Stats are great for football analysis. Naturally, you shouldn't use them in place of actually watching a game, but a look through the numbers can put key points into context. For example, if Reading won a game because they dominated the ball, how many passes did they make? Conversely, if they lost it because they couldn't create enough chances, shots/shots on target/expected goals are great for illustrating that point.
I was planning on doing this for Saturday's 2-0 win over Nottingham Forest. Reading have played really well and won a game! I thought, so, what interesting stats are there about the match? Annoyingly, on the face of it at least, there aren't any stand-out statistics that put into context just how comfortable Reading's win was. Ask any Royal fan who saw the game, and they'd likely agree that it was a very strong performance that more than earned three points.
However, a look at WhoScored's review of the game reveals that Reading: had 13 shots (5 on target) to Forest's 9 (3 on target), saw 56% of possession, edged the away side for interceptions (11-8) and clearances (12-11), made the same total dribbles (13), and were beaten for fouls (14-16) and aerial duels (12-13).
All in all that's not too bad, particularly Reading's clear advantage in possession and shots. But those stats aren't consistent with the reality of how the game went - Jose Gomes' side managing the game well and thoroughly deserving the win.
What's more, some things can't really be picked up by pure data. Reading looked far more energetic and determined than Forest, who themselves seemed frustrated and struggled to properly get into the game. Those are obviously big factors in deciding a result, but how would you illustrate them with numbers? After all, Reading attempted fewer tackles than Forest (21-23), with more of the away side's challenges being successful than ours (12-17).
I'd also argue that we looked a much bigger goalscoring threat than the visitors, particularly in the second half when Reading seemed to find space more easily and get into more dangerous areas. Modou Barrow, John Swift, Lewis Baker and Leandro Bacuna all found themselves in positions where they should have at least worked the 'keeper, even if they would all fluff their lines.
Again though, this advantage isn't really picked up in the stats. Yes, Reading had more shots than Forest, but 13-9 isn't as big a gap as I'd thought. A better way of assessing threat in front of goal is to look at 'expected goals' (xG) which, rather than simply counting shots, grades shots according to distance, angle and so on. Therefore a tap-in from six yards carries much more value than a hopeful punt from 30 yards out.
According to the website Experimental 3-6-1, Reading only created chances worth one goal, whereas Forest were expected to score 0.6 of a goal, going by the opportunities they made.
You can see on that graphic how Reading's xG creeps up after each shot on goal; Swift's first-half opener taking the total up quite a bit due to that being a better chance than, say, one of the long-range efforts from Ovie Ejaria or Modou Barrow.
On the whole xG is a better metric than, say, the total number of shots on target, but there are other points we need to bear in mind. Firstly, the own goal has no value on the graph. I'd argue though that a low, hard cross into the area can be an effective way of forcing a goal, even if you need a bit of luck for the ball to actually go over the line.
Similarly, there's no accurate way of plotting goalscoring opportunities that didn't end in a shot. If a Reading player goes through on goal but is tackled before they pull the trigger, is that chance worth as much as an actual shot? None of this is to knock the xG method, which does give us a great insight into how matches have panned out - it's simply to say that no stats should be taken at face value.
So, what role should stats in match analysis? Of course, they shouldn't replace the good old-fashioned value of watching a game with your own eyes in real time, but that's not to say they should be taken out of the equation (pun unintended for once) completely. Rather, we should not only use both sides (stats and viewing) to analyse matches, but also use them together where possible.