I took a rare few days off social media this weekend but before I shutdown my various streams on Friday, I noticed a curious exchange between a couple of agitated Nottingham Forest supporters squabbling with a football analytics academic about the Tricky Trees’ potential for automatic promotion from the Championship this season.
The suggestion from the Forest fans was that the league table was the most accurate assessment of their current campaign, and so naturally, an assault on the top-two was imminent. However, the “numbers nerd” (as he was callously called) suggested performance data pointed to a possible drop-off in results between now and May.
I won’t tackle that particular Forest-based debate head-on but the same archaic and humourless clichés were rife in the replies to the data expert. “The league table never lies” and “football is played on grass, not spreadsheets” are the classic hard-nosed traditionalist retorts, and you didn’t need to scroll far down the thread to tick them off your bingo cards.
I understand the resistance to modern day analysis, but the hostility towards performance data and its value is often slightly bewildering. Even after a 46-game campaign, the end-of-season league table can produce a couple of outliers and unexplainables, largely down to our old friend lady luck’s influence in a low-scoring sport. That’s just football.
But more often than not, the underlying metrics used by professional punters and bookmakers are considered to be the best barometer of how a team’s performing, more so than plain points on the board. And this was born out when looking at how the 2018/19 campaign concluded across the top four tiers of English football.
Last summer I compared top-three end-of-season league positions in the Premier League, Championship, League One and League Two with the top three performers across three key performance data indicators – Expected Goals (xG), xG from open play and shots attempted inside the penalty area – using ratio returns, and the results were remarkably accurate.
The top-three finishers filled 30 of the available 36 data slots – five more of those positions were taken by teams concluding the campaign in fourth, one of which only missed out on goal difference (Portsmouth), and another on the final day of the season (Mansfield). Colchester (8th) were the only team to finish below 4th and take a top-three data position.
Not bad, eh? It doesn’t require a rocket scientist to understand that if Team X are taking more shots than Team Y it will give the aggressors a stronger chance of succeeding. However, if Team X are also attempting efforts at goal from high-probability locations to improve shot conversion then naturally their prospects of winning are enhanced further.
As punters, we can’t watch 20+ matches every weekend and it’s impossible to see all 92 teams in action. Here, data is your friend. Would you continue to back a team that’s won 10 consecutive games 1-0 if you knew they had been convincingly out-shot in every game, comfortably losing the xG battle? You shouldn’t, so why ignore the evidence available?
So ignore basic results and don’t focus your attention on the league standings or form tables. The more accurate assessment in team strength can be found in performance data rankings so instead inspect shot tables and familiarise yourself with xG – after all, these are the metrics that bookmakers are using to set the odds and rate clubs in each division.
As for 2019/20, we’ve around a third of the regular season still to play and roughly 10 weeks of league football to come, so which sides are occupying those 36 aforementioned top-three berths for xG, xG from open play and shots attempted inside the penalty area?
Manchester City, Liverpool and Chelsea (in that order) take top-three crowns across the three metrics with Leicester in pursuit. The numbers suggest defending champions City are drastically underperforming, as well as Chelsea, who trail the Foxes by eight points in the league table following their enjoyable 2-2 draw at the King Power Stadium on Saturday.
Leeds and Brentford dominate the top two positions in the Championship for all three attributes but Sheffield Wednesday surprisingly take two of the nine slots. League leaders West Brom can only call upon a ratio ranking of 3rd in the xG standings, occupying 5th and 6th slots elsewhere, hence the markets reluctance to make the Baggies outright favourites.
League One has proven devilishly difficult to second-guess this season but Portsmouth and Peterborough are hogging five of the nine top-three positions across the three metrics. Oxford have squeezed in two top-three berths with current table-toppers Rotherham joined by upwardly-mobile Gillingham, whereas second-placed Wycombe’s highest rank is 12th.
Finally in League Two, Crewe and Swindon’s eye-catching campaigns are reflected by their positions in the performance data rankings with both clubs occupying the top-two for all three indicators. Newport’s rock-solid defence earns the Exiles two top-three positions, leaving Colchester to fill the ninth available slots thanks to their ever-improving efforts.
Based on the evidence to-hand at the time of writing, the underlying data points to four potential longer-term selections: Brentford to win the Championship (8/1 Bet365), Portsmouth (5/1 Bet365) to take League One honours, Crewe (4/1 Betway) to be crowned League Two kings and Newport (12/1 Bet365) to seal promotion from League Two.
Of course, I don’t expect you to blindly follow the above recommendations – far from it – but we can revisit the wagers in May, and also review the final data rankings alongside the corresponding league tables to see quite how accurate (or not) the underlying metrics proved to be mid-season. My presumption is, they won’t be a million miles away.