SPFL xG for Matchday 16

GriffithsICTI have been trying to come up with a good way to share expected goal totals each week for each SPFL match. I have been playing around with a few things, but nothing has satisfied me aesthetically, so for the time being I am just going to put them here in black and white and some thoughts about each match. You can find a quick refresher on what expected goals are and how I go about calculating them here.

St Johnstone 0.91 – Dundee 0.47

Aberdeen 1.15 – Ross County 0.67

Dundee United 0.06 – Hamilton 0.28

Kilmarnock 1.08 – Partick Thistle 1.71

Motherwell 1.73 – Hearts 1.24

Inverness CT 1.33 – Celtic 1.47

Boy howdy, doesn’t that Dundee United-Hamilton game seem like a real exciting affair?! How they got three goals out of that game is beyond me, but hey that’s football. I have been exploring some of Dundee United’s issues and am looking to post something hopefully explaining some of their troubles with analytics, but from this game, it doesn’t take Pep Guardiola, Arsene Wegner and the entire staff of Opta sitting around eating Nando’s to figure out Dundee United are struggling to create chances.

Celtic have been averaging 1.975 xG per match and earned 1.47 xG on Sunday when facing off with Inverness Caley Thistle, but watching this match live, I had to admire Caley’s first half defensive performance. Celtic got an early goal off of a nice run and finish from about 20 yards from Callum McGregor but for the rest of the half, Caley’s defense frustrated Celtic’s potent attack to great effect, holding Celtic to an xG of 0.32 in the first half. Celtic’s attack got going in the second half, with a much better xG total and were rewarded with two more goals to get the three points (including a goal that bounced off Carlton Cole’s foot, face, and off a Caley defender before going into the net, which roughly has an xG total of ¯\_()_/¯). Caley’s defense should be credited with a great effort, though moral victories don’t stop a team from being dragged into a relegation fight. Speaking of…

This week’s award for “Hey, Scottish Football isn’t Boring!” match of the week goes to Kilmarnock and Partick Thistle. The match had a modest xG total of 1.71 to 1.08 in favor of Thistle, but the goal total went 5-2 to Thistle. One of the five was an absolute screamer by Dundee United loanee Robbie Muirhead. With second last place in the SPFL leading to a relegation playoff with a Championship side, this was an important mid-year six pointer. After struggling in the beginning of the season, Thistle have won two and drew two in their last 5 matches, and with Dundee United floating towards relegation island, one has to wonder what United were thinking loaning the likes of Muirhead to Partick Thistle in the first place.

Expected Goal and the SPFL on the Team Level: St. Johnstone Hates You and Your xG Model

Last week I described how exactly I have been gathering xG data for the SPFL with the limited information that is available out there. I then shared the top performers individually in the SPFL in expected goals and discussed if those players were over/underperforming. We can use the same data on a team level and see which teams have been over and underperforming so far this season. We can also explore how St. Johnstone spit in the general direction of the concept of advanced stats and may force Seth at Fitba Fancy Stats  give up on football and focus on fly-fishing analytics.

You can see every team’s xG totals, xG conceded, and xG differential, among other stats on my Advanced Stats Team Table, which is updated weekly. Or just look down an inch.

Advanced Stats Table-2

https://public.tableau.com/views/SPFLAdvancedStatsTable/AdvancedStatsTable?:embed=y&:display_count=yes&:showTabs=y

https://public.tableau.com/views/xGpoints/xGTotalPoints?:embed=y&:display_count=yes&:showTabs=y

When we look at the xG total over points in the table, it should be no surprise to see Celtic far and away the leaders in both categories. Celtic’s chance creation has been great all year and has catapulted them to the top of the table. Surprisingly, the second highest xG total in the SPFL through 15 games has been Ross County. I have waxed on about my admiration for County Striker Liam Boyce, but the whole team has done surprisingly well in creating quality chances for the Staggies. If the Dingwall team can continue to produce offensive chances like they have, they have a strong chance of finishing in the top 6.

At the other end of the table, Inverness CT have really struggled to create chances and their position in the table reflects this. With Ryan Christie heading to Celtic early after an injury and Dani Lopez’s future with the club uncertain after a spitting incident, Caley need to focus on finding someone who can create some chances or they could find themselves fighting relegation either in last place or in the relegation playoff spot. Could this current squad hold off Rangers or Hibs in a playoff? That is probably a gamble the Highlanders do not want to take.

https://public.tableau.com/views/xGConcededpoints/xGConcededPoints?:embed=y&:display_count=yes&:showTabs=y

xG Conceded_Points

On the other end of the field, Celtic and, perhaps surprisingly again, Ross County are among the top xG Conceded teams in the league. Another surprise team at the top of the table, Hearts, owes much of their success to their stout defensive play. With an xG total of 13.01, good for third best in the league, Hearts back line have restricted and frustrated opponents and gotten enough from Juanama up front to stock up points. Hearts have shown that they have what it takes to fight for a European spot all season. With all the financial troubles of the end of the Mad Vlad Romanov, could even the most fervent Hearts supporters picture European football back at Tynecastle next year at the beginning of this season?

https://public.tableau.com/views/xGDifpoints/xGDiffPoints?:embed=y&:display_count=yes&:showTabs=y

xG Diff_Points

Combining xG for and xG against provides us with a chart that resembles the above charts, with many of the same high and low performers in similar spots as they are in the previous charts. I want to take this space to discuss St. Johnstone’s performance thus far in the season. In the table, the Saintees find themselves in fourth place, but only 3 points behind third place Aberdeen. For the longest time, St. Johnstone and Tommy Wright have been the scorn of the Scottish Football analytics community (all five of us), with much of their success reliant on Alan Mannus often pulling 3 points out of a hat. Whatever methods they have used and however unsustainable St. Johnstone’s performance was, it yielded them a Scottish Cup and a few Europa League qualifiers. This season, Mannus has had a decent but not spectacular season, with a save percentage of 68%. St. Johnstone have instead relied on the forever young 33 year old Steven MacLean (mentioned here) and Graham Cummins (quit giggling in the back!) to propel them up the table. The Saintees have a similar xG total and xG conceded totals as Aberdeen, but have one of the lowest Total Shot Ratio and Total Shot on Target Ratios in the league. With such a low amount of shots, one would think that St. Johnstone cannot sustain this success, but then again people have been thinking that for years now.

SPFL Expected Goals Leaders: Or Why Does the BBC Hate the SPFL?

One of the most talked about advanced stats in football today is expected goals. Often referred to in its shorthand abbreviation as ‘xG”, simply put expected goals is the likelihood of scoring based on where and how a shot occurs. While many guard their expected goals model with secrecy, there is a lot of great public work available on expected goals in football. The most famous of this work, is the expected goals manifesto Michael Caley wrote. Michael has been doing great work with expected goals for a long time, and I strongly suggest you give him a follow on twitter for his xG maps. With all this great public knowledge available, surely we can use it to track the progress of all the great strikers in the SPFL like Leigh Griffiths, Adam Rooney, Liam Boyce, and others, right? Erm, not quite. If you are reading this, you likely know that Scottish football is at the scrap end of the TV money table, meaning that a few selected SPFL games are on tv. With so few games on TV, anyone (like your humble author) wanting to calculate xG for the SPFL are reliant on the media to provide the data necessary to calculate it. And with the BBC and Opta teaming up to form a human ¯\_(ツ)_/¯ emoticon towards Scotland, we have to make due with what the Beeb gives us. Anyone who’s ever seen a SPFL live match report on the Beeb knows they use about six or seven descriptions to describe chances. Terms like “in the 6 yard box”, “to the right of the box”, and “outside the box” are the most that we get, the rest is left up to our imagination. Based on this, the great Ben Torvaney of Analytics FC created an “expected goals model” for the limited info we do get from the BBC (I believe the Championship in England gets the same cold shoulder live match descriptions as the SPFL). Friend of the blog, Seth Dobson of Fitba Fancy Stats has done some great work on the scoring rate of each of these locations, which I highly recommend. I have used Ben’s model to get the expected goal totals of every team and player in the SPFL and below are the top 11 (I can’t say no to you, Ali Crawford). Along with their xG totals, I have their xG compared to their actual goals, xG per 90 minutes compared to their actual goals per 90, and xG compared to the number of shots taken.

xG Table

Looking at expected goals compared to goals, the likes of Adam Rooney of Aberdeen and Steven MacLean of St. Johnstone are far and ahead of their goal totals based on their expected goals for the season. Rooney got a lot of those goals at the penalty spot during Aberdeen’s eight game win streak at the beginning of the season. All but one of MacLean’s goals came from open play, though it seems that this type of performance from him seems tough to continue. Chad Murphy wrote a very interesting piece about clubs selling overachieving strikers compared to their xG totals, and after reading that piece, Rooney and MacLean’s names jumped out in my head as perfect examples of this. Neither club seems to need to sell from a financial perspective, and who knows if anyone is banging down the doors in Aberdeen or Perth to get either Rooney or MacLean, but it might be smart to sell based on the likelihood of the two not being able to sustain these results.

xG_G

When we look at expected goals per 90 minutes compared to actual goals per 90 minutes, MacLean and Rooney stick out like a sore thumb again. However, if we look at the other end of the spectrum, we see a player who made headlines when he was subbed out earlier than he thought necessary. Kris Commons’ contributions to Celtic have been debated greatly since that day. He has the 2nd highest xG total in the SPFL, the 2nd highest xG per 90 minutes in the SPFL, yet only has 3 goals in league play. Taking a look at the SPFL Goals + Assists leaders, we see Commons with 5 assists in the league thus far, putting him at 0.9 Goals and Assists per 90 minutes, so Commons is definitely contributing to the Celtic attack in various ways, but should he be scoring more goals? Probably. Now please don’t throw your jacket at me, Kris.

xG p 90_G p 90

Leigh Griffiths has been a one man (not two man, much to the chagrin of much of the Celtic support who would like to return to the days of sewn leather balls, “getting stuck in!”, and playing two strikers up top) wrecking crew. Leading the league in goals, xG, and shots, Griffiths great play has moved Celtic to the top of the league. Much has been talked about the turn around last year Griffiths made when he drove himself to a Celtic reserve match against Sunderland just to get some game time. Since that trip down South, Griffiths has been fabulous and is a definite SPFL Player of the Year candidate (and his hair should be up for Comeback Player of the Year. It took awhile but the hair transplant is finally yielding results!) While Griffiths play has been great, let me take this opportunity to confess my undying man crush for Ross County’s Liam Boyce. Could Boyce have easily been included in the group of strikers over performing that likely won’t be able to replicate that form going forward? Yes  NO! YOU SHUT YOUR DIRTY MOUTH! HE WILL ONLY BE SOLD TO REAL MADRID TO REPLACE CR9! Boyce has been a big part of Ross County’s unexpected success thus far, and while his stats don’t scream “PROBABLY A LITTLE LUCKY!” like some other SPFL strikers do, I wouldn’t be shocked to see the former Cliftonville man come back to earth. In fact, I think we have seen a little bit of that, with Boyce not scoring in his last 4 matches (though those matches included fixtures against Celtic, Hearts, and Aberdeen the top 3 clubs in the table) and only 1 shot in 3 of those 4 matches. Could clubs be focusing their defensive efforts on Boyce?

xG_Shots

Advanced Stats for the SPFL After 14 Matches

Welcome to The Backless Rule. With this being the first real post of this blog (the actual first post on the page was created after a night of debate on twitter and originally posted on another site), I wanted to give a brief introduction. While there is a “About” section, I just wanted to re-iterate that I wouldn’t consider myself anything close to an expert on “advanced stats” or “analytics” in football. I have been interested in the various different new ways to quantify success and failure for football that have been developing lately, but when I searched for these new stats for the SPFL, I found very little. Opta provides data for the BBC, but, in perhaps a sign of the British Broadcasting Corporation’s interest in Scottish Football, they provide very little for public consumption for any fans of Scottish Leagues. This was frustrating, and since I grew up a punk rock kid I figured I might as well do it myself. In very punk fashion, there may be so experimentation and mistakes, but as long as we’ve got our three chords (xG, PDO, and TSoR?!) and the truth (England losing), we will see where this thing takes us. I am envisioning this site focusing on questions about Scottish Football rather than developing an expected goals model or some new revolutionary stat to answer all of football’s questions. I am more interested in finding out (or tying to at least) what on earth happened to Aberdeen after winning their first 8 matches or why Celtic fans demand their team play a two striker system, despite having one of the hottest strikers in Europe (who may be overachieving).  This post will serve to show what I have worked on and gathered 14 matches into the SPFL Premiership season thus far. I am also going to create a page with these graphs that are updated after every match. Continue reading “Advanced Stats for the SPFL After 14 Matches”