It was clear that for some people the absence of Shots on Target (SOT) is a valid reason for staying out (or maybe for getting out ) of a football trade. Now, I'm nothing if not democratic, and each to his own. But, I've some doubts, and therefore, some questions....
I have two main issues: a) definition and b) averages
What is a SOT? Who defines it? How accurate is the assessment in-play? Is there a universally accepted standard? If we say a SOT is a shot (or header???) that either results in a goal or is saved / blocked then I suppose that's OK as a standard. What about a shot that comes off the post or the bar? Or one that is millimetres out? Were they on target, or were they just 'shots'? What about deflections? See my point?
I can live with all of that, but what I don't get is how the averages work. Below is a table showing SOT and goals in the Premier League last season. From this we can see that Manchester United, the Champions, have the lowest SOT:Goal Ratio needing an average of just 3.48 SOT to score a goal. The 20th ranked team, perhaps similarly unsurprisingly, were QPR (relegated) who needed over 8 SOT per goal. We can also probably conclude that Reading are an efficient scoring machine, ranked second only to Utd and needing only 4.16 SOT to score. The fact that they also were relegated suggests they just need to up their shooting rate! Is Premier League soccer that simple, I wonder?
Be all that as it may, we can see that on average a Premier League side needs 5 SOT to score a goal. So in theory a game with 10 SOT should, on average, produce 2 goals.
I've said previously that maths isn't my strong point. Statistics are even more of a Dark Art to me. But, and this is the crux of what I don't get, how do those who rely on in-play stats allow for previous and forthcoming games and goals? If that sounds double dutch I'll try to explain what I mean.
August 17th 2013 marks the start of the new season for the Premier League. Team A need 5 SOT per goal and they are playing Team B who need 6.5 SOT per goal. Let's assume the game results in a 0-0 bore draw, Team A having had 5 SOT and Team B 7.
When those two teams kick off in their second match, do we count SOT from the first shrill of the man in black's whistle, or from the 5 and 7 figures above? By the same token, should we have taken into account the SOT since the respective last goals scored by both teams last season in our in-play stats for the 17/8/13 game?
Even a mathematical dunce like me can accept there's a causal relationship between SOT and goals. In a golfing situation, if my putt doesn't reach the hole, it ain't going down! But until someone can explain to me (in a way that I can understand) how these stats can be usefully employed in-play I'm going to carry on not worrying about them!
P.S. The SOT / Goal data comes courtesy of www.football-data.co.uk . Their statistical information is drawn from a number of sources - a site well worth a visit if, like me, you're stats mad ;)
Good post Dave.
ReplyDeleteIf you're advising people against an over-reliance on SOT in-play stats, then I have to agree with you. They should be just one tool but certainly not the sole tool.
But surely those who look at Bet365 (or other) in-play SOT data are just using them to get "a feel" for how the match is panning-out and nothing more. Most people use them as a single snapshot.
There will always be the odd case where there are a truckload of on target shots, but the game ends goalless. Generally, however, the SOT stats should provide a reasonable indication of how the game is going - and it seems eminently reasonable to assume that good SOT in-play figures suggest an open game.
Straight pre-match betting is another thing altogether. It's there that, in my opinion, shots on/off target come into their own. They are a great help when building a formula for straight betting, as they can act as that final ingredient to really enhance an algorithm.
Thanks, Eddie. I wouldn't presume to advise one way or another but think that as a sole arbiter of staying out, getting out, or getting in in-play stats are not, on their own, a strong enough indicator.
ReplyDeleteAny in-play trades I enter are as a result of pre-match assessment - which is sometimes right, sometimes wrong and most of the time somewhere in between brilliant and useless! But, once I decide to enter a match based on that assessment I at least like to give the trade a chance to work!