Presenting Further Research on When Free Agents Ink Their Deals

Earlier this week, I published my findings about the relationship between when free agents sign and the size of their contracts. As a quick refresher, in recent years, the last 20% or so of free agents to sign have been settling for contracts meaningfully lower than pre-offseason expectations. But that finding raises more questions, some of which I hope to answer today.

First, there’s an obvious question: Did the free agents who got those late, discounted deals perform worse than expected during the following season? In other words, did their low-dollar-value deals foreshadow lower-than-projected production? To examine this, I took the upcoming season’s projections for the players ranked on my Top 50 Free Agents list in each of the past three years, 150 players in all, to come up with a projected WAR for each segment of players. I then compared it to how they actually did in the ensuing year. There is indeed a drop-off for those who signed late:

First things first: Every group underperformed its projections. That comes down to playing time. Our projections use Depth Charts playing time, which approximates the most likely distribution of playing time across a given roster without accounting for the likelihood of injuries. Just as an example, non-catcher batters were projected for an average of more than 600 plate appearances in this dataset, and they came in closer to the mid-500s in practice. So don’t pay too much attention to the absolute numbers; the relative differences are what to look at here.

The last 10 free agents to sign saw huge shortfalls in production relative to expectations. One reason: They played less. The average hitter in this group of 150 free agents batted 70 times less than projected. Hitters signed among the last 10 free agents in their class batted 100 times less than projected. Likewise, the average pitcher in the group came up 25 innings shy of projections, but pitchers among the last 10 players signed came up 40 innings short.

But that doesn’t explain the whole drop-off. Both hitters and pitchers in the last group of signees just played worse, too. Their WAR/600 PA (or WAR/200 IP for pitchers) missed projections by more than any other group in the cohort. In other words, free agents who sign last often perform worst relative to expectations. There’s no clear causation here. Maybe signing late means abbreviated ramp-up time and a worse early season. Maybe players who sign late are disproportionately those identified by teams as having red flags not caught by public projection systems. Either way, the gap in production is inarguable.

That brought up one more question for me: Does the market balance out? In other words, we know that players who sign last generally do so for less than expected. We also know that they perform worse than expected. Do these two effects balance out to make some kind of idealized economy for free agents, where try as you might, teams get the same bang for their buck and all that changes is the number of bucks?

To answer this, I had to switch from the lens of total guaranteed dollars to average annual value. After all, if a hitter signs a five-year, $100 million contract and produces five wins above replacement in year one, the team didn’t pay $100 million for five wins – they paid $100 million for five wins plus four more seasons from that player. Putting everything on a per-year basis does a good job of tying the deals together on a reasonable axis.

I thus took the average annual value of every contract signed by my population of free agents, compared it to the WAR they produced in the first year of their deal, and got a striking result:

“Waiting out the market” doesn’t appear to pay dividends for a team. The free agents who sign last might sign smaller-than-expected deals, but their delivery of worse-than-expected performance has overwhelmed the savings in recent years. If you’re a team that wants to maximize its free agency dollars at all costs, waiting until the end of the market to offer deals to the players no one else has yet signed doesn’t appear to be the solution.

It’s worth noting that the second group isn’t clearly the best bargain despite having the lowest dollar-per-WAR outlay in year one. As I discussed in my previous article, that bucket has more superstars getting huge deals than all the other groups, as you can see from the average years column. With those deals, getting a first-year bargain in terms of wins-per-salary is nearly guaranteed. The riskier part of those contracts is the later years. For that group specifically, this comparison isn’t apples-to-apples.

But compare free agents signed in the first, third, and fourth groups with the stragglers, and you’ll see a huge difference. The AAVs? They’re broadly similar. The on-field value produced by those groups? Not so much. Altogether, players in the first, third, and fourth groups were projected for 1.8 WAR and produced 1.5 WAR. Players in the last group were projected for 1.8 WAR and produced 0.9 WAR. The differences in average annual value ($1.7 million) and guaranteed years (0.5 extra guaranteed years on average) were minimal. Mostly, those players who signed last just offered less performance per dollar.

In terms of broad conclusions, I have two main takeaways. First, the last group of players to sign in free agency really does appear to be different from the rest of the class across the board. Those players rack up less playing time and perform worse in that time. Even though they’ve broadly signed contracts that don’t meet their pre-free agency expectations, they haven’t played as well as their counterparts who signed earlier in the winter.

Second, teams who are attempting to wring the most out of a limited free agency budget probably shouldn’t wait out the market and attempt to sign whoever is left at the end, even if the contract they sign represents a team discount compared to pre-offseason expectations. Those players might command lower salaries, but they also perform worse. If efficiency is the goal and marquee free agents are off the table, the data suggest that almost any time is better than the very end of the offseason to add new players.

Finally, some caveats, because oh boy are there caveats. These data only encompass three years. It’s statistically significant over those years, but there’s no guarantee that behavior that holds one year will continue to do so in the future, particularly given that the bargaining between teams and players for free agency contracts is a dynamic process. If the contours of the market change in the future, the results easily could too.

Not every win above replacement is created equal, and WAR is a blunt instrument for our purposes. I’d prefer a player who’s a stone figurine for two months and then turns into an All-Star for the rest of the year rather than a player who’s just blah all year long, and my test absolutely isn’t measuring the distribution of WAR. It’s entirely possible that the miss relative to expectations comes down to less time to get into in-season shape; teams that plan for that might find their realized-value calculations quite different from what I’ve outlined here.

These are only broad conclusions. They might be true in an aggregate sense, but they’re hardly destiny. Matt Chapman signed at the very tail end of last year’s offseason, and his performance in 2024 represented the third-greatest outperformance relative to projections in the entire sample. Carlos Rodón signed 30th in his class and then was below replacement level the next year, the largest miss in the set. Broad aggregates are exactly what they say on the carton: broad and aggregated. Individual assessment of players is far more important in any given case than a boilerplate rule about when deals are signed.

With all of that said, I still find this conclusion fascinating. Yes, the last free agents to sign generally do so for less than we expected coming into the winter. But that doesn’t appear to be some economic force suppressing their compensation so much as a selection mechanism. The last free agents to sign are sometimes last for a reason, in other words. It’s something to think about the next time your team is looking to add the missing piece to its roster.