Valuing Major-Leaguers



How do we calculate these values?

On the surface, calculating trade value is relatively simple:

Field value – salary = surplus

Let’s dive into that a bit, because there’s actually a lot there.

Calculating “field value”

What do we mean by “field value”? (By the way, that’s our term. We don’t know if anyone else uses it.) It’s an estimate of how much that player is worth on the field. The easiest way to think about it is: what would this player be worth in free agency?

And how would we know that? Well, if you’re familiar with the concept of replacement value, you know that, in theory, a team could replace all of its starters with random fill-ins from AAA, who are being paid the league minimum salary, and still win about 50 of its games, because of the randomness of baseball. No team actually does that, but the fact that they could represents a floor, and a starting point by which to measure a player’s value. In other words, that’s point zero.

That’s also where WAR comes in. Wins Above Replacement is a bit of an uber stat that attempts to measure the whole of a player’s contribution, in wins that would be generated above that theoretical minimum of about 50. So if our team of random AAA nobodies subs in a 2-WAR player, they would be predicted to win 52 games. Sub in 10 2-WAR players and you’ve got a 70-win team, and so on.

So then, how do you translate that into field value?

It’s complicated. There are many, many parts.


First, there’s been a lot of research on how much a team pays for 1 WAR. As of this writing (summer 2019), we think there’s a baseline in the low 9s – that is, a little above $9M. But we also noticed that’s not consistent. A DH, for example, doesn’t get paid that rate because he doesn’t play defense. Starting pitchers, we’ve noticed, get a bit more (before adjustments).


In our modeling, we’ve found that there isn’t one true projection system available that accurately matches what the teams are using. So we played around with our own combinations of statistics, using our own formulas and algorithms, until we found consistent correlations to real-life transactions. We consider this part of our story proprietary, so we’ll leave it at that.


Inflation also plays a role, and this matters when forecasting future value. A few years ago, there were estimates that dollar-per-WAR was inflating at a rate of 8% per year. But we noticed a significant drop-off, particularly in the offseason of 2017/2018 – the year of the great free-agent freeze. That year, in aggregate, free agent salaries stayed flat instead of inflating, likely due to the market adjustment of teams fully embracing data-driven models.

In the offseason of 2018/2019, inflation grew a little bit from there – we estimate 3%. We also take into account the rate that MLB uses to apply to qualifying offers, which was also 3% after 2018. This number also squares well with the general rate of inflation in the US economy. So we’re using 3% until we see any significant change.

Years of Control

When you trade for a player, you’re not getting that player for an open-ended timeframe. You’re getting him for a finite number of years. At the beginning of a player’s career, teams have them for six years — the first three at league minimum salary (which is $555,000) or slightly above, and the next three at arbitration cost, which typically increases each year and depends on their performance.

Importantly, none of these first six years are guaranteed to the player. Each year, the team has an option to tender them a contract or not. So in our modeling, we factor in not only the total years of control, but also the value-positive years, because this gives us a sense of whether they’re likely to be tendered down the road, based on projections. For example, if a young player is projected to produce above league-minimum salary for the first three years, but under their projected arb salary after that, we assume the team would non-tender them at that point; therefore we value them only for those first three years of positive value.

For more established veterans, the amount of control is dependent on the terms of their contract, which vary from player to player. In most cases, they are guaranteed a certain salary over a certain timeframe, which we use to calculate value. In some cases, teams have an option (or more) to extend them beyond those guaranteed years. In our modeling, we treat those non-guaranteed years the same way we do for younger players above: if they project as value-positive, we include that value in our results; if not, we assume the team would decline the option.

Injury risk

This is a big one. In case you haven’t noticed, baseball players get injured. A lot. Some are minor, some major, some catastrophic. We noticed that many popular projection systems do not properly account for this (they do account for it a little, baked into playing time estimates, but not nearly enough, in our view).

Mind you, we’re not doctors or insurance actuaries. But we know this is a thing that baseball front offices care about a great deal, so much so that they’ve baked in a significant margin of safety in their valuations, which appears in the on-paper gap you might see between a projection system’s WAR estimate and the contract the player received.

So we’ve tried to approximate that amount, which tends to vary both by position and player. In this area, the riskiest players are starting pitchers. And we also know that as players get older, their injury risk increases – and it’s not a straight line. The risk adjustment just gets bigger and bigger each year.

Roster risk

When the Brewers traded Domingo Santana to the Mariners in December 2018, Brewers GM David Stearns acknowledged that one of the reasons for the trade was that Santana was out of options:

“At this stage, given where we are from a roster perspective, and given he was out of options, we felt this was a move that made sense,” Stearns told the Milwaukee Journal-Sentinel.

That means Santana could no longer be sent to the minors if he wasn’t cutting it at the MLB level.

That lack of flexibility affects a player’s value, since he’s taking up a roster spot that could potentially be given to someone else who’s better. It means the trading team has less leverage.

By the same token, teams value the flexibility of being able to move players up and down from MLB to AAA, because of the added depth it gives them, as Stearns also acknowledged at the time to MJS: “As you guys know, we value flexibility and versatility.”

So players who are out of options carry what we call roster risk, which is a negative adjustment to their field value. We apply a discount for players with that status. (Note that this is only true of players who are marginal – typically somewhere just above replacement value; stars or above-average regulars don’t have this risk.)


Most salaries are publicly available on multiple websites. Simple, right? Not always. Quite often, we have to estimate it beyond the current year. A veteran player may have a contract that includes a team option. In that case, we have to estimate whether the team is likely to pick up that option or not. If the projection is negative, we assume the team would not, and therefore don’t include it in our valuation (although we do include it in our Years of Control stat).

For players with less than six years of service time, we estimate their arbitration costs based on the 25/40/60 model. That is, for their first year of arbitration, they are likely to earn about 25% of their projected market value, then 40% of that in their second arb year, and 60% in their third. (Note: The Point of Pittsburgh study linked here is the most recent and, we believe, the most accurate of the estimate models, more so than previous estimates of 20/40/60, or, confusingly, 40/60/80.) As noted above, we don’t calculate negative-value years in our estimates.

Finally, we assume that player salaries do not decline in arbitration years. The arbitration system is based on giving raises, not pay cuts. We also found that there seems to be an unspoken minimum salary for first-year arb players of around $800K. So if they’re projected to produce less than that in their first arb year, we assume for our purposes they would be non-tendered. If they’re projected to produce less than 40% of their market value in their second year, and/or 60% in their third arb year, we also make the same assumption. (We’ve noticed a lot more players are being non-tendered than in previous years, likely due to teams getting smarter about valuation projections.)

Surplus value

So, after factoring in all of the above, we return to our simple formula:

Adjusted field value – salary = surplus trade value

That surplus trade value is the default estimate, and it’s commonly referred to simply as “trade value” in the media and among baseball experts. But there’s one more adjustment we need to make at this point.

Market adjustment

Because there are only 30 teams, when it comes to transactions, major league baseball is considered a closed market. It’s not as efficient or as liquid as other types of open markets. So there’s a wider variance of outcomes – especially in the trade market, where it’s sometimes difficult to find a trade partner for a certain player.

It also means that imbalances of supply and demand tend to be magnified. In the offseason of 2018/19, there was an oversupply of veteran 2Bs on the free agent market. Jed Lowrie, Brian Dozier, Ian Kinsler et al were all proven quality players, but they weren’t that much of an upgrade over cheaper (often in-house) options, so their free agent price came down. In other words, their price-per-WAR was lower than the baseline. This affected the trade market for 2Bs as well – we saw guys like Aledmys Diaz and Jurickson Profar go for less than that baseline (even though they’re versatile enough to play multiple positions).

Conversely, there was an undersupply of affordable quality catchers available, which raised the trade acquisition cost for those few who were in demand – such as J.T. Realmuto and Yan Gomes.

Further, you’ll notice in our trade simulator and team rosters that we indicate “availability,” which is our estimate of the likelihood that a given player would be moved by his team. To be sure, there is some subjectivity here, but it is informed by our continuous reading of the markets and reporting done by various news outlets, including MLBTradeRumors.

So we keep an eye on those trends and adjust accordingly. That is why, in our estimates, we provide a low, median and high range. These supply/demand factors strongly influence those numbers, and are the last adjustments we make. (Or are they?)

Except… in the summer

The summer trading deadline is its own animal. In the winter, the factors we mentioned above are relatively stable, because more teams are open to making roster improvements. There’s less of a distinction between buyers and sellers.

But in June and July, contending teams start to separate from non-contending teams, resulting in clearer cases of buyers and sellers. The buyers start looking for short-term improvements to carry them into a potential playoff run, while sellers look to trade current quality players whose team control period is running out in exchange for prospects to help their team in the future.

From a valuation standpoint, we account for this phenomenon in several ways:

We first adjust the valuations of all MLB players based on their production so far in the current year. We do one update in early June, based on the players’ numbers through the end of May, which gives us 1/3 of a season. We then blend that with their full-season projection to get a new, updated projection. We then do the same thing again in early July, based on the players’ numbers through the end of June, which gives us 1/2 of a season, which again is blended with their original projection.

We also identify the veteran players most likely to be traded to contenders. We adjust the valuations of the most impactful of those players by giving them an “October premium” – assuming they’re good enough to be traded to a contender to help them, they’re good enough to anticipate playing in October. After all, that’s what the acquiring team has in mind. It’s effectively a bonus month. So we assume that if they’re traded in early June, the acquiring team is trading for five months of service, not four; if they’re traded in early July, it’s for four months, not three.

We then prorate their salary amount down. So if they’re traded in early June, they’ll be paid for the remaining four months of the regular season; if in early July, for three months. Note that we don’t include a potential October run in the salary estimates, because it’s unpredictable; and, if they make the playoffs, they’ll get paid for that separately, more from the league than the team.

So the fact that we account for October in the playing time estimate, but not in the salary estimate, changes the equation, and often results in a valuation that is similar to the original full-year estimate. We believe that’s why teams that are potential sellers of star players prefer to wait until the summer deadline to gauge the market, because the odds are usually in their favor to do so – the valuation often doesn’t go down, and depending on performance, sometimes even goes up.

Finally, we apply one last look at the market at the summer deadline to see what types of players will be more in demand than others. It’s a given that pitchers – both starters and relievers – are in demand. That’s always the case. But position players are more variable, and the summer market for them depends more on the needs of the buyers than the wants of the sellers.

Thinking probabilistically

Overall, this is a probabilistic model. That means nothing here is absolute, but rather, a best guess based on the data available to us. We believe that’s the way MLB front offices view their transactional decisions as well. “You’re placing bets,” Yankees GM Brian Cashman told The Athletic. “And there are risks to every bet. It’s like investments. There is no such thing as a 100 percent guaranteed return on your investment.”

Questions? Contact us.