On Newcastle during Pardew’s touchline ban

I know Newcastle have had poor results recently so I thought I’d take a cursory look at their underlying numbers during the seven game span. What I found is pretty staggering. Here’s a split of the season:

Screen Shot 2014-04-27 at 12.23.30 AM

The metric there with the largest skill component is TSR (see here for more information), and that hasn’t really changed much.

What has changed? Well the more luck based metrics, shooting and save percentage (goals/shots on target), which have been poor all season, have simply fallen off a cliff in the past 7 games. A PDO (Sh% + Sv%) of 580 is astoundingly low – Newcastle just haven’t had a break go their way at all over the period.

To give a sense of how low this number is I took the worst Premiership team I have on record – Derby in ’07-08 – and looked at their PDO over the 32 7 game spans they had during that wretched season (i.e., games 1-7 are span 1, games 2-8 are span 2, etc.). Their lowest PDO over a 7 game span? 684, which would have been worth about 3 goals to Newcastle over the past 7 games.

I’m not quite in a position to quickly pull the 7 game PDOs for all of the teams I have on record, but I’d be amazed if Newcastle’s wasn’t solidly in the lowest 1%. That’s something to bare in mind if Newcastle get a result against Arsenal today and the commentators attribute it to the presence of Pardew.


3 thoughts on “On Newcastle during Pardew’s touchline ban

  1. Which kind of conversion rate do you use for the PDO in your rating?

    The thing is that I have created and analysed my own metric consisting of sh%, sot% and number of shots. (To be exact: I found the formula elsewhere, but I did research on my own and tried to upgrade it).

    As far as I was able to put the formula to test,If found that my rating isn’t much worse than yours (I don’t find SoT conceded numbers, which are fundamental for the index)

    One question regarding upgrades was: What kind of conversion rate works better, Goals/Shot or G/SoT?
    Although G/SoT itself has worse statistical characteristics than G/S, it seems to improve the qualities of my index. Perhaps this could be an interesting note regarding your rating and possibilites for improving it (if you do not already you use G/SoT).

    PS: It would be interesting to read something general about the new rating: Not necessarily about the entire formula, but about statistical questions such as “How quickly does it regress to the mean?”. As you did it with the other metrics. A concentrated reader might have already been able to learn few things from the previous articles, but the infromtation isn’t concrete yet.

    • For PDO I use shots on target. I can’t think of a solid argument for using total shots in it’s calculation.

      Re how fast it regresses towards the mean – I haven’t done the backward work required to know exactly, but given that it’s more repeatable season over season than TSR I’d guess it’s more repeatable within a season too, so I’d expect it to regress more quickly than that (which can be found here https://jameswgrayson.wordpress.com/2012/11/27/how-quickly-do-the-advanced-stats-regress-to-their-final-values/). When I do the work I may write it up.

      • Thanks for the answer.

        Yes, the first pieces of information you put into the articles looks very promising. Definitely promising enough that the rating can be used although the extensive examination isn’t complete yet.

        By the way, I like your approach because it works without any data that are hard to gather (as passes, duels or shooting positions). I often think about the fact that your simple metrics could easily be employed to amateur leagues. Shots and SoT are infrequent enough (there’s approximately 1 shot every 3-4 minutes) and easy enough to define. Every private person who is interested could collect shot data on his own. It just needs (not much) effort, a notebook or only pen and paper and it isn’t difficult. The maths behind is not very complicated, too.

        In my opinion it would also be an interesting approach to use the rating for measuring a player’s offensive skills (with e.g. Shots per 90 instead of TSR). How many shots did a player take and how well did he convert them? Certainly, it would also make sense to employ the rating to key passes.

        This would be not even a simple & easy way of gathering information about players’ or teams’ skills, but this information is also very useful and of high quality, as you’ve proved often enough. This is what economics call “effective”.

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s