The Castelen conundrum

Next up in having a look at some transfer related players and numbers for this A-League off-season is former Western Sydney Wanderers winger Romeo Castelen. A hat trick in the final win over Brisbane Roar wasn’t enough to win a new contract from Popovic, but with a list of clubs lining up for his signature lets have a look.

We all know he is a great dribbler of the ball and it’s little surprise he has completed the most dribbles per90 in both of the last 2 A-League seasons. But to put it into a bit more context of how he compares to his peers:

Screen Shot 2016-05-12 at 10.57.18 PM.png

He is really, really good. His dribble success rate of 50% and 63% are above average and his 3.71 and 4.11 dribbles per90 blow just about everyone else out of the water. If you want a direct winger who will get behind his man to create chances, then he is going to be a good provider.

In a poor attacking side in his first season at Pirtek his job was more provider than finisher. He provided 2.08 chances per90, of which 0.15 were assists. In terms of shots, he took 1.63 per90.


Moving onto the most recent season, his creativity remained similar – 2.18 chances with 0.24 of them assists. But his shot output increased enormously – to 4.02 shots per90 with 7 goals at a rate of one every three matches.

While he was guilty of being wasteful in front of goal – no one with 5 or more goals (min 810 mins) took more shots per goal than the Dutchman – his combined shots and shots created (6.2 per90) ranks 5th among individual seasons over the last four years in the A-League behind only Aaron Mooy (2016), both of Del Piero’s seasons and Carlos Hernandez (2014).


open play creativity.png

There are things to be wary of though – his age being the most obvious and perhaps the reason he did not remain at the Wanderers. He turned 33 two days after the Grand Final and while his performances this season were better than last season there would be concerns of a big contract if it were for two years.

The second question would be whether he would fit the team style and structure. With Adelaide United losing Goodwin, Sanchez and Kamau he could be a good fit on the right for the Reds with the ability to hug the touchline and beat a man but also capable of coming inside and being a threat himself. While his ability on the counter could be useful to say Perth Glory who will have good finishers in Andy Keogh and Adam Taggart – reducing the need for his speculative efforts on goal.

All stats and data via Opta.


Nick Cowburn and the numbers

Interesting read in the Newcastle Herald last week with Jets coach Scott Miller pumping out some numbers to back-up the re-signing of Nick Cowburn on a 2-year deal. Pushing aside the fact that Miller has a questionable relationship with numbers, it is good to see that he is at least looking at some kind of data in making these decisions. But what data is it and is it useful?

Miller said: “I looked at his stats against [Scott] Jamieson and Scott Neville, for example, and he’s on the money with them in terms of key passes, tackles won, duels won.” The Herald then listed the stats for Cowburn and Jamieson on things such as tackle won percentage.

Few things to keep in mind here – Opta tackle definitions are a bit misleading. A tackle won means successfully winning the ball and possession being transferred to your team or the ball going out of play. A ‘tackle lost’ is still successfully winning the ball but instead the ball goes to another player from the opposition. All tackles are really a success. What is really a failed tackle is a ‘challenge lost’ where a player attempts a tackle but is dribbled around. So with this in mind, how does Cowburn stack up?

I have filtered the stats for matches where the players started at fullback only & tackle success is total tackles / (total tackles + challenge lost). I’ve done this up quickly and this is the most efficient way to do it – ideally you include all minutes at fullback but a) time and b) sub-effects is a thing to consider too.

Stats per 90 mins

So Cowburn makes more tackles, but he is also dribbled past nearly twice as often and in percentage terms is worse too. The least important number here for me is the first one. Making more tackles could mean lots of things  – he wasn’t marking tight enough, he was left exposed by the winger in front of him, the team didn’t press so there was more defensive output by the back four, they had less possession so were in the defensive phase more often. All of these would lead to a different evaluation of the player.

Strangely, tackles and intercepts is probably more important when looking at forwards (are they hard working, could they fit into a pressing structure) or midfielders (what style are they – defensive, box-to-box, playmaker etc).

So while he may not compare perfectly against Jamieson and Neville, it’s firstly not a particular wide scope for context and possibly not a fair comparison either. How does he compare across the whole league?

stats for matches started at fullback in season 15/16 (min. 500 minutes). Size of circle is total tackles per90

So from a defensive side – not quite in the more elite grouping but for a fullback who only turned 21 in March, those are some pretty good numbers.

It’s also not hard to see why at the Roar Stefanutto was dropped in favour of Brown.

The ‘key pass’ one is the most interesting though. In the matches where he started as a fullback he didn’t create a single chance – neither an assist or a key pass (shot assist that doesn’t result in a goal).


attacking stats for matches started at fullback in season 15/16 (min. 500 minutes). Size of circle is assists per90

From an attacking perspective Cowburn was the poorest fullback in the league. Fewest completed passes in the final third and not a single chance created.

I’m not saying Cowburn isn’t a good (re)signing for the Jets and it’s good to see a coach using numbers but it’s important to remember that with defenders there are few more things to consider – more isn’t always better like it is with strikers and goals. You wouldn’t want your centre-back making 10 tackles a match for instance. Cowburn’s defensive numbers stack up pretty well for a young player, but there is definite room for improvement going forward.

All stats and data via Opta.


PDO, regression and Newcastle v Wanderers

Before Newcastle beat the Wanderers 5-0 and render this post entirely useless, here is a quick look at PDO and why I’m keen on referencing it. It is by no means a perfect metric that can explain everything. But it does have its uses.

PDO is the sum of a side’s Score% and Save%. Each calculated as:

  • Score%: Goals For / Shots on Target For
  • Save%: 1-(Goals Against / Shots on Target Against)

Over the course of a season, most side’s convert their accurate shots into goals at a similar rate. And likewise, they save them at a reasonably similar rate at the other end. There are outliers of course – Melbourne Victory had a Score% of over 44% last season – more than 6% better than the next best effort over the last three years. The majority of sides will have a Score% between 25% and 35%.

From a  defensive perspective, only 2 out of 30 sides in the last three years hasn’t had a save% between 63% and 75%.

So why does this matter? There is a luck in football. Because games are decided on such small margins, it is possible for a team to perform in the short-term, and win games, at a level that is not sustainable over the rest of the season. At the start of the season, a side may convert a high percentage of their chances in front of goal, but over the course of the season their conversions will regress to the mean. Converting 50% of your chances for the first month, does not mean you will for next 23 rounds.

This weekend’s A-League clash is a perfect example of team’s with contrasting fortunes so far. We are four weeks in to the season and Newcastle find themselves in equal second position on the table. Scott Miller deserves credit, he has taken a team that won only three matches last season to as many victories in four weeks this time around. They look more defensively secure and the players are on the same page as each other.

But. And it is a big but. They’ve saved 83.33% of the shots on target they’ve faced. Only the Wanderers’ remarkable debut year comes close (Save% of 80.5%). It’s unlikely they can sustain this over the season. They could turn out to be one of the most defensively sound teams we’ve seen and they certainly restrict side’s from taking good chances. But to keep saving at 83%? I’m not convinced.

Not only have the Jets rode their luck in a defensive sense, they have also managed to score from 60% of their shots on target.

You can point to the resolute, compact defence as being hard to break down, or the fact the chances they will create are on the counter and often at an exposed opposition keeper. But to put it in perspective.. The last three year’s in the A-League look like this:

Screen Shot 2015-11-07 at 12.46.13 AM

At the other end of the spectrum though is Western Sydney. They so far have a score% of just 16.67% and a Save% of 28.57%. Things can only get better from here. They are ranked #1 in the competition for Total Shot Ratio – taking 68.54% of the shots in their matches and #2 for Shots on Target Ratio and Shots Inside the Box Ratio. They are dominating shots, but with little luck in either box.

Newcastle are the 2nd worst side in the league in all three of those metrics.

Score% and Save% takes a few weeks to settle and regress and this Saturday’s clash could be a starting point. See this graph here (press play bottom left corner) to see how each team’s PDO has evolved throughout the season over the last three years.

To see an interactive version of the above image click here.

Leopold Method – Melbourne Derby: Georgievski dominates down the right

Leopold Method – Melbourne Derby: Georgievski dominates down the right

In what’s become a familiar pattern, Melbourne Victory were victorious against their Melbourne City counterparts in front of 40,000 at Etihad Stadium on Saturday night, recording their fourth win in five meetings since the beginning of last season. Victory were the stronger team for the majority of the match – the goals shared across the front three with Fahid Ben Khalfallah, Kosta Barbarouses and Besart Berisha all finding the net in the 3-2 win.

The match was defined by Victory’s dominance down the flanks in the first half, particularly on the right. But it wasn’t all one-way traffic, with Melbourne City surging in the second half off the back of a better defensive shape and increased support for new striker Bruno Fornaroli in the attacking third.

Continue reading…

Leopold Method – Wanderers rue defensive errors

Leopold Method – Wanderers rue defensive errors

The 2015/16 A-League season kicked off on Thursday night with Brisbane Roar winning 3-1 against Western Sydney Wanderers at Pirtek Stadium. Less than eighteen months ago these two sides were playing against each other in a grand final, but after both experiencing disappointing domestic campaigns last season, each side were looking forward to a fresh start – Wanderers with a predominantly new playing squad and Roar with a new coach in John Aloisi.

It was the home side that started the strongest, and despite being down 3-1 at the break, were arguably the better side for the majority of the first half. Out of possession they employed a medium block that slowed Roar’s defence playing the ball to the feet of midfielders. With the ball, Nichols was the key player, finding space between the lines to receive the ball and pushing beyond the Roar’s back line with dangerous runs in behind.

Continue reading…