend of mine used to say, "This is a very simple game. You throw the ball, you catch the ball, you hit the ball. Sometimes you win, sometimes you lose, sometimes it rains." Think about that for a while.
by DyingQuail on
Nov 23, 2008 8:58 PM CST
up
reply
actions
0 recs
I thought you made up the word heteroskedacity, but now I see that it’s real.
by AstroAndy on
Nov 24, 2008 8:59 PM CST
up
reply
actions
0 recs






















