In Praise of Jill Ellis

It's not clickbait if you're already subscribed to the newsletter!

In the intro to one of his not-quite-as-well-known books, The Undoing Project, Michael Lewis writes about Houston Rockets general manager Daryl Morey. Morey had been at the vanguard of the NBA’s “Three Is Worth More Than Two” era, not quite solving basketball, but pushing it toward a new tactical approach that created more value with every single possession. 

However, when it came to drafting players, Morey was still a long way away from anything close to an “optimal” approach. He’d created a statistical model, but it didn’t lead to a higher hit rate than the average team would expect. As for traditional scouting methods, Lewis writes:

A scout watching a player tended to form a near-instant impression, around which all other data tended to organize itself. “Confirmation bias,” he’d heard this called. The human mind was just bad at seeing things it did not expect to see, and a bit too eager to see what it expected to see. “Confirmation bias is the most insidious because you don’t even realize it is happening,” he said. A scout would settle on an opinion about a player and then arrange the evidence to support that opinion. “The classic thing,” said Morey, “and this happens all the time with guys: If you don’t like a prospect, you say he has no position. If you like him, you say he’s multipositional. If you like a player, you compare his body to someone good. If you don’t like him, you compare him to someone who sucks.” Whatever prejudice a person brought to the business of selecting amateur players he tended to preserve, even when it served him badly, because he was always looking to have that prejudice confirmed.

I think about the seeming absurdity of this idea a lot -- to better understand sports, you should ... stop watching sports. Morey’s thinking here is a more advanced version of what Bill James, one of the pioneers of baseball’s stats revolution, once wrote:

One absolutely cannot tell, by watching, the difference between a .300 hitter and a .275 hitter. The difference is one hit every two weeks. ... The difference between a good and an average hitter is simply not visible—it is a matter of record.

When I was watching the US-England World Cup semi final last week, some time in the second half, the American keeper, Alyssa Naeher, took a goal kick. She lumped the up ball toward midfielder Lindsey Horan, who leapt up and flicked the ball on to one of the three attackers ahead of her. This was after Horan had created the US’s second goal with a combination of supernormal spatial awareness and deft dexterity -- a little leg-shake to buy herself time and then a hooked chip so perfectly placed that it looked like it was buffering in mid-air:

Now, she was dunking on England’s midfield in a show of sheer, brute athleticism. For all those reasons put together, I love watching Horan play. And so, my next thought was, Man, Horan is dominating England in the air today! It seemed in keeping with Horan’s overall-stellar performance on the day ... but it wasn’t. She was one for eight in aerial duels against the British.

One of the reasons that’s often given for why soccer resists quantification is that it’s too complex and too dynamic. To me, this is exactly why we should try to find ways to better understand it! I enjoy reading all kinds of tactical analysis. The historical inflection points (when a new, never-before-seen style becomes unstoppably dominant before its co opted by everyone else, and then the cycle begins again) and the ongoing philosophical debate about how the game should be played are two of my favorite things about the sport. However, I’ve heard the practice described as something more akin to art criticism, rather than analysis. And I think that’s right, as all tactical analysis is a work of interpreting and then value-judging. Even if you hooked the manager and all of the players up to a lie-detector test, I’m not sure you’d be able to paint a perfect picture of why they all did the things they did in a given match. Intentionality is impossible grasp. Take it from Dani Alves:

I also don’t think you can really understand what happened in a game just by watching. TV camera angles don’t show the whole field, and, on average, there are close to 150 individual possessions in every match. There are 22 players to keep track of at all times. There were nearly 500 passes completed in the World Cup final on Sunday. My brain certainly isn’t capable of processing all of that and then accurately presenting it any kind of coherent matter.

Despite that, I certainly fell into the camp that said the US won the World Cup in spite of its manager, rather than because of her. I mean, this US side was perhaps the most talented soccer team of all time, so that was always going to be the case, no matter who was on the sideline. The goal of a coach for a group this good is simple: don’t fuck it up. Jill Ellis, though, frequently benched Horan, who, pretty clearly was one of the best players at the tournament. And it certainly felt like the team kept eschewing the kind of high-degree-of-difficulty approaches that an uber-skilled team can take, in favor of easier-to-execute-but-less-likely-to-pay-off options: build-up play on the ground through the midfield rather than up the wings, trying to create numerical mismatches in the final third rather than opting for low-probability crosses.

But we’re a couple days removed now. And I think if you step back and take a look at everything that happened, a clearer picture starts to emerge: The defining feature of the USWNT’s tactical style is that it changed with their opponents.

STATS LLC have developed a system that aims to objectively calculate a team’s playing style. Here’s how they describe it:

Football is such a diverse sport with numerous tactics, unique scenarios and systems of playing. In order to capture the different ways a team plays, STATS have developed a framework that captures the playing style of a team throughout a match.

The framework has moved away from the traditional accumulations of single events to provide an impression of the way a team plays. STATS Playing Styles is a multivariate approach, therefore taking into consideration numerous events and factors that determine a team’s style of playing.

And here’s how the US looked across their seven wins in France, compared with their opponents. Bold numbers represent the number of possessions per game. The percentages represent how far below or above average for the tournament those numbers were. And the dotted-outline is the tournament average:

STATS define “maintenance” as “possessions in which a team looks to maintain and secure possession of the ball within the defensive area of the pitch”. “Build Up” is roughly defined as possessions between the midfield line and the penalty area. “Sustained Threat” is a possession that lasts more than six seconds in the attacking third. And “Fast Tempo” is a possession in the attacking half where a player releases the ball within two seconds and/or a player dribbles at a speed above 4 meters per second. The other phases are self-explanatory, I think, but these four have some overlap and a single possession can count as multiple phases.

So! On the whole, the US played at a fast pace and frequently won the ball back in attacking positions, compared to the field. They also racked up a bunch of attacking possession, while spending very little time on tame passing in their own half or direct passes up to the attackers. Outside of the heavy emphasis on crossing, that’s a really effective playing profile, especially when you combine it with a defense that allowed very little of anything outside of sterile possession and some long balls.

Of course, that profile includes a demolition of Thailand, an easy win over Chile, and a match against an under-staffed Sweden side. As the competition increased, the US’s style shifted. This is how it looked in the knockout rounds:

They still maintained average levels of attacking possession, but significantly cut back on crossing the ball in favor of more long balls and counter-attacks. The defense, meanwhile, still managed to face below-average levels of possession in their own half and their own final third. The new, less aggressive attacking approach surely played a part in keeping some of the best teams in the world at arm’s length. The US averaged 66 percent possession in the group stages, and that number dropped to 48 percent from the Round of 16 onward. Well, it worked! (Dotted line is the tournament average.)

As for the final against the Netherlands, it was sort of a hybrid of everything the US did in the previous six games. They possessed the ball in dangerous areas, and they counter-attacked as frequently as any team had all tournament.

The US allowed very little dangerous possession, and for the moves that did make their way into the American half,  you can clearly see how they countered it: with counter-attack after counter-attack. Otherwise, the Dutch were forced to keep attempting balls over the top -- a losing strategy against the most athletic side in the tournament.

This team won it all because of a generation of iconic players who never stopped talking shit and haven’t really ever stopped backing it up. The absurd meta-narratives that kept cropping up around them, to me, were more of a symptom of their dominance than anything else: first, people got mad that they scored too many goals, and then they got mad about tea, and then they got made that they wanted to get paid. But maybe their coach does deserve some credit, too. She’s won two World Cups in a row, she had a plan, and the plan came off. The team shape-shifted throughout the competition, but there is one aspect of their play that remained constant: they spent very little time in the “maintenance” phase, unconcerned with low-value passing in and around their own goal. Fittingly, in other words, they didn’t mess around. I know of at least one NBA GM who can appreciate that