July P Score: Reactive Games Lead to Fewer Goals

/

Change was good for some teams in July and not for others. FC Dallas and Colorado Rapids had perfect months while playing more proactively than earlier in the season. Meanwhile, Orlando City SC and Seattle Sounders had difficult months and were more reactive than they had ever been.

Below is the latest P Score table. Orlando still maintains their top spot, but Montreal caught up by being the most proactive team in July. This month I added each team’s opponent's P Score. Interestingly but not surprisingly, teams play more reactively against the New York Red Bulls and Sporting Kansas City.

Read More

How DC United continues to beat the odds

/

MLS will not be a top league until its coaches find better ways to connect projection metrics to tactical changes. Ben Olsen is showing he may be on the right track. The more data that is collected on variables such as game variability, individual player contributions and spatial frequency, the more information there is to digest and to affect playing, teaching and coaching. Therefore the issue lies in the translation of such data into meaningful, impactful resolutions that can be measured by on-field progress.

Read More

Weekend Kickoff

/

In most sports after an All-Star game, the remaining portion of the schedule is called the “second half of the season.” But with the manner in which Major League Soccer administers it's schedule and the general accumulation of games that are played during that time period,. I look at it slightly different. I look at pre-All-Star game as the first act, the All-Star game is intermission, and post All-Star is the second act.

In other words, Soccer is straight drama. The great Bill Shakespeare once said:

Read More

Value Available vs Value on Field: Evaluating MLS Salary Cap Management

/
Value Available vs Value on Field: Evaluating MLS Salary Cap Management

This post sets out to use expansion side Orlando City as an example of how value in MLS can be calculated, and how a teams’ management of the salary cap can be scrutinized using a simple visualization.

With the release of the salary cap last week there are a hundred and one things I want to do with this data. The first idea is something that myself and Ben Torvaney thought of when discussing what we would do if we had the salary data of the Premier League and other European leagues if it was available like it is in MLS (fun topic of conversation I know). Evidently, that data is unlikely to ever be released for any of the big leagues, so MLS is the only league these ideas can be applied to.

Read More

Sizing up the Silva for Saborio swap

/

Last week, the Alvaro Saborio for Luis Silva trade kind of took the league by surprise. Nobody saw this coming, but after the dust settled this trade makes perfect sense for both parties involved. For DC, they give up a promising youngster for a proven goal scorer they badly need. For Real Salt Lake, they pick up an up and coming midfielder who can help rebuild an aging RSL side.

But for United, this is a move to win now and to take advantage of their favorable table positioning to make a serious run for the Supporter's Shield in a weaker Eastern Conference and a possible deep run for the MLS Cup too. It's a "win now or never" kind of mentality and it's one that will most likely pay off.

Read More

Introducing MLS ELO Ratings

/

Today I'm excited to debut my MLS Elo Ratings. The Elo Rating system itself was originally invented by Arpad Elo to rate chess players more effectively. It's not a perfect system for soccer, as the game is more fluid and random than chess, but it's a good indicator of under/over performance than basic stats alone provide. 

Elo is a zero sum system, meaning that sides are allocated equal points depending on the outcome of a game. It rewards more points to teams who cause upsets - so when the Chicago Fire beat the Seattle Sounders they were awarded a large number of points, and the same number of points were taken away from Seattle. 

Read More

USMNT 1-2 Jamaica: The Reggae Boyz clear the dance floor

/

Hey U.S. fans, look on the bright side. We get an extra soccer game this fall! The USMNT will be in a one game playoff against either Mexico or Jamaica for the privilege of representing CONCACAF in the 2017 Confederations Cup. That bit of fun was brought to you by a loss of stunning proportion to Jamaica. The U.S. gave up two goals in five minutes off of set pieces in the first half and couldn’t mount a useful attack against a determined Jamaican defense. The 2-1 loss, the first to a Caribbean side on U.S. soil since 1968, will sting for a long time, especially for yours truly who was looking forward to going to the Gold Cup Final to watch the U.S. with his son. Not all stories have Hollywood endings. And certainly sports wouldn’t be sports without the heartbreak.

This game was a perfect example of why soccer statistics can sometimes lie. If you didn’t watch the game and just looked at the box score you might think that the U.S. was simply unlucky. They held 60% of the possession and completed 82% of their passes. They outshot the Jamaicans 20-10 and put 7 more shots on goal (10-3). The U.S. won the expected goal battle by a score of 2.3.-1.0, but looked up at the scoreboard at the end and saw the final score was actually reversed.

Read More

An Early Look at The 2019 USWNT World Cup Roster

/

We’ve got the three stars. We’ve had ourselves a ticker tape parade. And while I don’t think the celebration should end anytime soon, it’s been long enough since the final whistle blew in Vancouver to start thinking about the future of the USWNT.

Using logistic regression on data from all previous WWCs and factoring in position, age, and whether or not the player started the final competitive match in the previous World Cup*, I constructed a model that produces a very rough probability that a player that made the 2015 roster will make the cut for France 2019. Obviously, four years is a long time and a lot can change, so these probabilities are hardly precise and should be taken with a large grain of salt. I have not included goalkeepers, as the very small sample size makes things difficult to predict, and have also left out players that we already know will not play in 2019. Furthermore, I have included forwards but the sample size is also very small, making the probabilities even less precise than that of defenders and midfielders. So at the risk of getting way ahead of ourselves, here are the very rough probabilities that the members of the current squad will be back to defend the Cup in France:

Read More

USMNT 6-0 Cuba: The U.S. Smokes Cuba

/

On Saturday the U.S.A. cruised to a 6-0 victory over Cuba in the Gold Cup quarterfinals and advanced to play Jamaica in the semifinals. When a country with the population and the financial resources of the U.S. pounds on a country whose players are much more intrigued with the idea of leaving the team, it’s hard to get too excited about the victory. It’s an even harder match to break down statistically. How do you analyze a drubbing? Let’s just all feel good, right? Believe it or not I’ve found some statistics that will sober you right up.

Read More

Measuring Goalkeepers: The Limitations of Goals - xGoals

/

Last month we took a look at how bad the Goals Against Average (GAA) stat is for evaluating goalkeepers and how the save percentage (S%) is slightly better. To show the inefficiencies of the GAA and S%, I compared them to ASA’s own unique stat of Goals Minus Expected Goals (GmxG). The GmxG looks at where shots are taken, calculating the likelihood of a shot going in from that distance and angle to goal, which ends up telling us if a goalkeeper is giving up too many goals given the circumstances. This is great for a couple of reasons, the main one being that we have a more accurate reading on shots on goal than the ol’ shot percentage. If one goalkeeper consistently sees shots from distance while another is routinely left out to dry by his defense, the GmxG will let us know how many goals each goalkeeper should be conceding even if their SOG and goals are similar. However there are some shortcomings with any stat and the GmxG isn’t any different.

Read More