Welcome to part 11 of the EPL 2011/12 season review. I took a couple of days off to let the England v France game run it's course and to take part in some gambling and predictions for the Euro 2012 tournament (see here).
Today we will be looking at a new stat that has been touched on very briefly by two or three different bloggers in regards to football. But, this is a stat which I saw four years or so ago (being a big hockey - Edmonton Oilers - fan) and has been around for a good few years in hockey, being invented by a talented blogger and statistician named Vic Ferrari in about 2007/8.
It's premise is thus: PDO = scoring % + save %. PDO is the combined total of a teams ability to keep the ball out of their own net and the ability to put the ball in the opponents net.
Shortly I shall run through the EPL PDO table in overall standings and then for home and away. As I said before the jump, PDO is scoring% +save% - but what does it tell us about a team if the number is high or low?
A high number indicates excellency in keeping the ball out of the net (save%) and/or putting the ball in the opponents net (scoring%). A low number can indicate a poor ability to score or to save. Some teams have a high PDO on the back of an excellent scoring% (Stoke) and an average save%. PDO, in essence, is a marriage of these two abilities to give us a wider view of a specific team in terms of looking at defensive and offensive ability.
Is PDO a good measure of a good team though?
Well, the best EPL teams in terms of league position didn't necessarily have the highest PDO number. However, there are other factors to league position than just scoring % and save % (thus PDO). A high volume of shots on target for, for example, can mean a lower PDO but indicate a good team who scored lots and who finished high in the standings. Conversely, a team which conceded few shots on target against and conceded few goals against also can have a lower PDO but still excel in terms of points.
Overall, good teams will score at a higher rate and make saves at a higher rate but this isn't a revelation. There is an element of luck - due to the data sample size - when using PDO for a one off game, but when looking at a 38 game season (still a smaller sample size than is ideal) that luck should even itself out. A team with a high PDO doesn't make it a great team, it just shows us who was efficient in converting the number of shots on target for into goals, and who was efficient in keeping out the shots on target against.
Will PDO regress toward the mean?
I looked at City's scoring%'s and save%'s over the 2011/12 season and the both regressed downwards toward the mean - not to the mean - but only by a couple of percentage points. I have also read blogs that have looked at teams' scoring and save% and they are shown to decrease toward the mean over a seasons length if the %'s (PDO) are high and increase toward the mean if the teams %'s (PDO) are low. But the regression is not by a significant number.
We also have to understand that not all teams have the same quality of players at their respective clubs. United will always have a PDO above 100 and a relegated side will always be in the 90's. No relegated side would have posted a very high PDO number. Why? If the team had posted a high PDO number over 38 games they would amost certainly have not have been relegated.
In the end the quality of a team over 38 games matters to a PDO number despite the 2, 3 or 4% regression to the mean over the seasons length. A PDO number may well bounce back toward the mean the following season, as I expect Liverpool's number to do, but that is the next season with different players and circumstances and can't accurately be assigned to luck returning to, or deserting a team.
2011/12 EPL Table By PDO (overall):
We see Man City and Man Utd head this table. This is unsurprising as they were run away leaders atop the Premier League with both teams having high save and scoring %'s (refer to the links at the top of the article on save and scoring %'s). They were the best and most consistent teams but in United's case their PDO slipped slightly at seasons end as they tired and stumbled in the final six games.
Stoke are third, and it's this number that really maintained their Premier League status considering how poor (bottom four for the most part) in most of the metrics we have looked at so far. They had the leagues best scoring % and a slightly above average save %. Can it continue into next season? Possibly. They are an outshot team who play a unique style that gets them close to the opponents goal for high quality scoring chances. If their scoring % slips they could be in trouble.
Newcastle are fourth.They were stubborn at home and shocked us all by getting into fifth place, but this PDO needs an warning sign. Newcastle posted the fourth best scoring % on the back of firstly, Ba and then Cisse's incredible individual scoring %'s (Cisse's scoring % was nearly double what Robin van Persie's individual scoring % was). Will those hot strikers cool off and thus reduce Newcastle's scoring ability? In all likelihood, yes.
Sunderland were fifth off of the back of the fifth best save %, they were also only a -1 in goal difference when all the other bottom half teams were heavily in the minus column.
Everton are the final team to break even (100) by PDO and after Everton we see Fulham and West Brom a shade below it. The real shocks are Arsenal in seventeenth and Liverpool in nineteeth. Arsenal were fine by scoring % but crippled by the league's second worse save %. That save % didn't cause too much damage due to the few number of shots on goal conceded. The save % must be better next year to gain top four placing.
Liverpool with a PDO of 92.4 were the real disappointments of the league. Their save % was par but their scoring %, despite expensive additions, was the worst in the league by a distance. Could it be as bad again next year?
Despite the introduction where I stated good teams with good players may have good PDO's, we see a few teams who look out of place. The Liverpool's and Arsenal's under performed in just one of the save or scoring %'s, but not both. In Arsenal's case that impact was limited, but the PDO is what it is.
In the less fancied teams' case, they have a high PDO because the individuals played well or the their tactical system worked well in one or both categories.
2011/12 PDO (home)
We don't see a great deal of change for home record by PDO. City maintain first place (as was to be expected), whilst United slip a few places by PDO measure but still post an excellent number.
We can see that the mean line has moved slightly from 100 to about 102 or so as we are looking at home form.
Arsenal improve greatly when we look at home PDO and almost break even with a 98.5 score, but Liverpool still struggle. The three relegated sides are all bottom five in home PDO. Don't the pundits always say the home form is crucial in a relegation fight? All three sides struggled here.
2011/12 PDO (away)
Man Utd absolutely romp away with first place here in PDO terms. They had the leagues best away scoring% and the second best away save %. They were the league's number one away team and had the volume of shots for and lack of shots against to really back this PDO number up.
Stoke and Wolves, two unimpressive sides away from home, are third and fourth by PDO and important words need to be said here about the usefulness of PDO.
As there are relatively few shots on target per game in football, a team who doesn't have many shots on target for per game (away) but lots against per game (away) like Stoke can have a high PDO record but poor form in terms of points. The shot totals for are lower, thus if Stoke had only two shots on target in a game and scored one, the scoring % would be 50%, and that will give a high PDO score when added to the save %, which may be only two goals conceded out of six on target-66% save %. That will give the team a 116 PDO but importantly no points.
Extending this theory to a team that had the third best away record (Arsenal) but the second worst away PDO we can say that the volume of shots on target Arsenal conceded was so low (4.26 per game and 6th best) and thus the goals conceded was so low, despite the poor save %, that it didn't really hurt their away points total. The samples were too small to harm them in a big way. Their PDO may have been poor and the save % poor, but it didn't matter enough to derail all the other good things (high volume of shots on target, possession) that indicated Arsenal were a good side away from home.
PDO is a fine method of seeing how teams performed in terms of converting their shots on goal and preventing the opposition from converting their opportunities and adding them up into one quick and easy number. It's not perfect and there are teams who are special cases due to low volume of data that need further explanation as to why a low PDO doesn't fit with a high points total or vice versa.