Monday 2 November 2015

The Curse of the Unexpected FA Cup Finalists.

As last weekend was Halloween and next weekend is the 2nd round of the FA Cup, I thought I'd make up a FA Cup Curse, so here goes.

Four out of the last six finalists, who were members of the Premier League's threatened thirteen has managed to get themselves subsequently relegated, either in the same season in the case of Portsmouth and Wigan or very soon afterwards for Hull and (stretching the point a bit) West Ham.

So Stoke and particularly Aston Villa should be very afraid.....Or perhaps not.

Of the two, Villa are making the best attempt to validate an over fit, selectively chosen, coincidence,

They've won just four points from their first 10 matches, sacked their manager and are on course to amass around 15 points come May of 2016.

Of course only their most ardent supporters actually believe that Villa will only gain 15 points from their 38 matches. Small sample results are a mixture of luck and skill and where extremes of poor results occur it is likely that particularly bad luck, along with a lower than average talent level is responsible. And given more usual levels of luck, results over the longer term will be less extreme, while still remaining relatively poor.

Keeping with the superstitious theme, over the last 13 seasons, 18 sides have been cut adrift or joint bottom after 10 matches.

Crystal Palace have been in the most parlous state, with just three points. Manchester City shared the worst ten game record with Bolton and Sunderland in 2002 and surprisingly, Derby weren't bottom ten matches into 2007 when they eventually recorded a Premier League low of 11 points after the full 38 games.

A simple scaling up of the points per game record after 10 matches for each of these 18 sides would have predicted an average of 21 points in a full season.

However, in reality these slow starters won an average of 35.5. Still low enough to put many of the sides in the frame for relegation, but a considerable improvement compared to a simple scaling up of their early record.

16 of the 18 teams won more points than this naive up-scaling. Only Sunderland in 2005 and....Sunderland in 2002 won fewer than this simple projection after 10 matches.

The easiest way to adjust a team's projected final points total to allow for the vagary of random chance is to include a percentage of league average performance into their projected rate over the remaining 28 matches.

The average points won in the Premier League per game since 2002/03 is 1.37.

Simply scaling up the bottom team's rate of points gathering over the first ten matches and using this rate over the remaining 28 gives you a rmse against their actual final total of just over 18 points for the 18 sides. Adding 10% of league average of 1.37 points per game to the projection for the remaining 28 games reduces the rmse to 16 points, slightly improving the model.

The minimum is reached when around 65% of the projection comprises of league average and just 35% is derived from the actual ten game record of these struggling teams, reducing the rmse to just under 10 points.

So the good news for Villa is that they will almost certainly have better Premier League days ahead in 2015/16. The not so good news is that their unlucky and poor start will likely see a final points total centred around an average of about 33 points and that's going to make them odds on to continue "The FA Cup Curse".


No comments:

Post a Comment