Dropping Jaaskelainen

A few weeks ago Jussi Jaaskelainen was dropped by Sam Allardyce after keeping 9 clean sheets in 16 Premier League games. He was top of our shot stopper rankings for the season too. It seemed a bizarre decision at the time and the Finn was quickly re-instated after a couple of Premier League games in which his stand in, Adrian, conceded 6 goals. Jaaskelainen came back in against West Bromwich Albion only to concede 3 goals. He was immediately dropped again.

It’s not the first time Jaaskelainen has suffered this ignomy in recent years. Half way through Bolton’s 2011/12 relegation season he was dropped for Adam Bogdan, a youngster who had conceded 5 on his debut and then been beaten by Everton keeper Tim Howard in his second game. Jaaskelainen still didn’t get his place back.

Using the positions and outcomes of thousands of shots recorded in our database, we can give each one an expected goal value based on what’s happened before. This allows us to compare how many goals a keeper has conceded to how many the average Premier League keeper ‘would’ have conceded based on the same type of shots faced. Here’s Jaaskelainen’s plot since 2010/11:

JJRollingWe can see that Jaaskelainen was concededing about the ‘right’ number of goals for the kind of shots he was facing all the way through the 2010/11 season. But then right from the start of 2011/12 he starts to concede more goals than expected until he eventually gets dropped for Bogdan. Looking at this chart the decision to bin the Finn looks correct. In real terms , the model suggests he had conceded 7 more goals than he should have done in 18 games.

The below graphic shows the league average save % from different pitch zones and Jaaskelainen’s save %s at the point he got dropped in 2011/12:

JJ Avg ComparisonWe can see that his worst performance comes against shots from Zone 1. And this is the Zone we’ll concentrate on for this post. In real terms, the difference between Jaaskelainen’s 28% here and the league average 44% is four goals. We’ve discussed in previous posts that the fact that the average save % here is 44, generally means a keeper has less chance of saving one of these shots on target than he’s got of winning a heads/tails coin toss. The model therefore suggests that Jaaskelainen may simply have been unlucky to concede these goals. We wanted to test the model verses ‘real life’ so had a look at the footage of the goals he conceded from Zone 1 (x the adverts when it comes on):

Everyone’s going to have their own opinions on what could/should have been done to prevent the goals here. In our opinion the only goal here that looks vaguely poor from Jaaskelainen in terms of stopping the actual shot is the Gareth Bale goal. He tries to get down with his hands to a ball very close to his feet.

Given the numbers and the footage, we have to seriously question whether what we’re measuring here is actually goalkeeper skill in stopping these shots. We can plot Jaaskelainen’s rolling save % from Zone 1 from the start of our database:

JJ Zone 1 RollingWhat we’re seeing here is that it took almost a season for save % from here to stabilise for Jasskelainen. This pattern of stabilisation is much the same for every other keeper we’ve studied.

We looked at the MCFC analytics data release from Opta. Opta classify what they think is a player error leading to a goal. They only had Jaaskelainen down for one in his 18 game run and it wasn’t from Zone 1. They also classify situations where a player should reasonably be expected to score – usually in a one-on-one scenario or from very close range. They call these ‘big’ or ‘clear cut’ chances. Jaaskelainen faced more of these than anyone else bar Ali Al-Habsi at Wigan. In short, either his defence wasn’t really protecting him too well, or he himself wasn’t commanding his area enough. From the footage it’s difficult to definitively say one way or another – especially as we don’t know what the manager’s instructions are.

In short, it looks like Owen Coyle made a huge gamble in throwing rookie Adam Bogdan into a relegation fight. Bolton still got relegated and Bogdan’s end of season numbers were nothing to shout home about either with a below average save % in three of our four Zones, including Zone 1.

We said earlier that all keepers had similar rolling save % patterns from Zone 1. We lied. There is one exception – Swansea’s Michel Vorm:

MV Zone 1 RollingVorm looks so different from other Premier League keepers here, he’s almost from a different planet. Other information in our Chance Creation Model (based on where shots are hit from and how the ball was delivered to the shooter) suggests Swansea don’t appear to be doing anything particularly special defensively to help Vorm out. The graph above tells us that Vorm is continuously ahead on the heads versus tails coin toss of saving these shots.

Here’s another look at all the main Premier League keepers from Zone 1 over the last 3 and a half seasons:

Zone 1 AllVorm is way out on his own here. He’s not statistically significant – he’d have to be outside the 2nd curved line (2 standard deviations from the mean) for that. However, looking at the last two graphics, we’re not confident in saying that he’s showing no skill beyond that of his peers. Can anyone be that lucky? We guess we’ll find out over the next season or two. In any case, next time Swansea are on telly, look out for this.

Follow us here on Twitter.

This entry was posted in Sports and tagged , , , , . Bookmark the permalink.

4 Responses to Dropping Jaaskelainen

  1. Alan says:

    I found this to be very interesting. While I understand there is data to determine the chance a goal is scored from each zone how do you take into account defensive pressure? As an example, I would imagine that the chances of scoring a goal from Zone 1 without defensive pressure is greater than the chances of a team scoring a goal in Zone 1 with heavy defensive pressure as the shot is taken. I apologize if this has been discussed previously but curious to know your answer.

  2. MR says:

    What is Tremmels Zone 1-percentage? I guess a swansea-factor here.

  3. Its a good question. Unfortunately the data for where opponents are in relation to shooter is not publicly available. *some* of this will be accounted for by the fact the avg is taken from 1000s of diffierent situations. My experience in other areas is that over a season, adding lots of other bells and whistles is rarely significant unless the odd team is extreme in its style

  4. 25% after 23 games….

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s