How I’ll Evaluate MLB Free Agency Predictions This Offseason

The annual guessing game around MLB free-agent contracts has become almost as compelling as the signings themselves. This piece digs into a fresh way to compare FanGraphs’ much-watched contract predictions with the actual deals players sign—especially when the contract lengths don’t match up.

By normalizing years and assigning a fair value to those extra or missing seasons, we can finally get a sense of how close those predictions really are. No more hand-waving away the tricky math.

Why Comparing Free-Agent Predictions Is So Hard

Every offseason, FanGraphs drops contract projections for the top 50 free agents, right alongside crowdsourced estimates from fans. Those crowd numbers, built from thousands of submissions, have quietly become one of the most reliable tools in the industry.

The issue? Actual contracts rarely line up with projections in both years and dollars. One player might be projected for eight years and end up signing for ten. Another gets a three-year prediction, then settles for a one-year “pillow contract.”

That makes it messy to compare total guarantees, or even the average annual value (AAV), in any way that feels fair.

The Limits of AAV and Total Dollars Alone

AAV gets tossed around as a shortcut, but it doesn’t tell the whole story when contract lengths differ. A nine-year deal at a lower AAV might still be a bigger commitment than a shorter deal at a higher AAV.

Total dollars, meanwhile, ignore how quickly teams pay that money and what it says about long-term risk.

To really measure prediction accuracy, you have to account for both money and time. That’s where a simple but clever normalization method comes in.

A Simple Formula to Normalize Contract Lengths

Here’s the workaround: when predicted and actual contracts differ in length, extend the shorter deal to match the longer one. Price the added years at a fixed fraction of the original annual salary.

This way, you get an “apples to apples” comparison over the same number of seasons. The key assumption? Extra years at the back of a long-term contract are less valuable than the early ones, but not worthless.

The method uses a two-thirds multiplier on the annual salary for any added seasons.

How the Two-Thirds Multiplier Works in Practice

Here’s the basic idea:

This doesn’t need aging curves, WAR forecasts, or complicated modeling. It’s a rule-of-thumb solution based on the idea that back-end years are discounted but still matter.

Is it perfect? Not really. But it trades a bit of precision for consistency and transparency, which is crucial when you’re trying to evaluate dozens of contracts at once.

Juan Soto: A Case Study in Long-Term Deals

Juan Soto’s mega-deal is a perfect test case. FanGraphs projected Soto to land a 12-year contract. Instead, he signed for 15 years, stretching deeper into his 30s than a lot of folks expected.

With the new method, the 12-year predicted contract extends to 15 years, valuing those three extra seasons at two-thirds of the original predicted AAV. That gives you a “normalized” 15-year prediction, which can stand toe-to-toe with Soto’s real 15-year deal.

Why This Matters for Evaluating Market Valuation

This approach does more than tidy up the math—it clarifies how closely the market and the predictors value a player over the long haul. For someone like Soto, where teams pay for both prime years and late decline years, the contract’s structure and length say as much as the total guaranteed money.

Teams often stretch deals to lower the official AAV for luxury tax reasons, or to spread risk. The normalization method helps peel back those layers and get at the underlying valuation in a more even-handed way.

Handling Pillow Contracts and Short-Term Bets

The method shines when a player signs a short “pillow contract” to reset his market. Suppose a player is projected for four years but signs a one-year deal instead. The one-year actual contract gets extended to four years, with the remaining three seasons valued at two-thirds of that one-year AAV.

That gives a reasonable equivalent of what the short-term deal represents in long-run value—without pretending a one-year flier is the same as a four-year commitment.

Balancing Simplicity and Realism

The two-thirds discount is a compromise. It admits that late years usually deliver less surplus value, but it keeps the formula simple and repeatable.

Most importantly, it skips specific performance projections, which can vary wildly across systems and introduce their own biases.

Measuring Prediction Accuracy: Average vs. Absolute Miss

Once contracts are normalized to the same length, you can measure how far off the predictions were in two main ways:

Both metrics matter. Average miss tells you if the model tends to be systematically too high or too low. Absolute miss shows how tight the predictions are overall, regardless of bias.

What It Reveals About the 2024–25 Free-Agent Class

Applied to the 2024–25 free-agency predictions, this normalization method sharpens the picture. It better captures the effect of teams adding extra years to reduce AAV or trimming years to minimize risk.

Contract length itself is a form of valuation—teams express confidence (or caution) through years as much as through dollars. By equalizing the time horizon, we finally get to evaluate predictions on the same playing field.

What Comes Next for Contract Evaluation

This tool keeps changing—it’s not set in stone. The two-thirds multiplier is just a starting point.

The author seems open to tweaking it with more feedback and fresh data from future offseasons. As a practical framework, though, it already gives us a clearer way to judge how well we (and the crowd) read the market.

Front offices, agents, and fans are all playing the numbers game these days. Having a simple, transparent method to compare predicted and actual contracts isn’t just for stat nerds—it’s a real window into how baseball values talent over time.

 
Here is the source article for this story: Here’s How I’m Planning on Evaluating Free Agency Predictions