I love a fast counterattack. The moment when a soccer game breaks open and suddenly five or six players are all racing toward goal, attackers looking to fill the right lanes and defenders desperately planning a last-ditch tackle that could score a goal. And these are not just exciting moments. Recent research has shown that attacking at speed correlates strongly with odds of scoring.

The faster an attack is, the higher the chance of scoring along a roughly linear scale. But weirdly, at above about eight yards per second, the relationship breaks down.

Attacks that cover ground at a rate faster than eight yards per second are extremely rare. This is because it’s not reasonably possible to run with the ball much faster than that. Take Mame Biram Diouf’s goal against Manchester City last week:

From the clearance to the shot, Stoke and Diouf cover about 80 yards in 10 seconds. It’s a perfect example of the utility of attacking speed. By driving up the heart of the defense, Diouf gets a one-on-one chance against keeper Joe Hart. While Hart perhaps could have done better against the shot, the chance was an excellent one precisely because the speedy counterattack prevented City from putting any defensive pressure on Diouf at the point of the strike.

To go faster than that, the attacking team probably has to use less effective attacking methods. It would take a very long pass, moving much faster than a player is likely to run, to create an attack at 12 or 15 yards per second. That means, in turn, that the attacking player needs to be well ahead of the ball, and thus that he is probably already defended. In my new expected goals methodology, I adjust attacking speed for whether the attack includes a very long pass.

So how do we measure which teams attack at speed? I do not think average attacking speed will be a very useful metric because it will be skewed by those few very fast but not very useful attacks. I think the simplest way to do it is just to collect the total number of fast attacks. As you can see from the above chart, at about five yards per second, the odds of a shot being converted rises to over 10 percent. That seems like a reasonably good cut-off. So which teams last year took the most shots after fast attacks?

Liverpool tops the list, unsurprisingly. Chelsea is notable here for having slightly more fast attacks that did not involve a very long pass. Brendan Rodgers encouraged long-balling, and Liverpool’s 20 shots following a long-ball driven fast attack was the most in the Premier League.

The teams most hurt by their lack of attacking at speed were Sunderland and Swansea. You can see the effects of Swansea’s “Welsh tiki-taka” in the club’s near-total avoidance of long passing, but it also ends up with a plodding attack that produces few breakaway opportunities. Among the top clubs, David Moyes’s Manchester United was obviously a problem. Their slow, cross-focused attack created few good chances, and slowed down the game to little notable end.

Another way of looking at speed of attack is to consider the percentage of shots produced by fast attacks. This moves a few lower-table clubs up the rankings, but also shows that attacking at speed has some drawbacks.

Relegated Fulham with its massive goals-allowed tally tops the list. Among the top of the table clubs, Spurs and Liverpool probably had the weakest defenses. There appears to be a bit of a correlation between the ability to attack at speed and leaving yourself open at the back. It makes sense. To create breakaways, you need runners leaking out to receive those outlet passes and get the break going. Give too many attacking players the privilege of breaking out of defense, and you might find your own defense stretched and struggling.

Attacking at speed, then, is a good way to create chances. We can see in the struggles of teams like Manchester United and Sunderland how a lack of attacking pace hurts. But at the same time, focusing on creating fast attacks has a downside in its knock-on effects on a defense.