THE BOOK--Playing The Percentages In Baseball

I don’t watch much soccer but everytime I do I see the following sequence several times a game. The team with the ball will advance the ball up the field making pinpoint pass after pinpoint pass. Once they get close enough to take a reasonable shot they will miss the goal by about 20 to 30 feet. I’m sure there’s some reason this happens like there are defenders in the way or they gotta kick it a lot harder than when they’re passing, but it does look like they choke a little.

The best analogy I can give, for this blog, is that taking a shot is a lot like pitching and passing is a lot like making a throw across the diamond.

The pass needs to get to a general location and the receiver creates the illusion that the pass was right on the money by the way they trap the pass.

When these players take a shot, a lot of different things can be done (and go wrong) with the ball depending on the way it is struck.  They are able to generate many more times movement than a baseball and have it travel upwards of 60mph.  Curve left, curve right, dive down, swoop up, and knuckle (unsurprisingly the worst to stop).  Just like with pitching (or batting) being off by a fraction of an inch can have a dramatically different outcome than what was intended.

If you read the linked article it is not dealing with penalty shots in general (i.e. during the game) but specifically only in penalty shootouts.

Also the 60% is not just when behind but specifically when behind and taking a shot that if missed leads to instant elimination. Likewise the 90% is for shots when tied and a goal leads to instant victory.

In any case I would not be suprised if penalty kicks in soccer has a higher than normal amount of choke than most other sports trials. There simply is not enough physical skill (diffculty) involved to explain how often penalty kicks are missed by profesionals even in the “normal” cases.

My guess (cross-posted at Phil’s blog):

The tied teams have to be kicking last, which means they are tied despite one fewer attempt. Let’s say the scores look like this:
4-4 (35% frequency)
3-3 (50%)
2-2 (15%)
That means the second team is already shooting about 80%.

On the other side, a team down by 1 kicking second in the 5th round is probably just slightly below average (average = 70%).  However, the -1 teams kicking 1st in the 5th round will be weaker, probably below 60%, having had equal opportunities. So an average of 60% doesn’t require a very large choke effect (if any).

Penalty shot is a game between shooter and keeper where shooter must prevent keeper from guessing correctly. In high anxiety situations, shooter may give more information to keepers than normal.
Take, for example, England. I don’t know when their national team last won penalty shootout, but they must have lost six or so of them over the last two decades or so. I still remember the look on Steven Gerrard’s face when he had to shoot his penalty shot in 2006 against Portugal. Just looking at him, I had a feeling he would have 50-50 chance if goalie was removed from the net. He looked as if he was certain there was no way he would score.
Penalty shootout comes after 120 minutes of (presumably) hard play. Players are tired. It is hard to keep emotions in check. Players from nations and teams that have history of penalty losses have additional burden, that becomes especially hard when they are behind.
All in all, just thinking it through, 60% is not too little.
Has anyone calculated the penalty shootout success based on which team shoots first? Based purely on memory of games I watched, I feel that team that shoots first has better chance of winning it, all else being equal. I would like to see if somebody has analyzed it.

For international competition matches the data on penalty kick success by round is as follows:

The percentage success rate in the World Cup is 71.2% compared to 82.7% (CA) and 84.6% (EC), possibly reflecting the greater importance and consequent pressure of the world stage.

The success rate of each penalty kick changes throughout the competition:

First kick 86.6%
Second kick 81.7%
Third kick 79.3%
Fourth kick 72.5%
Fifth kick 80%
‘Sudden death’ kicks 64.3%

That’s purely driven by quality of penalty kick taker. Managers tend to put there best guy either at the start - to get momentum - or at the end, to covert the clutch kick.

John: 2 questions.

1. How many kicks are we talking about?
2. In the sudden death kicks, can they reuse the same players (if they want), or does it go to 6th and successive players?

Turns out the study only had about 400 total kicks.  So maybe 80 5th kicks.  So each “sample”—tied and goal = win, down 1 and miss = loss—must have had only 20-30 kicks in it.  We can’t tell anything from that.

Tango/Guy

Penalty shoot outs are reasonably rare - there are very few tournaments where they are used and even then there will only be a handful of matches where they are required. Hence any sample is meaningless as it is very hard to control for various factors. Non shootout penalties are completely different from a choke/clutch perspective (at least most of the time).

Here’s a link of penalty shoot outs by country: http://www.penaltyshootouts.co.uk/countries.html

You can see that the top countries max out at 11 shoot-outs - that’s 55 kicks (max).  It’s interesting to note the amount of variance between the teams. England has a success rate of 17%. Germany is 70%.

If England play Germany in the world cup the over under will probably be around 60% (Germany win). Even though, in theory, it should be 50%. Roughly speaking the players should have equal talent - (talent pool is roughly similar - pop size is similar, football is #1 sport).

On that specific paper I believe the n is 409, which is basically every penalty shootout in those tournaments.

In sudden death you cannot reuse the same players until you get through all 11 (or 10) if you get a man sent off ... hence why the % goes down.

John:  I read somewhere that the team kicking first wins 60% of the shootouts.  Do you know if that’s right?  And if so, do you think shooting first really confers an advantage.

Hypothesis: if you linked up closing betting lines to the shootout data you’d find that the better teams going in do have an edge in the shootout.

Weird Nats dream scenario: we make it through group stage. Tim Howard can’t go; gets hurt chest again. Marcus starts. We go through most of extra time tied. Bradley develops a set of balls and subs out Marcus for Guzan before the shootout and the Nats win because Brad is a beast on PKs.

Even though he has more hair than a Nats keeper should.

400 kicks, or 80 kicks per sequence number?

Let me give you a little trick.  Take the number of trials, and take the square root.  (In this case, about 9.) Take the recipricol (in this case, .111). 

That’s your range at 2 SD.

So, if the mean is 80%, then +/- 2 SD, it’s 70% - 90%.  The data from John for the first 5 kicks fit right in.

***

Seeing that’s 6th and later kicks is guaranteed to be from not the top 5 kicker on the team, then 65% sounds reasonable.

If the test from those researchers INCLUDED those kicks, then it’s going to be heavily biased.

I will be both disappointed (and unsurprised) if those researchers included those players.

When watching the World Cup, my #1 thought is “Wow, there is a TON of variance in soccer.”