Tuesday, 5 June 2012

Do Interceptions Matter to Teams ?

In terms of excitement and importance to the overall outcome of the contest,rugby of both codes and the NFL have cornered the market in interceptions.An interception is a relatively rare event in both sports,but it can be a game changing one.In a field position based sport such as the NFL the event always costs the transgressor points,both in the potential points lost by their offense and potential points handed to their opponent's offense as a result of the turnover.Similarly in rugby,the interception is relatively rare,but because the majority of the players will be stretched across the pitch close to the incident,it gives the intercepting team an excellent opportunity to at worst make considerable gains in territory and quite often it also leads to an unopposed score.

By contrast an interception in football is much more low key.It doesn't have the rarity value and unless a team can intercept an opponent deep in their own territory,it rarely produces anything dramatic or game changing.Admittedly,it's unlikely that Ebbsfleet's keeper,Preston Edwards would have ever been sent off just 10 seconds into a game,without ever touching the ball had it not been for an intercepted backpass and Stuart Pearce and San Marino would not be inextricably linked in World Cup history,but these types of interceptions are rare and uncommon.In truth most football interceptions pass virtually unnoticed.In a game where changes of possession is commonplace,they are simply another way of losing the ball.Sometimes the interceptor will receive a brief round of applause and sometimes the player getting intercepted will get the same for showing adventure and ambition.

Team and individual player interceptions are now readily available from such sites as EPLIndex,but the usefulness of these newly available statistics is still unclear.As with all counting stats they do an excellent job of recording on field events,but often their value can be obscured because they are often presented without context or connection to a team's success or failure.

How you use numbers such as interceptions is at the heart of the new wave of statistical analysis of football.Unlike set piece sports such as baseball and territory based sports such as American Football and to a lesser degree rugby,football has a much more diverse tactical approach.To take an obvious example,often soccer teams will go backwards to regroup,whereas except in particular endgame situations this is never the case in the NFL.Therefore,how football teams accumulate secondary counting stats such as interceptions can depend greatly on both match situation and their tactical approach.

If we take the example of three teams from 2011/12 who enjoyed varying degrees of success.Manchester City,Champions of course,as a team made 648 interceptions over the season,Stoke who just about finished midtable had 497 and Aston Villa,who lost their manager but held on to their Premiership status recorded 786 interceptions.

Interception Rates for Three Teams from the EPL,2011/12.

Team. Passes Attempted by Opponents. Interceptions
Made by Team.
% of Passes
Manchester City. 14524 648 4.5
Stoke City. 17115 497 2.9
Aston Villa. 17078 786 4.6

There's no obvious correlation between these three raw interception numbers and overall finishing position.So the first step is to try to introduce some type of context.Where the interceptions occurred and what the score was at the time would be a marvelous addition,but it will unfortunately have to wait for much more detailed figures.However,to effect an intervention you need your opponent to pass the ball,so at the very least we can change interceptions to a rate figure.Stoke and Villa's opponents each attempted 17,000 passes over the season against the two Midland teams,while Manchester City's opponents tried just 14,500.So we can start to add more context and more value to these raw counting stats.Both Villa and Manchester City intercepted around 4.5% of their opponents passes,while Stoke intervened on less than 3% of occasions.

The Relatively Rare Sight of a Stoke Player Attempting an Interception.

Next we need to try to deduce the on field distribution for these events.It appears that just under half of a team's interceptions are made by defenders,the vast majority of the rest are made by midfielders and less than 10% are made by strikers.This makes intuitive sense.When defenders are playing the ball out of defence they have more space and they really don't want to give possession away,but when strikers and midfielders are playing passes deep in an opponents half,the downside of an intercepted pass is merely a lost opportunity with little immediate threat to their own goal.Therefore we can speculate that interceptions are predominately made in a defensive context.

We can now look at a game level to see if an increased rate of interceptions,primarily we speculate in a defensive situation can help a team to achieve a favourable result.I've used avoiding defeat as the definition of an acceptable outcome and I've also required that the connection between more efficient interceptions and a good result is unlikely to have appeared by chance.

In the case of Manchester City there is no connection between improving rates of intervention and improving results.Such was City's superiority last season,they could have conceded an extra goal in over half of their games and still won the match,so even if being better at intercepting passes makes more likely to win,it seems likely that in City's case this possibly defensive effect would be overwhelmed by their attacking prowess.

By contrast Villa and Stoke were the two least potent attacks in the EPL,so if a talent to intercept helps a defense then we should be able to see a game by game correlation between efficient interception rates and results.Both teams conceded about as many goals as did teams who finished around 10th in the league and they bettered the goals against totals of half a dozen teams.So their relative success was based around a reasonable defensive showing rather than the dazzling attacking intent of Manchester City.

In contrast to the Champions,Villa show a very strong correlation between interception rates and success on the field.Their chances of taking at least a point from the game increases with increasing interception efficiency.Based on a regression of last years results,they had around a 30% of taking points from the game when they intercepted 3% of their opponents passes,but this figure jumped to a 90% chance of points if they intercepted at a rate of 8%.Of course correlation does not imply causation,teams may be attempting more adventurous passes when they are either stalemated with or trailing to a poor Villa side and Villa may be building up their interception rates after they have gained control of the scoreboard.Viewing Villa's 2011/12 games with an eye to this information should enable anyone to decide if aggressively chasing interceptions was a key factor in Villa achieving their limited success last season or whether it was a mere by product of a favourable game situation.All this may be moot next term with their change of management.

Stoke by contrast had virtually identical defensive statistics as Villa did last term,but very different interception rates.Not only did The Potters intervene at a much lower rate,only 3% of opponents passes were picked off compared to 4.6% for Villa,but there was no correlation between that rate and success on a game by game level.It would seem that interceptions play no part in Stoke's matchday game plan and defensively they are much more about team pressing to force mistakes or speculative shots from distance.The risk reward of going for a pick it would seem is too much for the ultra cautious Tony Pulis.

We've started with a raw stat that many consider of very little value and hopefully we've turned it into a slightly more useful figure that may indicate how some teams set up their defence.Any information is welcome and better than none at all,but potential gains can be made by simple manipulation of the figures.The sample of teams has been necessarily small because data has been require on a game by game basis,but we have been able to speculate that interceptions may have played a fairly significant part in the way Villa's defence performed under Mcleish.It was his attack that underperformed from the previous campaign and ultimately sealed his fate.

There is decent evidence from the NFL that shows that an ability to intercept the ball is a repeatable skill and while the current counting figure for individual players in association football are welcome,they do fall way short of those available on a team level.For example we can barely speculate on the value of large number of interceptions by a single player because we don't even know how many tempting passes have been played towards him.So if teams are being built to pick their opponents pockets,they presumably are using video analysis as a bare minimum to recruit their squad additions.

Interceptions may be lower in the pecking order of more advanced stats than other more universally predictive figures,but there is some reason to suggest that they do illuminate the bigger team picture some of the time.

1 comment:

  1. I like this post.

    Another reason why strikers/attackers do not intercept so much could be because if defenders are put under pressure they will most likely blast the ball over the sideline or (if they dare) pass it back towards the goalkeeper - it is highly probable that the chance of an opponent being behind the line of defence and between them and the goalkeeper is very small (though it would result in a spectacular interception!)
    Anyway ... the defender blasting it over the sideline gives his team a chance to move back and into a defensive position as usually it takes ages for the opposing team to take the throw in. Thus a loss of possession isn't really a bit deal, from a throw in one would imagine it fairly easy to regain possession or at least slow any attacking intent the opposing team may have by making them throw backward etc.
    You don't often see an attacking player kick the ball out of bounds because he is under pressure, you normally see them try to keep a ball in and maybe bounce it off an opposition player out of bounds.

    Two things pop into my head as I write this ...
    1) Teams/coaches should really insist that players take the throw in asap to capatalise on pressure situations that force the other team to play the ball out of bounds ... not wait for the right/left defender to come up and take the throw in.
    This is done a bit, but in my opinion when watching games, it could be done more often.

    2) Might it be interesting for an attacking team to actually deliberately play the ball out of bounds as far forward as possible if a player comes under pressure without any options of getting rid of the ball? A throw in is many times slower than an interception in changing the direction of play. This will give the ball losing team time to set up a pressure situation at a throw in instead of having to chase down the ball carrier.

    There were two interesting interceptions over the last few friendlies that I saw.
    1) Holland - Bulgaria - extra time of second half Heitinga plays a cross field pass (central defence to left back) that Bulgaria intercept (striker I think) and directly results in a goal and a loss.
    2) Holland - Northern Ireland - first half Robben intercepts a central defence pass out wide which directly results in a goal for a 4-0 score.
    Both interceptions were players seeing where a pass was going to go and sprinting to intercept ... both cases there wasn't much pressure and the defender thought they knew what they were doing. :)

    Anyway, again, like the post, got my mind working, good stuff. :)