Playing the numbers

We hear a lot about the triumph of advanced baseball statistics. We hear about teams using advanced stats, of course, but we also hear about how advanced stats have changed the way the game is covered and analyzed and rewarded. Take awards. In the old days, we all know, cigar-chomping sportswriters chose their Most Valuable Players based essentially on, you know, RBIs.

Let’s look at the American League MVPs in the 1970s just based on RBIs:

1970: Boog Powell, RBI leader on best team

1971: Vida Blue, Pitcher

1972: Dick Allen, overall RBI leader

1973: Reggie Jackson, overall RBI leader

1974: Jeff Burroughs, overall RBI leader

1975: Fred Lynn, RBI leader on best team

1976: Thurman Munson, RBI leader on best team

1977: Not RBI-related (Rod Carew hit .388)

1978: JIm Rice, overall RBI leader

1979: Don Baylor, overall RBI leader

So, using just RBIs and a little bit of savvy, you could have picked eight out of the 10 MVPs in the 1970s. It was like that. But, you might have heard, we have moved past all that. Sportswriters don’t chomp on cigars now, and we don’t care much about RBIs. Well, we understand that RBIs are really team statistics, not individual ones. We understand that, right?

Here are your last four American League MVPs.

2012: Miguel Cabrera, overall RBI leader

2013: Miguel Cabrera, one behind in RBI race and on playoff team

2014: Mike Trout, overall RBI leader

2015: Josh Donaldson, overall RBI leader

Huh, funny, that doesn’t look any different than the 1970s. I go back four years because by basically every single advanced statistical measure — Fangraphs WAR, Baseball Reference WAR, Baseball Prospectus WARP, Bill James Win Shares as our examples — Mike Trout has been the best player in the league each of the last four years. He has swept the stats each and every one of those four years.

And yet … he’s won just one MVP award.

As you probably know, this year, Toronto’s Josh Donaldson won the American League MVP award over Los Angeles’ Mike Trout, and there was a lot of angst. I voted in that election, too, and when I read that pal Ken Rosenthal of Fox Sports actually went as far as writing down Donaldson’s name a couple of times in the top slot before finally shifting at the last minute to vote Trout, well, it sounded familiar. My experience was eerily similar.

But, looking back after the vote and AFTER all the hype and narratives wound down, I realized something. It wasn’t that close at all. Mike Trout got totally screwed by the voters again.

Now, I should say: I don’t think that Donaldson actually won the award because he led the league in RBIs. And I don’t think that Donaldson won the award because his Blue Jays made the playoffs while Trout’s Angels did not. I’m sure both of those things played a role, but I don’t think those were the key things.

I believe the key thing was this: The two most influential advanced stats — Fangraphs and Baseball Reference WAR — made it seem like it was a toss-up between the two players.

Fangraphs WAR:

Trout: 9.0

Donaldson: 8.7

Baseball Reference WAR

Trout: 9.4

Donaldson: 8.8

Yes, Trout was slightly ahead. But the difference was almost nothing, especially in Fangraphs WAR. The difference was well within the margin of error, and my suspicion is that because it was SO close, the voters looked elsewhere for guidance. Donaldson led the league in RBIs and he also led the league in Win Probability Added, a fun statistic that breaks down each play and determines how much that player added to the the team’s chances for victory. The knock on WPA has been that, like RBIs, it is driven by the other members of the team. But some don’t see that as a knock.

In any case, when you have a toss-up — like this Trout-Donaldson matchup seemed to be — then maybe it’s cool to go looking for team context statistics.

But the question is — was it REALLY a toss-up? Here are those other two advanced statistics I mentioned:

Baseball Prospectus WARP

Trout: 10.0

Donaldson: 7.6.

Win Shares

Trout: 42 win shares

Donaldson: 32 win shares

Funny: Those aren’t close at all. Both estimated Trout to be a 33 percent-better player than Donaldson.

So what gives? The two most influential stats have them neck-and-neck while WARP and Win Shares had Trout being the runaway better player — how is that possible? The answer, as you will see, comes down to those murky estimations of defense.

See, all four statistics calculate Trout having a much better offensive season than Donaldson. It’s easy to miss that because Donaldson had more RBIs and total bases. But Trout had 31 more points in on-base percentage, 22 more points in slugging percentage, and he played half his games in that hitter’s graveyard called, bizarrely, Angel Stadium of Anaheim. When you play in a pitcher’s ballpark, like Angel Stadium of Anaheim, each run is worth more.

How much of a difference was there between the two? One thing that’s always good to do when comparing players is to compare their road numbers. It’s not a perfectly neutral comparison, but it’s closer than comparing their home ballpark.

Here were Trout and Donaldson’s road numbers in 2015:

Trout: .316/.400/.623, 20 doubles, four triples, 21 homers.

Donaldson: .263/.343/.487, 16 doubles, one triple, 17 homers.

That’s not close at all. Trout’s road OPS+ — which compares his OPS  to those across the league — was 187. Donaldson’s was 134. Again, not close.

And the statistics all agree: Trout was the significantly better hitter. Fangraphs, Baseball Prospectus and Win Shares have Trout as roughly 25 percent offensively. Baseball Reference has Trout as about 16 percent better — I’m not sure why their estimate is a bit lower. Still, nobody disagrees: Trout was way better with the bat.

Ah, but they really disagree on defense. Baseball Reference has Donaldson as a better defender than Trout. Fangraphs has Donaldson as a MUCH better defender. Baseball Prospectus and Win Shares, meanwhile, have Trout as the better defender.

So who is right? Well, it’s hard to tell, of course. One played third base, the other played center field. Their home ballparks are very different. We used to look at errors as a way to differentiate fielders — Donaldson made 18 errors, Trout made zero. We used to look at range factor — Donaldson’s range factor was 10 percent better than the league average, Trout’s was 30 percent higher.

John Dewan breaks down every fielder play in every play of every game — it’s probably the most involved and time-intensive system around —  and it estimates that Donaldson saved 11 more runs than the average third baseman and Trout saved five more runs than average.

Which leads us to … who knows? Maybe Donaldson was better. Maybe Trout was better. I guess you have to ask yourself: Do you really think that Josh Donaldson was THAT MUCH BETTER a defensive player than Mike Trout that it makes up for the sizable gap in offense? If you think yes — if you agree with the Fangraphs estimate — then, yes, Donaldson had a strong MVP case.

I don’t see it. I respect the Fangraphs math, but I don’t believe a good third baseman could make up that much ground on a good centerfielder. Donaldson was rated a slightly better baserunner too, but again, not enough to tilt a dominant offensive advantage.

In the end, I tend to believe that while we talk a good game about being more analytical, we’re really doing more or less the same stuff. Mike Trout was, by just about any measure, the best player in the league in 2012, 2013, 2014 and 2015. And while we can try to wrap it up in coats of many colors, the bottom line is that three times we gave the MVP award to someone with more RBIs.

Scroll Down For: