Fixture strength

We have reached one of those points in the season when we face a fixture swing. A fixture swing is where some teams that have had a poor run of fixtures (like Chelsea and Newcastle United) can look forward to a better run, while other teams (like Bournemouth and Aston Villa) will face a tougher test of their abilities.

Fixture swings encourage Fantasy Premier League managers to shift out players from teams that have had their good run of fixtures and buy players from the teams that are entering a good run.

Judgments about the attractiveness of upcoming fixtures often contain an element of subjectivity based on a managers’ perception of the relative strengths and weaknesses of the teams involved. Some managers though will strive for a degree of objectivity by creating tickers that calculate fixture strength by drawing on factors such as wins, draws, goals scored and goals conceded.

Over the last few weeks I have been noticing a disconnect between some of the underlying data I have been looking at and the strength of teams as indicated by respected tickers, such as the useful one on Fantasy Football Scout (this is not a criticism of  tickers, just an observation).  For example, Chelsea are considered a tough away fixture, but the Blues have conceded 22 Shots on Target at home, the joint fourth highest of any team in the Premier League. Meanwhile Bournemouth have conceded nine Shots on Target at home this year, the league’s lowest figure, but an away match against the Cherries is still regarded as a good fixture.

I have spent part of the international break looking at team Shots on Target (SoT) data. As I was gathering the data, I thought it would be interesting to create offence and defence tickers based on this statistic alone to tease out some of these disconnects I was spotting. I give a major health warning here that I am not a statistician and what I did with the data might not be what a trained statistician would do with it.

I have created a ticker which uses Shots on Target and Shots on Target Conceded data for each team to generate a value for upcoming gameweeks for those same metrics. Embedded within the calculation are modifiers to adjust for strength of schedule, and home and away performance.

Shots on Target ticker GW9-14

Shots on Target conceded ticker GW9-14

I would be interested to hear how useful – or not – you think these tickers are. The projected SoT would still need converting and those SoTC would need saving. These factors may help explain why Tottenham Hotspur have not been scoring more heavily and Bournemouth have been conceding more than the SoTC data suggests the team should. It’s also worth remembering that this data was generated from all the games this season, and Manchester City have demonstrated in the last three gameweeks the unsettling impact losing a key member of the backline (Vincent Kompany) can have on a defence.

The final health warning is the data sample used to generate these tickers is small – just four home games and four away games for each team – so something as simple as facing Manchester City could put a dent in a team’s data, even with the application of the modifier. I would expect the tickers to become more stable as the season progresses and the sample size increases.


5 thoughts on “Fixture strength

  1. Dear Diva,

    You can sort out your sample size, as I do, by delving into the past Season. 30 matches, to be precise.

    You may add the Championship promoted clubs by multiplying their last Season’s output by 38/46, and the resulting total again by 30/38.

    Remember, fixture rating, even for a trained statistician, as an art, not a science – so anything too complicated is almost certainly pointless. By all means, try to obtain the right result, but try to remember that even the best result will be a signpost, not a GPS reading …




    • Thanks Doosra. I was keen to see how teams were doing in 2015/16 alone and a carry over from last season could mask that. I’m not saying one approach is more valid than the other, just that I wanted to keep my data set pure to the Premier League 2015/16 and the cost of doing so is losing stability.


  2. Ruth_NZ says:

    Interesting to see Doosra’s comment, my immediate thought was that what you have done would be closer to how he calculates his ticker as well.

    Your work is interesting in any case, it confirms some impressions I had and challenges some others. Thanks.

    The big problem, of course, is changed circumstances. City began to leak goals like crazy as soon as they lost Kompany. Now, with Aguero and Silva out, will there be a similar effect on their attacking prowess? Chelsea have started the season very poorly. At some point a turn of form could be anticipated which may well throw recent statistics up in the air. As a Chelsea watcher I can only say that the absence of Ivanovich (on this season’s form) should only help them defensively.

    Statistics of any kind only report what has happened. In fact, because they are measured in a semi-artificial way, they don’t even give an accurate reflection of that. One example is “shots on target”. Different organisations measure this differently in the case of blocked shots for example. And no organisation that I know of measures quality of shot. A weak shot on target may therefore be used as a positive indicator whilst a great shot that shaves the crossbar with the GK totally beaten counts as a negative.

    It is therefore hard to use statistics to produce an accurate picture of the past. To use them to project future events is much harder to do because (a) a whole variety of changed circumstances (some of which are hard to quantify) have to be taken into account, and (b) the historic statistical picture is inaccurate anyway (because it uses arbitrary and semi-crude measures). That doesn’t mean statistics are of no use, not at all. But it does mean that they will never tell the whole story.


    • I agree with what you say Ruth. This can only be a signpost like Doosra said. The next step if I was to expand the ticker would probably be to include Shots in Box data, but I thought people would be interested to see Shot On Target data alone. My discussion on the blog about Dusan Tadic’s “goal” against Swansea illustrated the caution we need to exercise with the data, but I don’t think it is a big barrier that stops us looking at it.


Leave a Reply

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

You are commenting using your 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