AROUND THE LEAGUE WITH REINA: SIGNAL

If you took a break from your daily routine to read this article, I applaud you. Whether inadvertently or not, you’ve just entered an intimate circle of people who see the world differently. Thinking about sport probabilistically does not come naturally to most. In fact, it often lives at the fringes of fandom. Data and statistical analysis feel more at home in the offices and coffee shops rather than pubs and arenas, but we are here to change that.

Using data gathered from the league’s website, we can track whether teams outshot their opponents, how many points they should have accumulated based on their shooting performances, how they stack up against other teams over the season, and which players and teams convert chances into goals at higher rates than others.

These statistics allow us to peek behind the curtain of what it takes to win an indoor soccer game like never before, as we attempt to separate the signal from the noise. Noise refers to the unexplainable randomness inherent in data, while signal represents the skill and ability as exercised by the teams and players on the turf. The signal is what we’re after.

There is a lot of randomness that comes along when collecting data, but it’s our job to discern what is signal and what is noise based on our understanding of how the game is played and what each statistic tells us. Think of statistics like puzzle pieces that help us form a complete image of what happened in a game or a season.

In a sport like indoor soccer, you cannot simply read the final score to understand what happened in those 60 minutes. Goals are incredibly noisy, that is, beholden to randomness. A more effective way to assess a team’s ability to win games is to examine shot creation and prevention, utilizing a statistic derived from a close relative of indoor soccer.

Hoxie, Hoxie Points, and Hoxie Differential

In hockey, tracking each team’s shots for minus their shots against is a valuable determinant of team skill. Teams that consistently outshoot their opponents tend to win more hockey games, and vice versa. Indoor soccer shares many of the same characteristics as hockey, and as such, we can track the same statistics and gain a comparable understanding of team skill.

This statistic is named “Corsi” after a former Buffalo Sabres coach named Jim Corsi, simply because the analyst who created the stat liked Corsi’s mustache and wanted to honor him. To better recognize indoor soccer’s vast assortment of mustachioed participants, I have opted for “Hoxie” in honor of Andrew Hoxie, another famed and beloved mustache wearer.

With this in mind, we can look at each team’s Hoxie to see how many shots they took and how many shots they faced over the entire regular season. This allows us to identify which teams consistently outshoot their opponents, which typically indicates an above-average ability to win soccer games.

Hoxie Formula: Shots For - Shots Against = Hoxie Score

We can also examine each game and generate a HoxieScore by determining the degree to which a team outshot its opponent on any given night and assigning a point value based on the likelihood of that Hoxie performance resulting in a win, a tie, or a loss.

If a team outshot their opponent by more than three shots, they earned three Hoxie Points. Conversely, if a team conceded over three more shots than they took, they would earn zero Hoxie Points. If the shot totals were tied, each team earned one point, and if a team outshot their opponent by only one, two, or three shots, they earned two points. Totaling these up over the course of a season allows us to build a Hoxie table where we can see which teams “should” have won or lost more games than they actually did.

For example, last season’s champions, Chihuahua, clinched their first-ever MASL Shield title with 58 regular-season points, while Empire qualified for the playoffs in 7th with 37 points, 21 fewer than the Savage. Using our Hoxie Scores and Hoxie Points, we learn that the Strykers actually outshot their opponents at a comparable rate to the Savage, and based on each team’s individual game Hoxie performances, both teams would have been expected to finish the regular season with 58 points.

The variance, or Hoxie Differential, between what we would have expected to happen (Hoxie Points) and what really did happen (Actual Points) can be explained through several factors, including player skill, game state, play styles, luck, etc. This is where our ability to watch games and discern what is happening comes in. Like a flashlight in the dark, Hoxie tells us where to look and what to look for when analyzing the actual games.

Teams that overperform their expectations will likely attribute the aberration to skill, and they’re not entirely incorrect in that assessment. However, luck plays a more significant role in the sport than people might like to admit. Conversely, teams that underperform would likely prefer to attribute their struggles to luck rather than a lack of skill.

Scoring goals is incredibly random, but shot creation is much less so. Hoxie provides a basic understanding of each team’s ability to create and prevent chances, but another advanced stat helps us quantify which teams sink or swim in front of goal.

GPS

In our game, as with outdoor soccer, shooting conversion is extremely inconsistent at the individual level, but becomes more predictable when viewed from a broader perspective. One player may go through several phases finishing at higher or lower rates, but over the course of the season, the league-wide standard remains consistent. Teams are no different.

When taking every team’s conversion rate into account, the MASL average Goals Per Shot (GPS) was 0.223, meaning the average shot taken last season was worth that many goals. You could also say that if the average player were to take 100 shots, they would be expected to score 22 times on average. This acts as a measure of efficiency when judging attacking or defensive prowess.

GPS Formula: Goals / Shots = Goals Per Shot (GPS)

For instance, if Team A takes 10 shots in a game and scores two goals, they will have a GPS of 0.2, or 20% which is slightly below the league average conversion rate of 0.223, but not abnormal. However, if we say, hypothetically, that Team A has a GPS of 0.297 for the entire season, this game could be considered an underperformance.

GPS is a way to gauge a team’s performance relative to its own and that of other teams in the league. It contextualizes what we’re seeing when we watch not just one game, but an entire season. Teams may start the season strong, with a GPS of 0.316, only to hit a rough patch of games in January and February where they average 0.165.

Team A may win the same number of games in both stretches, but with our data in hand, we can tell when there are abnormalities. Perhaps they opted for a new system, or their star forward got hurt; it’s up to us to add context and nuance by combining what we’re seeing on the field with what we’re seeing in the data.

xG, GMxG, xGAMGA, xGD, and Kelvin

If you’ve learned anything about outdoor soccer or hockey data analytics, you’ve likely heard of expected goals, which track and attempt to quantify the likelihood of every shot becoming a goal. Advanced models consider various factors, including the distance from the goal, the body part used to take the shot, and the goalkeeper’s positioning, among others. That data is not yet available in our game, and as such, should not be considered when we discuss expected goals.

Based on the data from last season, our rudimentary Expected Goals (xG) model creates a league-wide baseline for shooting efficiency. Using last season’s league-wide GPS from earlier, 0.223, we can multiply each team’s total shots over the course of the season by that figure to visualize how many goals each team or player would be expected to score if it were the average player, taking the average shot.

xG Formula: Total Shots x 0.223 = Expected Goals or xG

Remember, this does not take into consideration the value of each shot; instead, it uses the average as a standard to measure against. By subtracting each team’s xG from their actual goals scored (GMxG), we can tell which teams are overperforming in front of goal. Teams in this camp may be generating a higher percentage of dangerous opportunities and, as such, converting them at a higher rate.

GMxG Formula: Goals - Expected Goals = GMxG

On the other hand, if a team is underperforming their GMxG, we can use other interpretive measures to understand why. Perhaps, they conceded several early goals and played three quarters against a team that showed no interest in attacking. These statistics provide a clearer understanding of which teams convert shots into goals at a higher or lower rate than the league average.

Defensively, we can also track xG against minus goals against (xGAMGA) to see which teams prevent goals at a higher rate. This could be due to a high percentage of blocked shots, a goalkeeper stranding on their head, or good old-fashioned luck. Nuance is everywhere for those with the eyes to see.

Now, let’s get complicated. We can also track each team’s expected goals against (xGA) for an understanding of their relative defensive proficiency. Using the expected goal differential (xGD) allows us to identify which teams excel in both phases of play. From there, we can subtract the teams’ actual Goal Differential (GD) from their xGD to determine which teams underperformed and overperformed in both phases of play (GDMxGD). I’m calling this stat “Kelvin”, in honor of Kelvin Oliveira, who overperformed his xG last season by over 11 goals.

Kelvin Formula: Goal Differential - Expected Goal Differential = Kelvin

Kelvin acts as a puzzle piece alongside Hoxie, GPS, and xG, and the analysis we get from actually watching the games. Individually, they don’t tell the whole story, but together, they’re starting to create an image.

These statistics can serve as a basis for understanding which teams are actually creating chances at an elite level and which teams are converting those chances at an elite level. I like to think of Hoxie as explaining what leads to a shot, and Kelvin as describing what happens when the ball leaves their foot, head, etc.

Thinking probabilistically could be the key to learning more about this sport than we ever thought possible. As we utilize these statistics in the future, we’ll review each team’s profile to identify areas of strength and weakness, explore individual performances, debate player accolades, and attempt to separate the signal from the noise inherent in our sport.