What the Crowd Tells Us About Free Agent Trends

Every year, FanGraphs asks our readers to provide contract predictions for the game’s top free agents and every year, our readers do an admirable job with their winter forecast. The predictions for this offseason’s most notable free agents can be found in our Top 50 Free Agents post; if you prefer a sortable, filterable table where you can easily see all the predictions, plus players’ actual contracts (when signed), we have that option as well. While their predictions are important and valuable to the site, I hope our readers will not take offense when I say that over the last few years, their contract prognostications have not been as good as they were in the past.

Going back to the winter prior to the 2014 season, here is what our readers predicted teams would spend, as well as the actual dollars spent, by year, with some figures coming from this 2018 post and Max Rieper’s prior research:

Free Agent Contract Crowdsourcing Results
Year Players Crowd ($/M) Contract ($/M) Difference % Difference
2014 43 1320.8 1366.9 $46.1 M 3.5%
2015 45 1396.0 1498.4 $102.37 M 7.3%
2016 52 2340.0 2215.0 -$125.0 M -5.3%
2017 42 1441.0 1147.0 -$294.0 M -20.4%
2018 50 1711.0 1279.0 -$432.0 M -25.3%
2019 63 2068.3 1707.4 -$360.9 M -17.5%
TOTAL (’14-’19) 295 10277.1 9212.6 -$1064.4 M -10.4%

Over six offseasons, the crowd’s predictions were roughly a billion dollars too high. That total is only about 10% off, which doesn’t seem so bad. Looking at the individual years above, we can see that the billion dollar difference is housed almost entirely in the last three winters. From 2014 through 2016, the crowd fluctuated a bit but with over five billion dollars in predicted salary, the crowd was off the actual mark by just $23 million, less than half a percent off the total amount. Over the last three seasons, major league payrolls have remained static. The lack of upward movement in spending has come almost entirely at the expense of free agents, who make up roughly two-thirds of total payroll. Based on the crowdsourced predictions, it’s fair to say that readers expected payrolls to rise. The lack of a decent increase in 2017 was a surprise, as was the fact that there was no upward correction in 2018. Last year, readers were closer than they had been the previous two offseasons, but still missed the mark by 17% compared to the actual contracts signed.

To take a closer look at the split, we’ll separate the seasons above into two groups, and break them down by predicted contract value. Here’s 2014-’16:

Free Agent Contract Crowdsourcing 2014-’16
Predicted Contract Players Crowd ($/M) Contract ($/M) Difference % Difference
$80 M+ 17 2185.5 2434.3 $248.8 M 11.4%
$40 M – $80 M 22 1141.0 1066.0 -$75.0 M -6.4%
$10 M – $40 M 79 1591.5 1387.6 -$204.0 M -12.8%
Under $10 M 22 139.0 192.5 $53.5 M 38.5%
TOTAL (’14-’16) 140 5057.0 5080.3 $23.3 M 0.5%

For the players with the highest predicted contracts, the crowd undershot by the totals by 11%. It came pretty close on the upper-to-mid-range deals and tended to overshoot the large middle tier of free agents. At the very bottom, players got more than expected, but the totals were so low as to not make a huge difference. Over the last three years, the crowd has accurately predicted the top contracts while getting considerably worse in the middle-tiers:

Free Agent Contract Crowdsourcing 2017-’19
Predicted Contract Players Crowd ($/M) Contract ($/M) Difference % Difference
$80 M+ 12 1594.0 1567.5 -$26.5 M -1.7%
$40 M – $80 M 24 1340.0 1075.7 -$264.4 M -19.7%
$10 M – $40 M 95 2131.0 1360.8 -$770.2 M -36.1%
Under $10 M 24 155.0 129.0 -$26.0 M -16.8%
TOTAL (’17-’19) 155 5220.0 4133.0 -$1087.1 M -20.8%

When it comes to the accuracy of the crowd’s wisdom, we can see a decently-sized gap on the predicted contracts from $40 million to $80 million, and a significant difference in the $10 million to $40 million tier. This group had the biggest gap from 2014-’16 as well; it includes 71% of the difference between the crowd and the actual contracts while only accounting for 61% of the players and 41% of the expected contracts. When we compare the differences from the last three years to the three offseasons prior to that, we can see the outsized role the $10 million to $40 million group plays:

Shift in Free Agent Crowdsourcing Accuracy
Predicted Contract Contract-Crowd (2014-2016) Contract-Crowd (2017-2019) Change % of TOTAL Change
$80 M+ $248.8 M -$26.5 M -$275.3 M 24.8%
$40 M – $80 M -$75.0 M -$264.4 M -$189.4 M 17.1%
$10 M – $40 M -$204.0 M -$770.2 M -$566.2 M 51.0%
Under $10 M $53.5 M -$26.0 M -$80.0 M 7.2%
TOTAL $23.3 M -$1087.1 M -$1110.1 M 100.0%

As one would expect, the change in that group occurred in part because so few of the players in it were able to land deals worth more than the predicted amounts, and because many players signed deals that fell far short of expectations. From 2014-’16, 10 players signed contracts worth at least $10 million more than was expected, for a total of $151 million, while 14 players undershot their predicted deals by at least that much, for a total of $282 million. Over the last three years, only three deals have resulted in the player receiving at least $10 million more than expected, for a total of $34 million, while the 40 players who received guarantees at least $10 million below expectations signed deals that amounted to $698 million. The deals landing far below the crowd’s predictions are responsible for the bulk of the downturn in free agency.

We can’t blame the crowd for these predictions; it’s difficult to anticipate the bottom of any individual player’s market falling out. Since 2017, more than half of the predictions for players in the $10 million to $40 million group have been within $10 million of the actual contract. When forecasting any one player, taking the one-in-seven shot that they end up with undershooting what might be typically predicted by $20 million or more isn’t the wisest bet. In the aggregate, we might know that a bunch of these players will do worse than what we might expect, but choosing precisely which ones will actually meet that fate is a bit more difficult given their roughly similar value in an increasingly difficult market.

Of course, the crowd here isn’t a static group, and it is certainly aware of how the free agent landscape has changed; these predictions aren’t and shouldn’t be uniform from year to year. But there’s been a significant overestimation of the size of their eventual contracts for the players starting around No. 15 through the end of the Top 50 Free Agents list. We don’t know who will be left holding the bag in February or March, but it is likely that some in that group will end up very disappointed in their contracts.

Pinpointing the exact cause of these changes to the free agent market isn’t the purpose of this post, but it is hard to argue that these players aren’t being hurt by two sorts of competition. The first comes in the form of competition from a large group of younger, cheaper, very talented players, who have benefited from improved scouting and player development and who are still playing on their first major league contracts. With no increase in the number of jobs, and these players’ salaries significantly lower than they would be were they themselves to hit the open market, many teams are content to fill their rosters with the young and the cheap, which brings us to the second kind of competition, or rather, its absence, namely the lack of competition for free agents’ services. Tony Wolfe just detailed how older players tend to be clustered on playoff teams, which will limit a player’s market in free agency if the only teams interested in their services are increasingly small number of clubs that view themselves as contending.

As we begin free agency, the crowd still provides a good starting point for how much players are likely to receive in free agency, even if its predictive power has waned some in the last few seasons. If the last few years of this exercise are any indication, expect the elite players to come pretty close to the predicted amount and the next tier of players to come up short but not outrageously so, while the most meaningful divergence will be for the huge middle tier of free agents to fall short of expectations.

Craig Edwards can be found on twitter @craigjedwards.

Newest Most Voted
Inline Feedbacks
View all comments
4 years ago

Love to ask people to put in the effort to do this then crap all over them

Famous Mortimer
4 years ago
Reply to  regfairfield

No idea why this was downvoted so much. It’s literally what happened in the article.

CC AFCmember
4 years ago

Cause that’s not what happened? Pointing out the fact that the crowd’s guesses differ from the results isn’t “crap[ping] all over” the crowd.

Also, maybe work on the use of literally, because it would be shockingly powerful work if Craig literally managed to take a crap on the entire crowd.

4 years ago
Reply to  CC AFC

Craig’s work has been impressive in the past, but boy has the bar been set up high this time. This feat would truly impress.

4 years ago
Reply to  regfairfield

I totally disagree. I like it when we see the results of what our predictions are. If we’re off, it helps to understand why, so we can come closer the next next time.

4 years ago
Reply to  regfairfield

It’s fair to say that the readership population has changed over time as well. Fangraphs isn’t niche like it used to be.