You know that old saying in football that the ‘league table never lies’? Well, the truth is that it kinda does.
Don’t get us wrong, after 38 or 46 games the league table DOES NOT lie. The best team wins and the worse three or four relegated – that is irrefutable. But in the early going of the campaign the league table DOES lie, and that’s for one very obvious reason: all teams start on zero points.
That’s the fairest way of course, but it doesn’t make any logical sense to punters, because….
- All teams AREN’T equal – Chelsea are better than West Brom, for instance, so the fact they both start on nil point is irrelevant
- All wins are worth three points, whether you beat Arsenal or Burnley. Is that a fair representation?
- Some teams have easier early season fixtures than others, so the league table is not a fair reflection of ability
The reason we bring this up is to highlight how early season football fixtures can do strange things to the psyche of punters. Everton may top the table after four games and some bettors will think ‘watch out for Ronald Koeman’s side this year!’, but over the fullness of time mindsets like those are shown to be hokum (barring Leicester City’s almighty anomaly in 2015/16).
So we want to introduce you to another method of ranking teams in the same division: ELO Ratings.
Say ELO to a New Way
The ELO system was founded by Arpad Elo, a chess player in the US who felt that the standard league ladder system used by the United States Chess Federation was not representative of the players’ abilities.
So he went about devising his own system that created rankings based on the perceived greatness of an achievement. The theory is that by winning more points for beating a team ranked first than twentieth, the league table would ultimately be a true reflection of the skill levels of each team (or player, in his case).
The actual mathematics of Elo’s model can be difficult for the layman to follow – we’re talking distributions, variance and standard deviation here, the sort of thing long since forgotten from school days. But the key point is that the general methodology can be tweaked to create a basic system that more accurately ranks football teams in their respective divisions; particularly in the first half of the season before all teams deviate to their norm.
So, we’ve devised our own system based on ELO Ratings, and you are welcome to crib it and modify it as you wish.
One of the bizarre things about sport, in our opinion, is that the new season does not take into consideration past events. That is only fair, of course, but for Chelsea to start on the same number of points as Watford, for instance, feels strange given that they ended up a whopping 53 points apart at the end of the 2016/17 Premier League campaign.
So our own ratings system takes into consideration past performance, and ensures that the new league table reflects that of the old, with Newcastle, Brighton and Huddersfield, the three promoted sides, nestled at the bottom.
The conclusion we can seek? If a team climbs our table, or slips down a few places, it is a sure sign that they have progressed or regressed over the summer.
Our system also ensures that every game is worth an appropriate amount of points based upon the qualities of the two sides. So, Tottenham will gain more points by beating Manchester United than they do Crystal Palace.
It is not an exact science by any means, and the more mathematically minded among you might want to tweak the system to your own specifications. But we think it should work quite well, and provide punters with a more accurate assessment of where two teams are heading at any given point.
The System in Action
So, first things first, we must create a ‘ground zero’ for the system to get started with. So we’ve created an arbitrary numerical system based on last season’s performance, with Chelsea -as defending champions – starting with 2000 points, and then each subsequent side having 25 points less, all the way down to Huddersfield (who were the ‘third’ team to get promoted from the Championship) on 1525 points.
It looks like this:
|West Bromwich Albion||1775|
|West Ham United||1750|
|Brighton & Hove Albion||1550|
So that’s our starting point, and while open to interpretation offers a reasonable accurate assessment of where the 20 Premier League teams are right now.
So how do we move it on from here? That comes by changing the points as matches are played. We need a method that is fair based on the teams’ relative quality, and so the easiest way is to create a bank that each team plays for based on their current points tally.
Let us explain:
Arsenal play Leicester on Friday night, and so we need to configure a way that accurately reflects the outcome of the game.
So our system requires teams to ‘give up’ 5% of their points tally, with the winner taking the lot. As such, Arsenal put 95 points (1900/20) on the line and Leicester 86.25 (1725/20).
If Arsenal win, they take Leicester’s 86.25 points and their tally becomes 1986.25, while Leicester lose their 86.25 to become 1638.75. The reverse happens if the Foxes triumph, and the league table alters accordingly.
Calculating the points payout on a draw is easy enough: we add Arsenal’s 95 to Leicester’s 86.25 to get 181.25, divide by two and get a bounty of 90.6.
Arsenal’s recalculation would be (1900 – 95) + 90.6, and Leicester’s would be (1725 – 86.25) + 90.6.
Ultimately, this model should show a better representation of relative quality than the standard ‘three points for a win’ format.