Stars or Depth? What Is the Best Way To Build an MLB Roster?

Building an MLB roster is anything but simple, to say the least.

It would be very convenient if it was as easy as playing MLB: The Show, but as we are well aware of, there are many complexities to roster construction. Not only do organizations need to have high-end talent, but they also need to have 26 quality big-leaguers as well other players in the pipeline when adversity hits.

In a perfect world, teams would be able to have tons of star talent as well as intriguing depth. However, we do not live in a perfect world, and for that reason, teams need to adopt a specific strategy when it comes to building the best roster possible in the most efficient way imaginable.

Teams have generally two courses: will they prioritize star talent, or will they look to have as deep a team as possible? The first option is typically known as the “stars and scrubs” approach, and it is one that you see often see in basketball. Meanwhile, the latter approach is one that you’ll see with sports with deeper rosters, primarily football. Overall, both methods are used frequently by teams, but it is unclear which one is the more efficient when it comes to roster building.

What good is there to posing a problem if we aren’t going to find the answer for it? We need to dig deep into these two approaches! Should teams prioritize star talent even if it means their depth is lacking? Or is quantity more valuable than quality? Let us try to discover the answer to this critical question!

Lineup Construction

As I’ve written about in the past, offense clearly wins championships in baseball. Teams need to do whatever they can to maximize the amount of runs they can score. But does that happen more by having elite players at the top of the lineup or a deep lineup? The results are mixed:

By the looks of it, it is equally as effective to score runs by having star hitters or having depth. Yet there appears to be limits to how much the star hitter truly matters. Just based off of the first scatterplot, having a truly elite hitter hasn’t guaranteed offensive success for teams, while the results are less volatile for clubs that prioritized quantity of quality. Could this mean that the “stars and scrubs” approach has a higher range of outcomes? I think that is likely the case, but remember, when you have more quantity, there is a greater chance that one breaks out. In the end, is it easier for “scrub” hitter to play up to average or for the average hitter to become a star? By the looks of it, there isn’t a clear answer.

In the end, teams need to hedge in the middle. One star hitter may not make the difference, but if you can have multiple high-end hitters, perhaps you don’t need to have a lineup that goes nine deep. Let us call this a “depth of stars.” The Boston Red Sox, for instance, only had five above-average hitters on their 2018 World Series lineup, but four of them posted an offensive runs above average (Off) over 20. Although the end of their lineup was lacking, they had enough depth of quality hitters at the front to make up for it. We’re always looking for binary answers, yet it appears to be that when it comes to lineup construction, the main contributions come from the first four-to-five hitters. If they are of enough quality, then the back end of the lineup doesn’t have to be sound.

Rotation Construction

When it comes to a pitching staff, we often hear about the importance of the “ace.” The thought is that you can hide warts with the rest of your rotation, and there potentially may be some validity to that:

Once again, we see that there is a relationship with allowing runs between the star player and the amount of depth a team has. Even more than with scoring runs, however, there appears to be much more volatility between the relationship of allowing runs and the ace pitcher’s wins above replacement (WAR). This, in my opinion, means that if given the choice between having an ace or multiple quality pitchers, I would choose the latter option. Possessing a high quantity of two-win pitchers, at least based on the scatterplot, nearly guarantees a team’s ability to limit the amount of runs scored, and the same cannot be said about having one true ace.

That line of thinking particularly rings true when considering the lack of relationship between having multiple four-win pitchers and being adept at run prevention. The Nationals may opt for star pitching rather than depth, but they also had three pitchers worth more than four wins in 2019, which is equivalent to the amount of 2+ WAR pitchers a team usually has. Meanwhile, teams like the Mets, Rangers, and Phillies have had issues with run prevention in the past despite having high-end talent at the front of their rotation. In other words, quantity over quality appears to be the way to go when it comes to building a rotation.

Bullpen Construction

There is some value to having a rotation ace, but it is definitely limited. Does the high-end reliever, typically labelled as the “closer,” fare similarly? Not exactly:

I have written before about the lack of value of a high-end reliever in today’s game, and these results back that proposal up. Even when dismissing the volatility of individual relievers, having one fantastic arm doesn’t matter much if the rest of the bullpen can’t do their part. This is yet another data point that suggests that teams should not pay top dollar for a reliever.

What makes the Rays so strong isn’t one reliever, but the amount of quality options they have, and the same is true for other clubs with strong bullpens, such as the Padres, Yankees, Brewers, and Dodgers. The White Sox may be happy with the results of their bullpen, but that may not have much to do with them signing Liam Hendriks and rather their abundance of relievers who could easily be worth over 1 WAR last year. On the other hand, the Dodgers do not need to feel pressured to sign a high-end reliever, as their depth more than makes up for not having a prototypical closing pitcher.

With the most volatile aspect of a team’s roster, it isn’t surprising that quantity easily trumps quality. Adding as many decent relievers possible, whether via waiver claims, low-level signings, or minor trades, should be a focus for every team when constructing their bullpen. For what it’s worth, the Phillies, who had the league’s worst bullpen 2020, have adopted this approach, and it should work out much better for them than simply signing Hendriks to be their top option.

Roster Construction

We have gone over how stars affect scoring and preventing runs, but how much do they impact the game itself?

There are multiple ways to look at this question. For starters, we can classify “stars” as both high-end position players and pitchers. At the same time, a high-end hitter may have a different affect than an ace pitcher, especially since position players generally have a slightly greater impact on winning games, based on previous studies.

Let us start by grouping hitters and pitchers together. All of the sudden, we start to see the clear winner of the “stars versus depth” debate. Remember, we are using expected wins rather than actual wins, as expected wins give us a better idea of the team’s true talent level.

First off, I would like to apologize to Mike Trout for not being part of a winning team! Seriously, it is sad how clear it is to see which three dots are him being the star player of an underperforming Angels club. Hopefully, with a new general manager in place, they can make strides towards being a playoff team in 2021.

In all seriousness, depth appears to be a greater part of a winning team than simply having star players. This all goes back to the foundations of a successful organization. Think of this like a house. Until you have the foundation in place, there is no point of having shiny external factors; the house will just collapse. This also applies to team-building. It is great to have Mike Trout, but if I am not surrounding him with quality talent, then his peak seasons are just going to be wasted. In fact, that is exactly what has happened in Anaheim. I do believe that stars can help put teams over the top and can be the finishing piece, but when it comes to the one feature that practically all winning teams possess, it is having a certain level of organizational depth.

Now for the second part of our question: does having a star position player matter more or less than having a star pitcher. What about above-average hitters versus pitchers? Based on the coefficient of determination (r^2) for each of the following relationships, the results give a somewhat convincing answer:

  • Star Hitter WAR to xWins= .399
  • #4+ WAR Hitters to xWins= .413
  • #2+ WAR Hitters to xWins= .365
  • Ace WAR to xWins= .315
  • #4+ WAR Pitchers to xWins= .245
  • #2+ WAR Pitchers to xWins= .350
  • Star Reliever to xWins= .051
  • #1 WAR Relievers to xWins= .129

Yes, star hitters have a slightly greater impact than star pitchers, which would line up with the idea that position players get paid more $/WAR than pitchers. There are plenty of approaches when it comes to lineup construction, but that is where stars make a difference, while depth has to be prioritized for pitching. Meanwhile, there seems to once again be little benefit between having a star reliever compared to a star player at a different position.


Is there a convincing answer to the “stars or depth” question?

In my opinion, both are valid ways to build a team. Top-heavy organizations such as the Red Sox and Nationals have adopted a “stars and scrubs” approach in the past, while several successful organizations prioritize quantity over quality.

However, I do believe that depth is more significant when it comes to team success than having star talent. Having one star player guarantees little, as seen with Trout, while having an abundance of quality players generally means that you are going to win a lot of games.

Organizational depth is the foundation of a team, and it is the main trait that clubs like the Dodgers, Yankees, Rays, Twins, and Cleveland have to keep them winning sustainably. Once you have that depth, however, you can look to take your team over the top with star talent, as that is the only way to upgrade the roster at that point. When given the choice, I would lean towards adding the star position player, though as long as you aren’t investing heavily in a high-end reliever, you could go either way.

A lineup appears to be as good as its fourth- or fifth-best hitter, a rotation is as good as its third- or fourth-best pitcher, and a bullpen is as good as its overall depth. These are the roster-building principles that I would go by. A “stars and scrubs” approach may let you have more flashy talent, but it also doesn’t necessarily mean you’ll win many games. Consequently, teams need to be responsible and make sure they build their organization with the proper foundation of players before “star chasing.” This is why teams need to go through rebuilds, and it is precisely why the Padres are on the verge of joining teams who can win consistently, while the Phillies and Angels have failed to make the postseason in recent years.

The best decision often isn’t the most exciting one, and that is the case with MLB roster construction. If general managers could build their roster like they can in video games, they would, but given the unpredictability of baseball and a limited amount of resources that each team has, they cannot easily have stars and depth (unless you’re the Dodgers or Yankees). You can never have enough quality players, and even if it comes at the expense of having a superstar who can help sell season tickets, decision-makers need to stick by that mantra.

More of my work can be found at

newest oldest most voted
Matan Kronfeld
Matan Kronfeld

There is a time and place for single variable linear regression. This is not it. A lot of nuance is being left out in this article. For example, correlating top reliever WAR to runs allowed per game and saying since the R2 is low therefore a top reliever is unimportant is a flawed analysis. The value of a top RP is that they perform well in high leverage situations, which isnt reflected in R/G. Also, there may be some correlation/causation issues here. Perhaps better teams tend to develop more decent players and stars are more random. If so, deciding to… Read more »


How does using the more appropriate nonlinear fits adjust the variance and therefore the fit contribution to the model.