Assessing free agent value through comparable players: offensive linemen

With only two teams remaining to compete in the Super Bowl, the offseason is in full swing for the rest of the NFL and football pundits. We at PFF are also getting into the action with the launch of our 2020 Draft Guide and new Free Agency Rankings tool, which allows you to sort by position and team needs and includes contract projections from OverTheCap as well as historical grades and wins above replacement numbers for the past three seasons.

I am also doing a series of articles that will go through our top-50 ranked free agents (excluding QBs), using their past stats to match them with historical players. Once we have a sample of similar players, we can see how those similar players performed in subsequent seasons to get an idea of the current player’s range of outcomes.

This system is similar to FiveThirtyEight’s CARMELO player projections, only it is applied to football and uses PFF’s WAR metric as the proxy for player value.

In earlier research, I looked at historical WAR and salary cap figures to derive the approximate cost of a win. Adjustments to our WAR calculation and the ever-rising NFL salary cap has moved the current estimate for the price of a win for spending on unrestricted free agents at 34.2% of the cap, or $68.4 million based on an assumed 2020 cap of $200 million.

Buying wins in free agency is an expensive proposition, but the value of each incremental win for a playoff-caliber team cannot be understated. This year, teams are going to have tough decisions on how to spend that money. The first installment on wide receivers and the second on running backs and tight ends have already been published. In this article, we’re forecasting free agents at the offensive linemen. Unlike other posts, I’m not projecting the entirety on those who fall into the top-50, as No. 47 Andrew Whitworth and No. 49 Jason Peters are too old to produce a meaningful set of comparable players.

Metholodogy

For each of the offensive lineman below, the closest comparable players were found by measuring the similarity between them based on several features — age (within one year), weight, WAR, previous-year WAR, PFF pass block grade, PFF run block grade, and WAR per snap.

The most similar players across these metrics are grouped together to calculate the expected WAR numbers for 2020 and 2021, with the weight of each comp reflecting the level of similarity. I didn’t use the same number of comps for all players, as it is more difficult to find a wide range of comparable seasons for older players.

The first plot displays past and future performance for the free agent based on the comp, the next shows the relevant WAR numbers for each of the comps and the final table gives the dollar-based value figures for 2018-2021, with assumed per-team salary caps of $200 million in 2020 and $215 million in 2021. I decided to restrict the forecasts to the next two years in order to maximize the number of comparable players we have in the dataset, reduce error for longer forecasts and reflect the fact that all but the top-tier free agent contracts in the NFL are glorified two-year deals that give teams the option to continue.

No. 17: Anthony Castonzo

The first thing we need to know from Castonzo is whether he’s going to retire this offseason. If he decides to give it a go, he’ll be a little over 32 years old to start next season. Castonzo’s expected WAR projections for the next two years are fairly modest, and even the ceiling projection isn’t much better, on average, than what he produced the last two seasons. Across the offensive line our WAR values are probably lower than most expect, and certainly for players who are beyond their prime years.

Castonzo comps range from Joe Staley, Duane Brown, and Donald Penn on the high side, to Michael Roos and Tyson Clabo on the downside.

OverTheCap sees Castonzo seeking a three-year contract worth around $13 million per season. That looks rich versus the expected WAR based on comparable players, though a slight discount versus the ceiling outcomes.