Does Winning the Rookie of the Year in Baseball Lead to an Elite Career?

What can past winners such as 2003 NL Rookie of the Year Dontrelle Willis tell us about recent winners?

Baseball fans have been spoiled with exciting young talent during the past two seasons. The past four Rookie of the Year Award winners, Mike Trout, Bryce Harper, Jose Fernandez, and Wil Myers, have taken the league by storm, posting elite numbers, especially for players of such a young age.

Fernandez and Trout won respective Rookie of the Year honors in their age-20 seasons while Harper took home his hardware at the tender age of 19. It's easy to project seasons of stardom for each of these players, but are these high expectations justified?

The typical career path for Major League Baseball players is a curve, with the peak years coming at roughly ages 26 through 29 and slopes on either side. Robinson Cano is a good example of a player who has followed this career path, as he averaged 2.7 WAR per season during his first four years before blossoming into the superstar he is today. Cano is likely near the end of his peak and should begin the decline phase of his career within the next few seasons.

If most players follow this traditional career curve, then it would make sense that players having elite seasons early in their career will be more likely to have an elite career. In theory, the curve would simply “move up” as young players continue to improve with age. Is this indeed the case?

How have Rookie of the Year Award winners fared after their rookie campaign? Are they bound for stardom or do they often regress? Do players with high WAR totals in their award winning rookie campaign tend to have better careers than those with a lower WAR total? With what degree of certainty can we make any predictions?

To answer these questions, I examined the careers of all Rookie of the Year Award winners since 1990 and looked for answers in the data. They were quite surprising.

What Does It Take to Win the Rookie of the Year Award?

The process for winning the Rookie of the Year Award is simple. A player must have rookie eligibility and perform better (in the eyes of the voters) than all other rookies in his respective league.

Rookie eligibility is defined by the league as follows:

A player shall be considered a rookie unless, during a previous season or seasons, he has (a) exceeded 130 at-bats or 50 innings pitched in the Major Leagues; or (b) accumulated more than 45 days on the active roster of a Major League club or clubs during the period of 25-player limit.

Because of these regulations, winners of the Rookie of the Year Award often accrue service time in one or more seasons prior to their award winning rookie campaign. Geovany Soto, the 2008 NL Rookie of the Year, played in parts of three seasons before his official rookie season. Contrarily, other players such as the Orioles’ Manny Machado extinguish their rookie eligibility in a partial season and consequently are not seriously considered for the award.

Though all major awards have their faults, namely that writers can't seem to agree on the proper way to determine a player’s “value,” the Rookie of the Year Award has an additional complication. Since only a small percentage of players are eligible for the award, the yearly difference in quality among the player pool is often extreme.

Consider the 2013 Rookie of the Year Awards as an example. Even though he didn't post the highest WAR total among AL rookies, Wil Myers was a reasonable choice for the American League Rookie of the Year Award. In the National League, seven players posted higher totals than Myers’ 2.0 mark including a few that were substantially higher.

The voting results and WAR totals of those receiving votes are listed below.

National League Voting Results

RankNameTeamVote Pts1st PlaceWAR
1Jose FernandezMIA142266.7
2Yasiel PuigLAD9545
3Shelby MillerSTL1203.3
4Hyun-jin RyuLAD1003.7
5Julio TeheranATL703.7
6Jedd GyorkoSDP202.2
7Nolan ArenadoCOL103.9
7Evan GattisATL100.6

American League Voting Results

RankNameTeamVote Pts1st PlaceWAR
1Wil MyersTBR131232
2Jose IglesiasTOT8051.9
3Chris ArcherTBR3512.2
4Dan StrailyOAK1511.2
5J.B. ShuckLAA400.9
6Cody AllenCLE201.4
6Martin PerezTEX201.6
8David LoughKCR102.7

This chart proves three things:

1. Awards are not always given to the player with the highest WAR total. WAR is still not a perfect statistic, but it is useful in this exercise as it is a simple way to compare players.
2. The National League was loaded with impact rookies last season. This is probably not news.
3. Since winners are picked from a rather shallow group of their peers, the level of performance necessary to win the award varies greatly from year to year and league to league. If Yasiel Puig, Shelby Miller, Nolan Arenado, Hyun-jin Ryu, Julio Teheran, or even possibly Jedd Gyorko would have played in the American League this past season they would have likely won the Rookie of the Year Award. On the other hand, if Wil Myers would have played in the National League this season he may not have even received votes!

This, along with the human element of letting voters choose a winner, explains the vast difference in performance among winners in their winning season. Mike Trout’s 10.9 mark in 2012 was the highest in this group while Eric Karros won the award with a mere 0.3 mark in 1992. That’s a difference of 10.6 WAR, which is equivalent to the difference between Barry Bonds in 2004 (10.6) and Placido Polanco in 2013 (0.0) or Snuffy Stirnweiss in 1945 (8.6) and Yuniesky Betancourt in 2013 (-2.0).

Necessary Rant About How Eric Karros Was the Wrong Choice for the 1992 Rookie of the Year Award

Eric Karros winning Rookie of the Year with 0.3 WAR is utterly ridiculous. I know that WAR was not available to voters in 1992, but his real statistics do not support his cause any better.

Karros’s debut season was very similar to a typical Mark Trumbo season without as many home runs, which is to say he is a one-tool player whose one tool was merely average for his position. Karros was like Tony Gwynn with an average hit tool, Michael Bourn with average speed, Aroldis Chapman with an average fastball, or R.A. Dickey with an average knuckleball. You get the idea.

He hit 20 home runs. That’s it. His .257 average was fine for a one-tool slugger, heck Adam Dunn has done just fine hitting in the low .200’s for quite some time, but the difference is that Adam Dunn hits lots of home runs each year while Karros hit only 20.

Justin Smoak hit 20 home runs this year while providing a significantly higher OBP, not to mention a higher WAR total. But wait, that same Justin Smoak has been dubbed a bust by many analysts and may be out of a job if the Mariners sign Nelson Cruz, resign Kendrys Morales, teach Jesus Montero how to effectively hit baseballs, or determine that Logan Morrison is incapable of playing outfield. Any one of those four events would relegate Smoak, whose 2013 WAR total was nearly four times as high as Karros’s 1992 total, to the bench or on a bus to Tacoma.

Putting aside the award, Karros should not have even received votes. Through some form of collective delinquency the voters bypassed five (Five!) players who posted at least 2 WAR: Moises Alou, Reggie Sanders, Tim Wakefield, Mike Perez, and Ben Rivera. One would assume that with a large collection of reasonable players from which to choose, Karros’s election must have been due to a fluke in the voting and the result of a very close election.

This was not the case. Karros received 22 of the 24 first place votes! I have no further comment on the matter, other than it's certainly easier for one to understand Immanuel Kant’s Critique of Practical Reason, Dennis Rodman’s recent actions, and how neither the all-time hit leader nor the all time home run leader are enshrined in Cooperstown than it is to understand how 22 writers voted Eric Karros National League Rookie of the Year in 1992.

OK, that’s all on that for now. Back to the numbers.

Can We Predict Success Based on Winners?

We have already seen how the difference between Rookie of the Year Award winners is extremely large, with some players posting WAR totals reminiscent of 1800’s pitchers such as Tim Keefe and Old Hoss Radbourn and others, well, others having less WAR than Switzerland during the entire 20th century.

Our next task is to examine the correlation between winning the Rookie of the Year Award and having a productive career.

This correlation will be examined in three ways.

1. Rookie WAR versus Second Year WAR
2. Rookie WAR versus Second and Third Year WAR
3. Rookie WAR versus Career WAR/Season

The goal with these comparisons is to find out if differences in WAR totals among Rookies of the Year can be used as accurate predictors of differences career success. In other words, will a high WAR total for a Rookie of the Year during his rookie season typically lead to a better career than a Rookie of the Year with a low WAR total during his rookie season?

I will supplement the analysis in the next three sections with relevant examples of Rookie of the Year winners following specific career paths discussed in the section.

The 2012 and 2013 Rookie of the Year Award Winners were omitted from the next three sections.

Rookie WAR versus Second Year WAR

Before calculating data, I typically form a hypothesis about my expectations for the results. For this section, I assumed that there would be a fairly strong correlation between Rookie WAR for Rookie of the Year Award winners in this data set and second year WAR, considering the same player should be somewhat likely to perform at a similar level in consecutive seasons.

That prediction was wildly inaccurate.

An R-squared value, also called the coefficient of determination, is used to find what percentage of the variation in the Y variable, in this case WAR in a player’s second season, can be explained by the X variable, in this case WAR in a player’s award winning rookie campaign.

The R-squared value for this data set was a mere 0.203, which means that rookie WAR can only account for roughly 20 percent of the variance in second year WAR.

This is an extremely surprising result and shows that in this data set there is little power in using rookie WAR among Rookie of the Year Award winners since 1990 as a predictive statistic for WAR in that player’s second season. Subsequent data will either confirm or reject this assertion.

Example 1: Jason Jennings, Colorado Rockies RHP and 2002 NL Rookie of the Year Award Winner.

Jennings had one of the most obscure career paths I have ever seen. In his rookie season, Jennings posted 2.0 WAR, but followed it up with a poor season where he was worth a mere tenth of a win. The next season, Jennings reverted almost exactly to his rookie-level production, posting 1.9 WAR.

Another season of regression saw Jennings post only 1.0 WAR, but he somehow followed that year of mediocrity with a 5.0 WAR campaign. Jennings left Colorado after that season and was never the same again, finishing his career with WAR totals of -0.6, -0.7, and 0.5.

The career path of Jennings includes a sophomore slump and wild inconsistency. I am also enamored by the fact that he had a 5.0 WAR season with the Rockies, then left Colorado and immediately became a below replacement level pitcher. Making predictions about Jason Jennings has proven more difficult than predicting Manny Ramirez's antics, Julian Tavares’s means of delivering a baseball from the pitcher’s mound to first base, or Moe Drabowsky’s pranks from the Orioles’ bullpen.

This example is useful in demonstrating the volatility of some player’s career paths. Jennings had a good rookie season and even a 5.0 WAR season later in his career, but was not able to sustain his infrequent success in subsequent seasons.

Rookie WAR versus Second and Third Year WAR

Continuing with the theme of proving my hypotheses wrong, the correlation between rookie WAR and second plus third year WAR was surprisingly high.

The methodology for this subsection was simple. I summed the WAR totals from a player’s second and third seasons then compared that total to the WAR total from each player’s rookie season.

The R-squared value for this data set rose to 0.334, which is still low, but significantly higher than the total from the previous subsection.

This number means that within this data set, WAR from a Rookie of the Year Award winner’s rookie season can more accurately predict WAR of the next two seasons combined than of only the subsequent season, although it is still fairly low.

Two possible explanations immediately come to mind:

1. Over the course of multiple seasons, fluke performances are likely to have a lesser impact on the data and a player’s true talent level will be present.
2. Players go through a “sophomore slump” in their second season before reverting to dominance in their third season.

There are many examples from this data set to illustrate these points. This section also gives me a chance to talk about some of my favorite Rookie of the Year busts.

Example 2: Geovany Soto, Chicago Cubs C and 2008 NL Rookie of the Year

One could reasonably argue that Soto’s career path didn't feature a sophomore slump, but rather outliers in his first and third full seasons. Soto posted 3.2 WAR in his 2008 rookie campaign, regressed in 2009 to 0.2 WAR, then bounced back in his third season to post 3.1 WAR. He has trailed off in recent years, declining steadily from 2011 through 2013 before rebounding last season.

This is a nice example of someone undergoing a traditional sophomore slump, starting with a very productive season, then having a terrible season, then reverting back to rookie-level production by having a very good season. This was a common trend among players in this sample, though none followed this pattern quite as cleanly as Soto.

Example 3: Raul Mondesi, Los Angeles Dodgers OF and 1994 NL Rookie of the Year

You may be thinking, wait a minute, Mondesi didn't have a sophomore slump, so why is he being featured here as an example? Mondesi is here because he is the exception, not the rule. Of the 44 players examined, only 13 increased their WAR totals from their rookie year to their second season, meaning that the sophomore slump has indeed been an issue for many of the these Rookie of the Year Award winners.

Mondesi follows what I will call the “ideal career path” for Rookie of the Year award winners, at least for his first four seasons. After posting 1.8 WAR in his award-winning debut, Mondesi increased his production to the respective levels of 4.8 WAR, 4.7 WAR, and 5.7 WAR in his next three seasons.

This is the career path often imagined for young stars but unfortunately this has not been the career path that Rookie of the Year Award winners in this study have typically followed.

The data below demonstrate this point.

Rookie WARWAR 2WAR 3
Standard Deviation1.582.192.80

WAR 2 and WAR 3 represent WAR totals for each player's second and third season.

The data shows that, for these Rookie of the Year Award winners, the production level in their rookie campaign is typically better than their production in their second or third seasons. And moreover, the production in their third season is typically better than their production in their second season.

Put more simply, for these Rookie of the Year Award winners, production levels during their first three seasons have been highest in their rookie season, second highest in their third season, and lowest in their second season.

The sophomore slump certainly exists in this data set.

Rookie WAR versus Career WAR per Season

In this section we will examine this correlation between Rookie WAR and average WAR per season for the course of each player’s career.

For the sake of having more accurate and useful data, I omitted any partial seasons prior to each player’s award winning rookie campaign. This means that despite debuting in 2005, Geovany Soto’s entry in my spreadsheet begins with his 2008 season. All other partial seasons for players such as Andrew Bailey, who has missed considerable time battling injuries, are included and considered as full seasons.

Additionally, I omitted each player’s rookie season from the WAR per season figure so that it was not counted twice.

The R-squared value of this correlation is 0.264, which is lower than the R-squared number comparing each player’s rookie season to their second and third season, but higher than the R-squared figure comparing each player’s rookie season to only their second season.

This number appropriately slots between the previous two R-squared values. It seems right that an elite rookie season in this data set is more effective predicting the results of a player’s second and third seasons than predicting the success of his entire career. It also seems right that an elite rookie season in this data set is least effective predicting the success of a player’s second season due to sophomore slumps and general volatility of the career paths of many of these players.

The following two players demonstrate this volatility and show that a Rookie of the Year Award does not imply a great second season, third season, or career. These players are commonly labeled as busts.

Example 4: Bob Hamelin, Kansas City Royals 1B/DH and 1994 AL Rookie of the Year

The other Rookie of the Year in 1994, Bob Hamelin of Kansas City, posted a higher WAR total than Mondesi during the 1994 campaign. Following that season, Hamelin’s production ranged from being good enough to maybe have on an active roster to being appropriately slotted in the seven hole of a local beer league softball lineup. He is the poster boy for the player that had no business having a great rookie season, and immediately reverted to mediocrity immediately following his award winning campaign. Mondesi, meanwhile, went on to have a productive career.

The career paths of these players illustrate the point that a higher WAR during a rookie campaign does not necessarily lead to a more productive career, and vice versa.

A final thought on Hamelin: He posted 2.6 WAR during his rookie season. He also posted 2.6 total WAR during his entire career. This means that Hamelin was worth exactly 0.0 WAR during his entire Major League career apart from his award winning rookie campaign, a span of five additional seasons.

Example 5: Pat Litasch, Milwaukee Brewers SS and 1992 AL Rookie of the Year

Litasch is my favorite example of a Rookie of the Year bust because his decline was so sudden and so consistent. The WAR totals during Litasch’s career are as follows: 4.4, 1.6, 0.2, -0.1, -0.7, and -1.2.

I am fascinated by Litasch’s ability to have a great rookie season then get consistently and significantly worse for the next five seasons. His decline was almost linear! This is to be expected from an aging star but not from the Rookie of the Year.

Litasch’s career path leads us to another point that is somewhat implied but still worth mentioning. To win the Rookie of the Year Award, one must have an excellent season (unless one is Eric Karros), which doesn't necessarily mean that one is an excellent player bound for an excellent career.

Just as great players sometimes have poor seasons, mediocre players sometimes have great seasons. This is certainly the case with Pat Litasch, Angel Berroa, Bob Hamelin and others who have had great rookie performances that were well above their talent level and evidently unsustainable.

Final Thoughts and Relevance for Current Young Superstars

Admittedly, my first final thought is “I wonder if anyone sells Pat Litasch jerseys anymore. I should check that out.” Updates on this pursuit will be announced via my Twitter account.

The big question for this study is, “What does this tell us about current and future Rookie of the Year award winners?” Well, unfortunately, not a whole lot. Since the R-squared values were low, WAR totals from each winner’s rookie season have proven to not be a great predictive statistic for player performance in any of the three categories we examined within this data set.

Nevertheless, let’s take a brief look at the recent Rookie of the Year Award winners.

Mike Trout is simply off the charts. His 2012 WAR total of 10.9 was over 3 WAR higher than the next highest entry on this list (Ichiro Suzuki in 2001), meaning that this data is not reliable for a player of Trout’s caliber. He is an outlier.

Trout did experience a sophomore slump as his WAR dropped down to a mere 9.2, which is still tied with Cy Young (1894), Shoeless Joe Jackson (1911), Ty Cobb (1912), Home Run Baker (1912), Jimmy Foxx (1933), Ted Williams (1949), Stan Musial (1949), Barry Bonds (2003), and others for the 270th best season of all time. Trout’s debut season is tied for 95th best all time, which is even more impressive considering many of the seasons in front of him are pitchers from the 1800’s.

Yeah, I admit - I just wanted to talk about Mike Trout's greatness.

As for the rest of the modern crew, the data from this study does little to confirm or reject their chance of enjoying an elite career. The high WAR totals of Harper and Fernandez during their Rookie of the Year award winning seasons certainly boost their likelihood for an elite career, but this still does not mean that either player is destined for greatness.

Myers had the lowest WAR total of the group, but 2.0 WAR is reasonable, especially considering he did not play a full season due to concerns with service time.

Unfortunately, this study doesn't end with a crystal ball prediction about the future of each of these players. Rookie WAR is only one of many ways to forecast success for a player’s second season, second and third season, or career, and at most this can account for approximately one-third of the variation in the data.

Put in the form of an answer to the title question, the answer is no, elite rookie seasons do not necessarily lead to elite careers.