There's been quite a lot written over the last few weeks (mostly in the ice hockey community, but also in the football one) about what PDO is and how it reflects team performance.
James Mirtle of the Globe and Mail came up with perhaps the best explaination of how PDO should be applied to general statistical analytics - noting it "highlights extreme outliers doing things that won't continue".
I felt this summed-up rather well what Jurgen Klopp and Borussia Dortmund are going through at the moment.
|Position||Team||Played||Won||Drawn||Lost||Goals For||Goals Against||Goal Difference||Points|
Football is plagued by small sample sizes compared to ice hockey and so PDO can have a much more noticeable effect and seemingly last longer into the season.
From an anylytics standpoint we should not necessarily be considering the number of games played (although that is relevant) but more the number of shots taken - this, of course, is what we are sampling here.
In ice hockey it's typical to see each team have around 30 shots (sometimes more) on target in each game. In football, the total number of shots can be just single figures (start watching Cardiff games, you'll get used to it) with barely a handful of shots on target.
So you can see how hard it is to build up a substantial sample size.
For example, Bournemouth, top in shots on target with 106, have as many shots on target in 19 games as most NHL teams do in four games.
The NHL season is now more than 20 games old and PDO effects are still being raised as a concern.
|Position||Team||Shots for total||Shots against total||Corsi/TSR||shots on target for total||shots on target against total||Shots on target share||Shooting % For||Save %||PDO|
As a result, teams enjoying a PDO boost, whether that be from a short term scoring hot streak or run of clean sheets, can see them climb the table rapidly and spend a decent amount of time at or near the summit.
The chart shows that as a whole after 19 weeks of play, the correlation between PDO and points is still quite strong at 0.6123. (1 is a complete positive correlation, -1 a full negative correlation)
Meanwhile, the shots on target (a fairly simplistic but suitable approximation for team skill) to points correlation is just 0.273. The overall shot share to points correlation is even lower (0.0684).
Curiously, the east midlands seems to have been one of the areas benefitting most from PDO this season.
Earlier in the season it was Nottingham Forest who roared to the summit and then came crashing down, now it is Derby County who look down upon the rest of the league.
Norwich, Bournemouth and Middlesbrough have proved the contrary, building their positions at the top on solid shots totals combined with a touch of positive PDO - and Norwich's descent has been more or less totally down to a collapse in PDO. The Canaries still have the best overall shot share (62.4% of all shots) and the second best shots on target rate (60.2%).
But looking down the table we can see this aforemention correlation being borne out.
Near the top Ipswich and Charlton are current darlings of PDO - and it did amuse me how their game last week played out. The injury time winner for Ipswich served to dent Charlton's then second highest PDO and give the Suffolk side the chance to userp them in that regard.
Ipswich's underlying numbers, rather like Watford, are a contrast. Comfortably below 50% for overall shot share, but comfortably over 50% for shots on target share.
If we go with shots on target being the better predictor of skill and future success, then it is possible to see both those clubs make a sustained challenge for promotion, as and when their PDO scores drop.
For Charlton it should be a much more worrying time.
The Addicks are rock bottom (yes, even below Blackpool) in total shot share and fourth bottom in shots on target share.
Should their PDO nose dive in the style Norwich's has, then it could get very messy very quickly for Bob Peeters' team.
Leeds are in an equally perilous predicament - although it may not appear so at the moment. We've seen earlier in the season that Massimo Cellino is prone to PDO reactions and although he may have bigger problems on his mind at the moment, a slump in the above average shooting % or a dip below average in save % could see Leeds drop rather quickly, especially with the team just five points above safety. Neil Redfearn has failed to address, so it seems, the key issues still facing this Leeds side in his second spell. Let's see how it plays out this time around.
Blackpool remain the biggest victims of PDO in the table, with Millwall the only other team still below 90.
As the season progresses we should see the range of PDO scores continue to tighten around 100 - that prospect alone means Blackpool and Millwall fans have something to look forward to.