Age Effects and Melky Cabrera

I know a lot of Braves fans (and media folks) were underwhelmed when the Braves traded Javier Vazquez and all they got back in terms of Major League talent was Melky Cabrera. I mean, Melky Cabrera? The same guy who has never been an above-average hitter in 4 seasons? The same guy who has put up these lines in his career?

Looks like just another overhyped Yankee prospect, right?

Looks can be deceiving, however. If Melky were a 28-year-old, it would be easy to dismiss him as nothing more than a complementary piece on a contender. He's only 25, however, and he actually played most of last season as a 24-year-old. It's easy to forget how young he is, since it seems like he's been around forever. It's also easy to forget just how big of a difference there is between a 24-year-old and a 28-year-old, in baseball terms.

To see just how big of a difference, I decided to see how some comparable players had done in their age 24 and age 25 seasons. In other words, I wanted to see how much improvement (if any) we could expect from Melky in 2010, his age 25 season. So, I searched Baseball-Reference for similar players from the last 20 years (1990-2009). I defined a "similar player" as someone who had qualified for the batting title as a 24-year-old while starting at least 10 games in centerfield, and who then qualified for the batting title again as a 25-year-old. I came up with 21 names:

Andruw Jones, Bernie Williams, Brian McRae, Carlos Beltran, Chad Curtis, Coco Crisp, Derek Bell, Grady Sizemore, Jacoby Ellsbury, Johnny Damon, Juan Pierre, Magglio Ordoñez, Marquis Grissom, Preston Wilson, Raul Mondesi, Ray Lankford, Rich Becker, Sammy Sosa, Shannon Stewart, Terrence Long, and Vernon Wells.

I compiled all of these players' statistics in their age-24 and age-25 seasons and got these averages:

So, on average, these players got slightly better at just about everything in the hitting arena. I did not find any evidence of an improvement (or decline) in the defense of these players, so I left that out of my analysis. Here's a visual of the hitting improvement:

Those % improvement numbers are relative to the league averages for the past 20 seasons (which just so happen to be almost exactly the league averages from last season). The age-24 seasons (in blue) are a bit better than average in every category but walk rate, but well below average in that category. In the age-25 seasons (in red), the players improved to well above average in power (ISO), and nearly league-average in walk rate. Their batting averages and strikeout rates improved slightly as well.

If Melky improves by the average amounts that these centerfielders improved, this is what his 2010 line will look like:

Not bad, right? That line is better than most projections have for him. In fact, CHONE is the only projection system I can find that doesn't expect Melky to get slightly worse in 2010. I know why--his 2008 was so bad that it skews their data--but it still seems odd, given how these centerfielders have generally improved in their age-25 seasons.

In fact, I think that projection above may be understating Melky's potential, perhaps significantly. The reason for this is that many of the players in the sample above were already stars at age 24 (A. Jones, Wells, Beltran, Sizemore, etc.), and several others were not stars, but had already peaked as regular players (Becker, Long, Curtis). Those two sets of players will frequently decline in their age-25 seasons simply because of a regression to their natural skill levels.

Introducing the "Melkalikes"

What I'm really searching for here is the likelihood that a player like Melky--who has not had a breakout year--would improve significantly and become an above-average player. So I eliminated all the players who had "broken out" in their age-24 seasons. I defined a "breakout" year as having either a .360 OBP or a .450 SLG. This eliminated 11 of the 21 players in my data set, leaving these 10 comparable players:

Bernie Williams, Brian McRae, Coco Crisp, Derek Bell, Jacoby Ellsbury, Johnny Damon, Juan Pierre, Magglio Ordoñez, Marquis Grissom, and Ray Lankford.

Now, obviously you could debate just how "comparable" a few of these guys are--Pierre, especially, is a different sort of player than Melky. But overall I think this gives us a good idea of the range of possibilities for Melky's career. And it has to be encouraging to us Braves fans, since all of these players went on to have long and productive careers (or look like they will, in the case of Ellsbury). Remember, none of these guys were very good at age 24. The amazing thing is that all of them improved at age 25, many of them significantly. How much did they improve? Here's their average stat line:

Compare that age-24 line to Melky's age-24 line. It's very similar (except Melky is slightly better in most categories). In fact, these players are so similar to Melky that I have dubbed them the "Melkalikes." The Melkalikes improved by leaps and bounds in their age-25 season almost to a man. In fact, of the 10 players, 7 had what I defined above as a "breakout" year--.360 OBP or .450 SLG.  Here's the graph of their improvement:

So these players at age 24 were well above average in K% and slightly above average in AVG, but well below average in ISO and basically inept in BB%. That sounds a lot like Melky, except that Melky's ISO and BB% aren't that bad. Yet, in their age-25 seasons, these players were much, MUCH better. They improved their already-good AVG and K% significantly, but even better, their ISO rose to league average and their BB% shot up to respectable levels, though it was still well below average.

Here's what Melky's line would look like if he improves in 2010 as much as the Melkalikes did in their 2010 season:

That, my friends, is the line of a star player. Not an all-star, maybe, but just short of it. Here's what Melky's graph would look like with the same improvements:

Above-average power. Above-average patience. Phenomenal strikeout rate leading to a very good batting average. Could Melky really be this good in 2010? Definitely. In fact, based on the Melkalikes, he could be even better. If he is anything like his 10 comparable players, he is a great bet to improve in 2010; the only question is how much. Obviously, I'm not saying that this will happen. It's impossible to predict when or if a player will break out. But if you look at the track record of these other players, it sure seems likely that Melky will be better in 2010 and beyond.

Below, I break down the improvements of the Melkalikes in 7 categories and choose the one player who most reminds me of Melky.

Batting Average

Of the 10 Melkalikes, all 10 improved their batting average in their age-25 years. The improvements ranged from 3 points to 59 points. 8 of the 10 improved by at least 18 points, and 4 improved by at least 40 points. None of the Melkalikes hit .300 in their age-24 year, but 6 did so in their age-25 year. Here is a chart showing each player's improvement:

On-Base Percentage

Of the 10 Melkalikes, all 10 increased their on-base as well. The improvements ranged from 1 point to 70 points. 8 of the 10 improved by at least 19 points, and 4 of the 10 improved by at least 40 points. None of the Melkalikes had even a .350 on-base in their age-24 years, but 6 of them did in their age-25 years (and another was at .349). Here is the chart:

Slugging Percentage

Again, all 10 Melkalikes improved their slugging. In fact, all raised their SLG by at least 19 points. 4 of the 10 raised their SLG by at least 50 points, with a high of 105 points. None of the Melkalikes slugged .450 at age 24, but 6 slugged that high or better at age 25. Here is the chart:

Walk Rate

Now, you may be saying, "On-base and slugging are both driven by batting average! How much did these players actually improve their patience and power?" That's a valid point, rhetorical audience member! To address your concern, I looked at walk rate and isolated power as well.

As it turns out, 8 of 10 Melkalikes saw a rise in their walk rate (1 stayed the same and 1 declined). The biggest raw improvement was 4.4%, and 6 of the 10 players increased their raw walk rate by at least 2% (that's about 12 extra walks in 600 PAs, by the way). The improvements were generally not quite as dramatic, but no less visible. Here's the chart:

Isolated Power

As with walk rate, 8 of 10 Melkalikes improved their ISO in their age-25 years (again, 1 stayed the same and 1 declined). The biggest gain was 66 points, and 5 of the 10 increased by at least 30 points (that's around 10-15 more extra-base hits in a 600 PA season). Here's the chart:

Strikeout Rate

With that noticeable increase in power, you might be thinking that the Melkalikes must have been swinging for the fences more often, and thus striking out a lot more.  You'd be wrong, though (silly rhetorical audience member!). Only 1 Melkalike dramatically increased his strikeout rate, and 7 of 10 actually decreased their strikeout rate. In fact, 5 of the 10 decreased their raw strikeout rate by at least 2.5% (around 15 fewer K's in a 600-PA season). Here's the chart:

Wins Above Replacement

Finally, we should consider WAR as a measure of the Melkalikes' overall contributions (this includes defense and baserunning, in other words). I took these WAR values from Sean Smith's historical WAR database, since FanGraphs' WAR only goes back to 2002. That means some of these numbers will be a bit different from the values on FanGraphs, though not dramatically so.

At any rate, 9 of 10 Melkalikes increased their WAR in their age-25 season, 5 of them by more than 3 wins. In addition, Bernie Williams' age-25 season was the strike year of '94; if the season had played out normally, his WAR increase would have been even greater. Here is the chart:

Ellsbury's decline was due almost entirely to his defense dropping off a cliff in his age-25 season (last year). On the other hand, Grissom's defense and baserunning were off the charts in his age-25 year. On the whole, though, it is clear that a typical Melkalike played significantly better at age 25 than at age 24.

Conclusions

Obviously, none of this is a guarantee of Melky's future success. My sample is small, and subject to selection bias. In other words, by picking players who had not played well, I was naturally selecting players who had played beneath their skills, and thus would regress upward. My study also does not include players who were so ineffective in their age-25 years that they did not play regularly (Chris Young, Gregor Blanco, and Willy Taveras are a few of these). Finally, these numbers are not adjusted for era or park factors. So those are my caveats.

Melky could easily prove all the projections (but mine and CHONE's) right and play slightly worse in 2010. However, I think we can draw a few conclusions from this data:

1. Players who have showed enough promise to play regularly at age 24, but who have not yet fulfilled that promise, are nonetheless very likely to become productive players.
2. Many of these players become productive at age 25, and even the ones who don't do so often improve somewhat.
3. This improvement is often across-the-board and not just driven by a higher batting average.
4. Based on his age alone, Melky Cabrera stands an excellent chance of improving in 2010, and a good chance of improving dramatically. (Of course, there are many other factors at work, since Melky is changing leagues, lineups, and home ballparks.)

Personally, I think that the league change will help Melky and the lineup/ballpark change will hurt him, so that those effects will more or less cancel each other out. I think that the two projections above provide a reasonable range of expectations for Melky's output this year--somewhere between 2 and 4 WAR, with a line of around .290/.360/.450, give or take. You could probably make an argument that Melky will get worse in 2010, but given the data above I'd find it hard to believe that argument.

Season at Bernie's

One last thing. You may have noticed Bernie Williams' name in the above charts. In addition to being probably the best player of the Melkalikes, Bernie is also probably the closest comparable to Melky. Just superficially, they were both Yankees, both CFs, and both switch-hitters. Both were known more by their first names than their last names and both have first names that sound like Coxian nicknames (I wonder if Bobby's disappointed that he can't add "-ie" to "Melky"?). And of course their age-24 stat lines are very similar:

Like Melky, Bernie played semi-regularly at a very young age (22) but didn't really show much... until his age 25 year, of course. At age 25, Bernie was good. At age 26, he was great. For most of the next 10 years, he was a cornerstone of the Yankees dynasty, helping lead them to 4 titles.

This is obviously a best-case scenario for Melky. I wouldn't suggest that it is likely that he turns into a borderline hall-of-famer like Bernie (in fact, if you look closely at their first few seasons, Bernie's are clearly a bit better overall). The point is that it is possible. Until Melky fails to improve in the next two seasons, we should not give up hope that he will become a star. He is young, and young players with his potential very often break out even after many short-sighted folks have given up on them.