It’s MLS playoff time, and that means the possibility of penalties and shootouts. Rodrigo Schlegel led Orlando to victory after Pedro Gallese was sent off. Zack Steffen earned a move to Manchester City, in large part, due to leading Columbus to two penalty shootout victories. Portland named a bar in Providence Park after the famous double post shot in the marathon 2015 shootout against Sporting KC. Seattle won their first MLS Cup without registering a shot on goal by winning a shootout. Landon Donovan blasted his penalty over the bar to hand RSL the 2009 MLS Cup. Penalties and shootouts are remembered in ways that few other plays are.

It’s likely that at least one game in the 2021 playoffs is going to be decided by either a penalty or a shootout. So who should your team have take these penalties? Or just as importantly, who shouldn’t take them? To answer this we turn to Bayesian statistics. Specifically, I’m going to use empirical Bayes estimation which was previously used by Tuan Nguyen Doan to look at penalties in European soccer. In this case, empirical Bayes estimation uses historical knowledge about player penalty conversion rates to determine what any given player’s likely true conversion rate (known as prior expectation) and then update this based upon an individual player’s actual conversion rates (known as the posterior). There are other ways to do this, as Kevin Minkus, now Director of Analytics for the Fire, did using item response theory here at American Soccer Analysis in 2017.

In order to determine the prior expectation of penalty conversion, I turned to FBref using jaseziv’s spectacular worldfootballR package to quickly grab the career penalty conversion stats from 7,409 players. It turns out that 73% of players in the database have never taken a penalty in any competition, and less than 10% have taken at least five (note: shootouts and international matches are not included). I then determined the prior distribution from these 691 players with at least five career penalties and calculated the posterior mean and probability distributions of penalty conversion rates for the 35 players involved in the 2021 MLS playoffs.

The way to read these is to compare the shape and height of each player’s curve to the dashed white one that represents the prior expectation distribution, that is the likely ranges of penalty conversion for a player before you know anything about them. In the case of Alan Pulido and Nicolás Lodeiro, the model is pretty confident that they are pretty good at taking penalties. On the other end are DPs Nani and Maxi Moralez whose curves are shifted left from the prior distribution indicating that they may not be great penalty takers given their history. It’s hard to be really bad at penalties and still be tasked with taking them, so you are not likely to see very many players with very low estimated conversion rates.

A surprising name near the top is central midfielder Justin Portillo, he of 19 total minutes played this season in MLS for RSL. Portillo did not even make the bench on many match days. While he has never taken one in MLS, he converted 92% of his penalties across his time with Charleston and Real Monarchs in USL. If I’m Pablo Mastroeni, maybe I count him among my 20 available players and be ready to sub him in if their playoff game against Seattle is going to a shootout.

Of course there is another person that has some say about whether penalties are converted, the goalkeeper. I next grabbed data from 759 keepers in FBref, of which 73% had faced at least one penalty, and the 387 that faced at least five were used to determine the prior distribution. For MLS keepers I manually added penalties and shootouts from playoff games, MLS is Back, and (why not?) MLS All-Star games to what was available from FBref.

Of the 17 keepers in the 2021 MLS Playoffs (with a bonus Zach Steffen), Tim Melia runs away as the best in MLS, and is the best penalty stopper in the world using this method. And this doesn’t even doesn’t include his US Open Cup shootout heroics while in USL. Perhaps Gregg Berhalter should call him up to Qatar, assuming the US makes it, just to have him available in the case of a knockout round shootout. However, Zack Steffen and Matt Turner are no slouches at saving penalties either, so maybe they’ll be ok. Since about 75% of penalties are converted anyway, it’s a bit hard for a keeper to be bad at penalties and most other MLS playoff keepers are pretty much average. That said, based upon their penalty dossier, we can be reasonably confident that Joe Willis, Steve Clark, and Stefan Frei will let in more penalties than the average keeper.

As with much of soccer analytics, the main limitation of this study is the need for more data. Penalties are rare and not many players end up taking the penalties that do occur. These small numbers also limit the analysis that can be done. To make things worse, FBref does not have a full accounting of penalties, as many leagues and seasons just don’t have any data. Additionally, shootouts and international games are not included.

It would be nice to break goalkeepers down more into actual saves vs. off target shots. The former may be more in the keeper’s control, perhaps they look at the way the shooter angles their wrists like Nick Rimando did (Rimando would be in the Steffen and Turner tier with a smaller credible interval). The latter is probably more out of their control, but perhaps a keeper’s “presence” or reputation would influence the taker. I did not take into account whether penalties were taken at home or on the road, perhaps that makes a difference.

I don’t know how much MLS teams think about this type of thing, or even if it is a good use of the team or an analyst’s time. But in the playoffs putting a bit of a thumb on the scale could mean the difference between advancing or going home. It seemed to work for the USWNT who learned from their 2016 Olympic shootout loss to get smart about penalties as former US Soccer Director of Analytics Tyler Heaps explained at NESSIS in 2019. So Óscar Pareja or Ronny Delia, if you are reading this, perhaps find a way to get someone else to take your penalties rather than Nani or Maxi Moralez. Taty Castellanos is a good choice for NYCFC. There isn’t a clear cut favorite for Orlando, but Daryl Dike and Tesho Akindele are three out of four, which isn’t terrible, so maybe give them a shot.