Evaluating Defensive Prospects for the Expansion Draft

By @kevinminkus

FC Cincinnati will face many difficult decisions over the next three months, as they build their expansion team to be ready for the 2019 season. Their next set of choices takes place today, in the Expansion Draft. What the team decides there may not make or break their season, but they do have the opportunity to add important pieces for their inaugural year.

One strategy, the path I’ll discuss here, is for Cincinnati to grab cheap, young players. The hope is that, while they weren’t key contributors for their former teams, those players will continue to develop. A team with enough of these works in progress, and with a sufficient capacity to develop them, might reasonably hope for a few to pan out into full-time starters.

I’d argue that defenders, specifically, make the most sense for these types of picks, for a few reasons. If we actually look at the last three expansion drafts, teams have generally followed this pattern:

A primary reason is that attacking spots tend to go to TAM players and DPs, making it much more difficult for a young, cheap, unproven attacker to break into a lineup. A young defender, however, generally has less formidable competition for his spot (though this is changing every year), and so has a better chance to actually play - Josh Williams (who really is a perfectly fine player) started half the season for the league’s best defense, for example. Further, some part of defending is system-dependent. A team can get by fielding average defenders if the coaching staff is able to manage them effectively to mask their deficiencies.

The difficulty in using these slots on defenders, though, is that they mostly look pretty indistinguishable. Available defensive stats generally aren’t great, so what’s a team (or a blogger) to go on if they want to sift through the Wyatt Omsberg’s and Tomas Hilliard-Arce’s of the world? I’ll add, too, that distinguishing among young defenders isn’t a problem unique to expansion teams. Many teams will face similar decisions on these types of players in the next few months, whether it’s in the draft, the re-entry draft, or as they come up in trades.

One method might be to filter defenders on a specific set of measurable skills. Jonathan Campbell, for example, is surprisingly good at playing the ball out of the back. But what if a team, especially an expansion team, just wants a defender that can get get minutes right away? It turns that a simple logistic regression using a player’s minutes from the year prior does a decent job of determining whether he’ll be good enough to play at all the next year. Here, from that model, is the probability of a player 25 and under (which I’ll use as my loose definition of “young”) of getting any minutes in a season, given his minutes the year before. A handful of players available in this draft are included along that line:

So Nick Besler has about a 85% chance of being good enough to get minutes in 2019.

This says that young defenders that played more minutes last season are more likely to still be playing this season. The model suggests something that is really pretty obvious, but good to quantify: if Cincinnati wants a sure thing, they should select players that are already a sure thing or close to it. Hilliard-Arce wouldn’t be a great pick according to this.

Depending on Cincinnati’s goal though, for example if they’re building for 2020 and the future, they may prefer a higher risk player that has some chance of being a cornerstone of the defense. To evaluate this, I put together a slightly more complicated model to assess the likelihoods of possible outcomes for young defenders. The model is a hierarchical, negative binomial model that incorporates an age effect as well as team defensive quality, to predict how many minutes a player will be capable of playing two years from the present. These estimates can then be used to evaluate the potential ceilings, or at least shorter term ceilings, on players available in the draft. Here are a subset of centerbacks available in the expansion draft, with a probability distribution given over their potential minutes in 2020:

Jonathan Campbell and Nick Besler on average rate as the most certain picks here (though the model is pessimistic on all available youngish defenders - probably because of the partial pooling). Both spent a significant chunk of time as starters this past season. Tomas Hilliard-Arce and Hassan Ndam, however, rate as potentially higher value options, because of their age and inexperience. If FCC is willing to take on some risk, selecting one of those two options could pay off in the long-run.

Here’s the same graph, for a subset of fullbacks:

Tristan Blackmon here rates the highest on average, and also likely has decent upside - last season was his first in the league, and he saw action in 10 games as part of one of the West’s best defenses (at least on xGA). Given that he fulfills both characteristics, he’d likely be a smart pick-up for Cincinnati.