Now that we’re at the tail end of the offseason, it’s time to reflect: How closely did our free agent estimates correlate to the market? Let’s find out.

 

The numbers

By our count, there have been 126 free agent signings so far. Sure, there are still a few stragglers out there (David Robertson, Alex Verdugo, et al), but we’re at the point where we’re close enough to an endpoint to start self-evaluating our model.

The tricky question is, what’s a good metric?

Note that our methodology is more flexible than those of other baseball sites – with our free agent calculator (available to all GM subscribers), you can toggle between years to see how the AAV changes (typically, the longer the term, the lower the AAV, to account for increased risk). So we can’t really use total dollars, as that is variable for us. (Other outlets make predictions with a specific amount of years and totals, and therefore only one AAV, so it’s not quite apples-to-apples.)

That suggests we should focus on AAV, and within that, there are two options: the delta in percentage terms, and the delta in absolute value terms. Both have merits – AAV percentage tends to be more reliable on the bigger contracts, while AAV absolute value does a good job across all contract sizes, as this recent article in Fangraphs suggests. So let’s do both.

 

BTV model vs. the market, top 42 deals, calculated by AAV delta %

This measures the 42 biggest names signed in the offseason (using the list of top free agents written by Tim Britton of The Athletic as a benchmark). We’ve divided our results into six tiers, with the best being estimates that came very close to the market result (below 10% delta of the average AAV), followed by those that were between 10% and 19% delta, and so on down to the ones that were the farthest away from the market result (any above a 60% delta). The results are below:

 

Roughly 31% fell into the top tier, another 19% into the second tier, and another 21% in the third tier. In other words, 30 out of the 42 major contracts signed (71%) were below a 30% AAV delta. Is that good? It’s not bad – and we’ll be the first to tell you we can always be better.

Expanding that out a bit to the next tier, we find another 4 deals, or 9.5%, fell into the 30-39% miss category, which seems to be where the numbers drop off. If we draw the line there, it means 34 out of 42 of the major contracts, or 81%, were off by less than 40% on an AAV basis. That feels a bit loose, though, and it suggests we can do a little better at the bottom. 

 

Here are the deals we were very, very close on:

 

Notes: If there are player or team options in the deal, we calculate the AAV based on the probability of them being exercised by one side or the other, depending on if there is surplus value on either side; if there are deferrals, we use the present-value number supplied by the MLBPA.

 

So those are the biggest wins, if you will. And that’s a pretty good-looking list, especially considering the size of some of those deals. 

 

Others we were reasonably close on (between 10% and 19% AAV delta):

A few outliers of note: Gleyber Torres’ deal was undervalued by 81%, based on our modeling. Two very late starting-pitcher signings, Andrew Heaney and Jose Quintana, were off by 249% and 154%, respectively, as at that point their markets had fallen apart and they just took whatever deals they could.

 

BTV model vs. the market, all 126 deals, calculated by AAV absolute-value delta

As Ben Clemens noted in his Fangraphs’ piece, “this category is the real thing we’re all trying to get to.” So let’s get to it:

Out of all 126 deals, the absolute-value delta in our model was 2.74. This compares favorably to the 3.25 miss of the Fangraphs’ crowdsourcing exercise. One caveat: As the value of the contracts go down, so do the values of the misses, and because there are 84 of them in the lower tier, their sheer volume pulls down the overall average delta to a smaller number.

Here are the deals we were very close on in the lower-level tier based on this metric:

So there you have it. Overall, not bad, but could always be better. One last thought…

 

Timing matters

Teams have a tendency to overpay a little to get the players they want early in the offseason. Demand for the premium players in early December exceeds supply, so prices are high. As those players come off the board, the numbers start to normalize, as supply evens out with demand. 

Then, late in the offseason, as the pickings become slimmer, there are cheaper deals to be had, as supply exceeds demand, which results in some surprising underpays, such as those of Heaney and Quintana mentioned above. At this point, teams are either tapped out on their budgets, and/or become reluctant to pay extra for a marginal player when their in-house options are cheaper (and usually come with more roster flexibility). Savvy front offices know this, which is why we see some recent bargain shopping.

For the purposes of our model (and for everyone, really), the lesson is to be careful not to overreact and adjust too heavily to the hotness of the hot stove early on, as it inevitably cools off as time goes on. 

On the whole, this free agent market was pretty normal – one big contract, which was predictable; a few surprises, as always happens; and a whole bunch of deals that fell within a reasonable range.