Is the AL winning the talent war?

Ten of the last fourteen World Series have been won by an AL team; during this period All-Star games were even more one sided, with the AL going 11 to 3 with one tie; in the last eight World Series the AL has won 29 games to the NL's 12; and in interleague play last year, the AL went an astonishing 136-116! Wow. Some gap.

Moreover, after the recent bout of off-season trading it looks like our run shy NL ball clubs will receive even more of a pounding in 2006. An analysis in The Hardball Times estimates that 2006 cross league trades led to a net transfer of over 100 win shares of talent from the NL to the AL. That is equivalent to 5 wins per AL team! Woe betides NLers aiming for Series glory. At least solace can be taken from the idiosyncrasies of the playoff system, which guarantees that four NL teams are elevated to the postseason and that the World Series, being the best of only 7 games, is a bit of a crapshoot anyway.

Things can't be that bad, can they? It is time to draw breath and take a more objective look. How can we better understand the relative strength of the AL and NL, and whether the gap between the two leagues has been widening? Well let's scrutinize the evidence that purports to support the talent gap theory, namely:

1. The odds of a particular league assembling a 10-4 World Series record
2. How the two leagues have fared in interleague play
3. Whether the AL has successfully managed to pillage the top NL players

Going 10-4 in the World Series

So the AL has won 10 of 14 World Series - a substantial number, right? Well, not really. Suppose both the AL and NL champs have equivalent talent, in other words they are both .500 teams. Then the probability of either league going 10-4 is 18% (adjusting for home field advantage makes little difference). Statisticians normally work with confidence levels of 0.05 (5%), so if this was a statistical test then we'd reject the hypothesis that a 10-4 World Series record is proof that there is a talent gap between the leagues. What about the results of the individual games of the last 8 World Series, where the AL is 21-12? Again, the probability one team notching up such a record is 12%. This is lower, but still above the magical 0.05 threshold.

Interleague Record

Since interleague play started in 1997 there have been precisely 2199 games, which is almost a full regular season's worth. Hooray, we are in meaningful sample size country, albeit over a 9 year stretch. What are the results? 1104-1095 to the NL! Yep, you read correctly, the NL has an edge - a very thin edge, but an edge nevertheless. Take a look at how the teams rank to see if the NL advantage is a result of a couple of uber-teams or is more broadly spread:

Two AL teams - the A's and the Yankees - romp home at the top. But these two aside the NL takes seven out of the next eight spots. That doesn't look like a league suffering from a talent drain. OK, but recently has the AL been catching up with the NL? Not at all. Take a look at the graph below which shows both yearly and cumulative NL win percentage since 1997, which was when interleague play began.

Sure, the AL has been on top in the last couple of years but there is nothing to suggest a fundamental shift in the balance of power. As recently as 2003 the NL had a 20 game advantage, almost replicating the 2005 AL surplus. Identifying a clear trend is not straightforward.

Trades

So far there isn't a lot of evidence to support the talent drain hypothesis, but it's worth trying a different tack. Looking at players who have transferred between the two leagues, and how their performance changed, could help us understand whether talent drain is real, and if it is to what extent it is accelerating. What exactly are we looking for? The theory is that the AL attracts the best NL players, which results in a growing performance gap between the leagues. Therefore if we look at players who move to the AL from the NL we would expect them to outperform the average NL player in the year preceding the trade; and on the contrary we would expect the AL to be able to retain its best players by shipping out below average talent to the NL.

To properly evaluate this it is important to use the correct statistic. Take pitching for instance. Most know that ERA is a flawed metric to use as it doesn't necessarily reflect true pitcher skill. In order to assess total talent drain the chosen statistic also needs to transcend both hitting and pitching. A good statistic to use is wOBA, which was recently developed by Tangotiger and is discussed at length in The Book. In short it is essentially linear weights per plate appearance adjusted so that it is equivalent to OBA. By measuring the average wOBA for each league and the players traded across leagues for each of the last 10 years, and adjusting for park, we should be able to discern whether significant talent drain has occurred.

First for pitchers:

Remember a positive number means that there has been a net transfer of talent out of the league. This is because hurlers who have a higher wOBA give up more runs! The first thing that jumps out is that every year for the last 10 years the AL has kept its best pitchers (the pitchers that have moved to the NL have a positive differential wOBA). This may be because AL pitchers don't want to move to the NL as they have to learn how to hit - life is easier if you don't have to worry about sprinting around the bases - and good pitchers can choose where they work. Turning our attention to the NL, there is clear evidence of a talent drain in recent years. Since 2001, the wOBA of hurlers leaving the NL for the AL was .10 below average - which over the course of a season equates to 5 runs per pitcher (~400BFP). We do need to be a little careful with small sample sizes, but there is definitely a trend. Now check out the same table but this time for batters:

For sluggers, a positive number indicates a net inflow of talent because a higher wOBA means a player contributes more runs. Again AL teams are remarkably good at retaining talent. Why do AL hitters prefer to stay in the AL? It's difficult to fathom an answer but it could simply be because that is where the money is. Four of the five richest teams (in terms of payroll) are in the AL, and the Yankees and the Sox spend so much they may also distort the picture. What about NL batters? Are they trying to hot-foot it over the fence to shower in the collective glory of the AL? Here the data is a bit more ambiguous. Some years there was an outflow of talent, in other years not.

What is the conclusion then? Things are pretty even. If anything the AL has had the upper hand in recent years but the gap probably isn't as wide as some people claim. If there is one thing that sabermetrics teaches us it is the danger of being blinded by small sample size. Ergo the results of the last 10 World Series, or the last interleague season should be treated with caution. Put your money on a little regression to the mean.