Welcome back for the 2018 Edition of MLS Fantasy! For those new players or those new to MLS Fantasy Boss, this space focuses on how to xG (expected goals) to give you a leg up on then competition. Each week I’ll put out weekly rankings to help identify the best matchups to focus when choosing your players. Fair warning, this will be one of the nerdier spaces for fantasy thoughts. I’ll do my best to explain along the way.
For this first article of 2018, we’ll look at what the 2017 xG data can teach us about choosing our squads in 2018.
xG vs. real goals
Let’s start with how good xG are at predicting actual goals and goals against.
The graphs show each teams average goals per game plotted against their average xG/per game for their offense/defense and home/away. The R2 value in the top right corner gives an indication how good xG are at predicting that corresponding actual goal numbers. The larger the number, better correlated they are.
Each graph shows a positive correlation between expected goals and goals, but the R2 values indicate the different magnitude of the correlation in each category.
The key takeaway here is that xG is best correlated to actual defensive performance. Fantasy takeaway: the weekly xG rankings are best at indicating what goalkeepers and defenders to choose.
Next, let’s take a look at each team’s xG and actual goal averages from 2017.
Sortable table available here: https://docs.google.com/spreadsheets/d/1acB5Mh3hehetb51L7-zKMwavmVXfT5CzuR2v26EaAGE/edit?usp=sharing
The interesting columns to look at here are the xG-G columns. For offense, positive differences mean that the team outperformed their xG in 2017 and may be due for a regression, while negative differences mean that the team underperformed their xG in 2017. For defense, positive differences mean that the team gave up fewer goals than expected and their defensive numbers may be worse than they appear.
For example, the top 3 underperforming home offenses in 2017 were DC United, LA Galaxy, and NYRB. The top 3 over performing home offenses were Toronto, Atlanta United, and Chicago. What you make of these stats all depends on what you think of the quality of finishers you think each team has.
xG and offensive/defensive consistency
Next, let’s take a look at the average xG numbers for 2017.
Sortable table available here: https://docs.google.com/spreadsheets/d/19U7EIWNh5lwX1-ptJI-ZDmJCxVLNxpF8HTfeVoqjJg8/edit?usp=sharing
There’s a lot of information in this chart, so let’s break it down.
The chart is broken down into the average xG for Home Team Offense/Defense and Away Team Offense/Defense and the “consistency” rating for each team in each category (stats aside: this is the standard deviation of the xG for each team/category. This means that 67% of the games last season, the xG for that team/category combination fell in the range of Average xG ± “consistency”). In each category, green is good, red is bad.
Home offenses score 0.5 more xG at Home than Away teams do. The reverse is true for Home Defenses, which give up 0.5 xG fewer at Home than Away teams do. In fact, no team in the league averaged a higher xG total than the average for home teams. Fantasy takeaway: choose players from Home teams.
So how should you think about the consistency values? The smaller the value, the more consistently that team scored around its average xG number. So a team like Houston with an average xG of 2.01 and comparably small consistency value of 0.72 means that if you chose a Houston player at home, the offense was more likely to perform well. Compare this to a team like NYRB who average 1.91 xG, but have a larger consistency value of 1.01. This translates to NYRB has some big games at Home but also some bust games. Then take a look a team like DC United. They grade out as the most consistent Home Offense, but it means their offense was consistently terrible! Fantasy takeaway: Look for teams with a high average xG and a small consistency value on offense and a low average xG and small consistency value on defense. (Look for green and green)
xG vs. PP90
Finally, let’s look at how xG and xG against relate to PP90 for each position on the fantasy pitch.
Sortable table available here: https://docs.google.com/spreadsheets/d/1dr1Q-ffRzFPoe_fLOyJYdU2LCGkCvgUPqriP_Q2L_rs/edit?usp=sharing
Looking at the xG and xG against data vs. fantasy points shows how strongly the PP90 of each position for each team correlates to the xG for or against in that game.
The strongest correlation (R2) is for xG against and DEF PP90. This is followed by GK PP90, FWD PP90, and MID PP90.
Fantasy Takeaway: xG against works well for choosing defenders. xG numbers are slightly less good at the other positions, but still shows a positive correlation, account for ~30% of the variation.
Week 1 rankings will be out in a couple of days!
Hit me up the comments with questions or for a discussion!