Before this 2013/14 Championship season began, back in July, TWTD ran my article forecasting the end of term result as a ranking of the 24 competing teams constituted by a scoring system using five key variables:

(1) The team’s momentum during the last quarter of the previous season and final league positions.

(2) The current manager: his overall track record; and performance and prospects at the club.

(3) The first team squad: squad quality; team balance, cohesion and resilience; likely net improvements through new additions.

(4) The club’s owners: their approach to managing the club, its finances, and all footballing matters.

(5) The '12th Man': impact of the club’s supporters and match crowds on the club and team.

As the season has now reached midway with all teams having played 23 games, my curiosity got the better of me and I have dusted off the cobwebs from that article to take a look at how it compares to the Championship’s real world competition.

Table one below shows the 24 teams ranked according to my end of season forecast, together with each team’s current league position and the variance between the two (in brackets).

The main question to answer when comparing forecast and actual outcomes is what level of accuracy is present in the forecast ?

This can be assessed objectively by a numeric method, as follows. With 24 teams all allocated a forecast position, the worst possible outcome would be where the forecast positions of the teams are the exact inverse of the real ranking. So for instance, Leicester would be 24th in the forecast, that is 23 positions away from 1st and Derby would be 23rd, 21 positions away from second.

If this procedure is followed and then the resulting variance number per team is added up for all 24 teams, the total(maximum) variance number is 288 (23+21+19+17+15+13+11+09+07+05+03+01 times 2).

In other words, if the forecast has the 24 teams in exactly the inverse order to the actual ranking, the total variance number will be 288. Which is equivalent to saying that the predictive power of the forecast was zero.

If all teams are predicted in their correct places the variance number is zero and the forecast perfect. With 288 representing 0% accuracy and 0 representing 100%, we have an objective accuracy measure.

Adding the variances for each of the 24 teams in Table One, the total is 148. This represents a predictive accuracy of 51% which is, I suspect, a likely 'average' predictive accuracy for a person who is a fan of a club playing in the Championship.

Table One

Championship Season 2013/14 League Table

End of Season Forecast vs Actual Rankings after 23 Games (50%) Completed

Forecast Successes

Moving from the overall picture to the specifics, we can consider the main successes to date.

Two teams’ forecast positions have 100% accuracy – Leicester and Reading. Nine of the 24 (ie, over one third) have forecast positions which are three places or less from their mid-season actual places. Encouragingly, one of these is Ipswich, forecast for fifth and currently sixth.

A variance of three positions or less is predictive accuracy at just less than 90%.

Forecast Failings

The major negatives outcomes are listed in Table Two. These top six negative forecasts account together for 78 variance points, that is, for 53% of the total variance (148 points) in the forecast for all 24 League positions.

Two of these – Blackpool and Burnley – have been surprise outperformers; and two others – Bolton and Watford – surprise underperformers for most observers, not only this writer. It will be fascinating to see whether, during the second half of the season, there is a reversion to expected performance for these four.

Table Two Top Negative Variances

Looking Ahead

By season’s end, is it likely that the forecast accuracy will improve or weaken significantly ? That depends on the degree of variability of teams’ forthcoming performances relative to their achievements to date.

Changes in trend are already active. For example, the Championship’s league table based on only the last six games played would have a top six that includes Huddersfield, Middlesbrough and Wigan.

One prediction alone is worth holding to though: a top six finish for Town!

Happy New Year to all Town fans and TWTD readers.

BillBlue added 12:43 - Jan 1 Methinks you must have had a boring Christmas Andrew but confess I read the article with admiration. I am quite astounded with your prediction accuracy at this point and do take my hat off to you. You are far more accurate than I have ever been probably because I am a born optimist and tend to base my forecasts on my hopes! Well done and it will be interesting to see your notes at the end of the season so thanks for an interesting read.

0

ITFCOYB added 10:28 - Jan 2 Impressive stuff. Could you run some randomly selected control groups to find out what variance score you could expect on guesswork alone? :-)
0

Hiltzkooler added 11:23 - Jan 2 Returning to work after the Christmas break this is an entertaining read..had to read it a couple times to get the gist of the math but loved it nevertheless...excellent AndrewPC!
0

commuterblue added 12:43 - Jan 2 good work, i enjoyed this and the initial blog. ITFCOYB - this is slightly worse than a random group would be expected to get. let's say you simply listed alphabetically (either A to Z or Z to A) you would expect to score 144. this might mean that either you're as well off just guessing, or allternatively it might mean it was just a "bad" half season for this set off predictions but over longer periods this methodology is better than just guessing. the only real way to know is to repeat many times!

0

Bergholtblue added 18:38 - Jan 2 Me and 7 other friends and family do this at the start of the season with the winner taking all in the form of beer or wine, depending on preference. Our scores at this moment are 110, 112, 120, 124, 126, 126 , 152 and 152. Afraid you are an also ran at the moment! (LOL) Like you Burnley Bolton and Watford are our high scorers. In addition Derby and Wigan are giving high scores as well. It all goes to show that the Championship is so much more difficult to predict than the Premier League. I'd be interested in your final score. All the best.

0

bluelodgeblue added 09:20 - Jan 3 Great Blog. thanks
0

Blogs 295 bloggers