What if the best team DID always win?
Written by: Jamie Kilday @FitbaInScotland
For those unacquainted with Expected Points I wrote an article (re)introducing it at the beginning of the season, which you can read here. In short it is taking the number of shots, the Expected Goals from those shots and simulating the result of the game through a Monte Carlo simulation. With this, you take the percentage of times a team wins and multiply it by 3, and add the percentage of times the team draws.
One way to think of it would be if you have ever played Football Manager and restarted a save after losing a big game. Imagine if when you closed your save, the game told you how likely it would be that you win on the second play through.
For some this sounds a bit too far-fetched and I get it. You are probably thinking that surely the outcome of the game is…well…just that, the outcome of the game. If you can give me a minute, I’ll explain why Expected Points is not only quite useful but could well be the advanced stat for Da’s (maybe).
When you start talking about advanced stats, sceptics usually say that it takes the fun out of the game but if there are one thing football fans love more than talking about football its being right! So, if you have ever walked out of a game and thought “We should have won that game comfortably” or “We were lucky to get a point out of that one” then Expected Points can give you that warm fuzzy feeling of being right.
As well as that, it serves as a quick marker for how well a side is performing compared to actual results and is something of a benchmark to be used early in exploratory team analysis. It can flag that there could be something in a team’s underlying numbers that could be exploited.
A large over performance in Expected Points is based on a high conversion rate. This could mean that a side is on a hot streak that is most probably unsustainable or that the side has a system that isn’t particularly dominant so they hit teams on the break, this generally can produce fewer chances but of higher quality. Conversely an under performance could mean that the side has a low conversion rate, either in general or as the result of playing a very dominant style that produces a lot of low-quality chances.
Here is how the Expected Points table looks going into the split;
At the top of the table there is very little change, with the only talking point being that Aberdeen should probably be sitting above an over-performing Kilmarnock.
If we look at the foot of the table, as we go into what looks to be one of the closest relegation battles the Scottish Premiership has ever had, the underlying numbers show that the gap should be even closer. Hamilton’s points are in line with what we would expect but they have benefited from the fact that St. Mirren and Dundee have had pretty significant under-performances.
The viz below hopefully highlights how significant each sides over/under performances are.
As we go into the split let’s take a look at the top 5 fixtures this season where the side that should have won comfortably came away with absolutely nothing.
5. Livingston - Expected Points 2.24
Livingston losing late to 10 men Accies back in November is the first of our upsets. On the face of it only getting one shot on target is poor and points to an obvious reason why the result didn’t go The Lion’s way, but it wasn’t due to poor shot selection. In actual fact their shot selection yielded an Expected Shot Accuracy (xShAcc) of 42.3% and an xG of 1.87.
A dominant performance from Livingston could have ended up being a routine win if they had taken their chances but they can also say there were unlucky to concede. Although having 5 of their shots inside the box, Hamilton’s high percentage of shots on target is way above their xShAcc of 38.6%.
4. Rangers - Expected Points 2.60
Rangers under performance compared to their Expected points is down to games like this one. It’s fair to say that Rangers have toiled against Aberdeen this season but their mid-week tie in December probably should have seen them take all 3 points.
Gerrard, who is not afraid to throw his players under the bus after a defeat, was characteristically critical of his side but his comments after the game were undeniably harsh.
Aberdeen benefited from scoring early and shut up shop, the issue is that Aberdeen are not particularly adept at sitting deep and allowing pressure at the best of times and they spent the lion’s share of this match playing with ten men. That is how Rangers still managed to get 9 shots inside the box with an impressive average of 0.22 xG per shot. Their SoT ratio was in line with what you would expect and it’s hard to see where Rangers put a foot wrong. Even after Morelos’ red card they still created 5 chances.
Interestingly Rangers changed their formation for the following game against Dundee and as a result created far fewer chances and only managed to get a point.
3. Dundee - Expected Points 2.60
The opening game of the season, in hindsight, looks to be a double whammy for Dundee. Not only should their performance justify a comfortable win, but their defeat was to the hands of fellow relegation battlers St. Mirren.
Whilst the Buddies looked to have matched Dundee for Shots, including those inside the box, the quality of their chances were much worse. An xG score of 0.71 - 1.68 (+1 Penalty) shows that Dundee were seriously unlucky not to come away with anything from this game.
All that said, I could analyse the game to an inch of its life but if you miss a penalty when the game is tied and then in the 83rd minute your goalkeeper decides to start showboating at the edge of his own six-yard box… then you can’t feel that hard done by.
2. Aberdeen - Expected Points 2.82
Coming only a month after a comfortable 3-0 win at The SuperSeal Stadium, this fixture from the outset should have been a routine win for Aberdeen and yet even 21 attempts on goal couldn’t stop Hamilton taking all three points.
So, chance creation wasn’t an issue for Aberdeen and an xG of 3.57 shows that not only was the quantity there but also so was the quality. Unfortunately, as only 3 shots on target would suggest (although they did hot the woodwork twice), the finishing just wasn’t there. With their xShAcc at 43.6% they could have justified hitting the target around 9 times.
If Aberdeen’s finishing was woeful, Hamilton’s was very fortunate. Oakley scoring from a speculative effort from an impossible angle is the sort of good luck a side fighting relegation needs.
As the Expected points table shows, their actual points and expected points are pretty even indicating that although they have had 2 of the 5 luckiest wins in the league, they have also been consistently unlucky.
1. Rangers - Expected Points 2.91
In his post-match interview, Gerrard claimed that Rangers gifted Kilmarnock 3 points and it’s hard to argue when this game goes down as officially the unluckiest result of the season.
A comfortable performance saw Rangers creating a high rate of good chances, with a reasonable level of accuracy but two quick counter attacks saw them come undone.
Whilst you can’t argue against Rangers players and supporters feeling aggrieved, this sort of performance is bread and butter for Kilmarnock under the stewardship of Steve Clarke.
Sitting deep and soaking up pressure is what they do and it is the main reason why they are where they are in the table. They will consistently overachieve their expected points for this reason. However, there is playing to hit on the counter attack and there is conceding 3.57 xG. On a cold night in January it delivered the results for Killie but there is enough evidence to suggest that if Rangers hadn’t given them so many ‘Gifts’ the tactics might not have looked so clever.