This week’s question comes from N.J., who writes:

I’m going to keep writing until you answer or persuade me this is a stupid

I’d like to see a metric that evaluates what a player has actually
contributed toward winning, not just how good the player is. For example,
Sammy Sosa has been a terrific player this year, the second-best
hitter in the NL. But how much has he helped actually win games? Have his
hits come in wins or losses? Has he driven in runs when it counted, or in
12-1 blowouts? Has his ability to take walks kept rallies alive, or merely
lengthened scoreless innings? The MVP should be about value, not just

Thanks for the question, N.J.

The reliance on the context-free assumption leads to a lot of consternation
among those relying on more traditional measures of player performance. In
some ways, this is understandable. To the extent that more traditional
measures like runs and RBI help record chains of actual events that lead to
scoring in real games, they are grounded in a concrete reality the way that
context-free models cannot be.

Make no mistake–I’m not abandoning my sabermetric principles. But N.J.’s
question gets to the heart of why many baseball fans still distrust the
"advanced" statistics. They don’t directly relate back to the
events and situations that occurred in the actual games. Runs and RBI
capture, however imperfectly, specific results that determine whether games
are won or lost.

Sabermetrics has its own versions of this approach, ways of looking at the
actual game situations faced by batters and computing the change in the
probability of winning the game as the result of that plate appearance.
Wolverton’s Relievers’ Run Expectation Report
here on Baseball
does this to some extent, considering the base/out state a
reliever faces when entering the game. Tom Ruane has done extensive work
analyzing play-by-play data and coming up with each player’s Value Added,
based on work originally done by Gary Skoog.

We can take a similar approach with the cruder measures found in box scores.
Bill James used a statistic dubbed "Victory-Important RBI" in some
of his early Baseball Abstracts to help address questions very
similar to N.J.’s. I went down a similar path in researching an answer to
the question, until BP colleague Rany Jazayerli pointed out that I’d
basically reinvented work done 20 years ago. However, there is one
significant difference between the approaches James and I took: I also used
runs scored instead of just RBI. Both approaches are similar in how they go
about assessing which players have actually had the most impact on winning
games without reliance on a context-free model.

The goal is to measure the "impact" each player actually had
during the season. Runs in close games mean more than runs in blowouts,
since each is more important to the outcome. Similarly, losses have no
value, so any runs scored in losses do not count.

We’ll measure a player’s production by Runs Produced (R+RBI), thus focusing
on runs that actually score, rather than potential runs. [Side note: Please
don’t fall into the trap of subtracting home runs from R+RBI, as is
sometimes suggested when Runs Produced is brought up]. I actually use the
average of runs and RBI, thus (R+RBI)/2, to get a number that is on a more
familiar scale, and avoids having to deal with double-counting runs. Now we
have to determine how many of these runs actually helped the team win: his
Contribution to Winning, or simply Contribution.

If the player’s team loses, his Contribution is zero, as the collective
efforts of the team failed to reach the desired goal (a win). If the
player’s team wins, his Contribution is his fraction of the team’s runs
produced in that game, scaled to the minimum margin of victory needed. If a
player has two runs and three RBI, he has 2.5 Runs Produced (RP). If his
team wins 8-3, his production represents 2.5/8 = 31.25% of the team’s
output. But since the team only needed four runs to win, the extra runs are
superfluous, so the team as a whole only gets credit for four runs. Thus our
player gets credit for a Contribution of 31.25% * 4 runs = 1.25 runs.

In addition to the Contribution, we’ll also show a value called
"Importance," defined as Contribution / Runs Produced. This value
shows what percentage of a player’s R+RBI actually impacted a win (as
defined in Contribution). By itself, this value isn’t that telling–a player
with many RBI will undoubtedly accrue a number of them in losses and
blowouts. However, since Bill James included such a calculation with his
Victory-Important RBI, I will do so for completeness.

There are obvious sabermetric problems with this as an evaluation method,
such as the aforementioned context dependence, positional adjustments,
lineup order effects, runner advancement not captured by R and RBI, park
effects, the fact that not all runs have a corresponding RBI, and more. But
given this week’s question, it isn’t so bad as a record of impact–whether a
player contributed to actual winning games or not. And, although it’s not
the sabermetric way to answer the question, it is arguably a rational way to
use traditional stats to help determine the MVP.

On to the stats. In the tables below, I show the team’s record in the games
where the player appeared, the player’s Runs Produced Total, his
Contribution (which could also be called Victory-Important Runs Produced,
following the James terminology), and Importance. Totals are through
September 25.

Top 20 AL Batters

Batter              Team    G     W    L      RP Contrib Import

Bret Boone SEA 148 107 41 122.0 59.40 .4869 Juan Gonzalez CLE 135 80 55 118.5 56.25 .4747 Ichiro Suzuki SEA 149 105 44 93.5 49.19 .5261 Corey Koskie MIN 142 74 68 96.5 49.03 .5080 Garret Anderson ANA 150 74 76 94.0 48.56 .5166 Jim Thome CLE 146 84 62 110.0 47.90 .4355 Alex Rodriguez TEX 152 70 82 128.0 47.49 .3710 Mike Cameron SEA 140 99 41 99.0 47.13 .4760 Edgar Martinez SEA 126 87 39 93.5 47.11 .5039 Bernie Williams NYA 136 84 52 89.5 47.01 .5252 Roberto Alomar CLE 149 86 63 104.5 46.82 .4480 Tino Martinez NYA 145 88 57 98.5 45.93 .4663 Jason Giambi OAK 144 87 57 104.0 42.77 .4112 Troy Glaus ANA 150 74 76 95.5 42.27 .4426 Paul Konerko CHA 146 76 70 91.0 41.86 .4600 Magglio Ordonez CHA 149 76 73 98.5 41.06 .4169 Derek Jeter NYA 140 81 59 86.0 40.87 .4752 Miguel Tejada OAK 151 93 58 100.0 40.64 .4064 Eric Chavez OAK 140 85 55 94.0 39.11 .4161 Manny Ramirez BOS 140 69 71 108.0 38.79 .3592

Hopefully, this article will appease some of the Bret Boone fans I
offended over on when I looked at how big a fluke Boone might turn
out to be this year. Boone has obviously been an essential piece of the
Mariners’ Cinderella season, and his presence atop this list testifies to
how often he’s produced in games they’ve won. Of course, because the
Mariners won so many games, it’s not surprising that several of their
players appear on a list that only credits players for production in wins.

Ichiro fans will be pleased to see him ranked higher on this list
than on most sabermetric rankings. Garret Anderson is probably the
biggest beneficiary under this system. He’s fourth in the AL, yet by a
sabermetric measure like Marginal Lineup Value (a component of VORP), he
ranks fifth…on the Angels. In the AL, Anderson ranks 70th.

Top 20 NL Batters

Batter              Team    G     W    L     RP Contrib Import

Jeff Bagwell HOU 150 89 61 120.5 57.88 .4804 Todd Helton COL 149 66 83 131.5 54.72 .4161 Barry Bonds SFN 143 76 67 122.0 53.87 .4416 Shawn Green LAN 152 81 71 116.0 51.29 .4421 Albert Pujols SLN 151 86 65 115.5 49.29 .4267 Ryan Klesko SDN 135 67 68 103.0 49.06 .4763 Sammy Sosa CHN 149 79 70 139.5 48.52 .3478 Luis Gonzalez ARI 151 84 67 125.0 47.13 .3770 Rich Aurilia SFN 146 80 66 99.0 46.22 .4668 Chipper Jones ATL 149 79 70 99.0 46.13 .4660 Lance Berkman HOU 145 84 61 113.0 46.01 .4071 Scott Rolen PHI 144 77 67 95.0 44.19 .4652 Bobby Abreu PHI 151 80 71 105.5 44.00 .4171 Phil Nevin SDN 140 71 69 101.5 43.11 .4247 Gary Sheffield LAN 133 69 64 89.5 43.07 .4812 Larry Walker COL 133 58 75 110.5 43.05 .3896 Moises Alou HOU 129 77 52 89.5 42.30 .4726 Mike Piazza NYN 132 67 65 82.0 42.28 .5156 Vladimir Guerrero MON 149 64 85 101.0 40.75 .4035 Craig Biggio HOU 144 85 59 86.5 39.97 .4621

Jeff Bagwell certainly hasn’t been getting the MVP attention that
Barry Bonds, Luis Gonzalez, and Sammy Sosa have been getting,
but if you are looking for tangible impact on games won, Bagwell outshines
them all. Craig Biggio moves from 51st in MLV to 20th on the list
above. Most hurt is Brian Giles, one of the best hitters in the NL
(ninth in MLV), who doesn’t appear in the top 20, as the Pirates as a team
have won fewer games than Bagwell’s Contribution alone.

Keith Woolner is an author of Baseball Prospectus. You can contact him by
clicking here.

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe