Well what a weekend that was, there were an absolute shedload of goals which was great for almost everyone who watches the premier league – the obvious exception being Sunderland supporters! From a fantasy point of view things appear to be looking up for me after a rather disastrous start to the season, I’ve averaged 118 each week for the last three weeks and I’m nearly up to the hallowed mark of 100 points per week with an average of 97.4.
So what’s the focus of this week’s column? Well there are now three well-loved prediction tools in my column so this week we’re going to cover those and a little look at how the Poisson distribution faired.
The Poisson Distribution
As you’ll all be aware, I introduced a new feature last week based on the Poisson distribution – even though this weekend was absolutely chock-a-block with an above average number of goals it did pretty well, even though I was concerned over the lack of data for the model to work on. Let’s have a look:
The good
Newcastle vs Leicester was a correct score. Can’t say fairer than that.
The pretty good
Burnley vs West Ham gave the correct match result
Crystal Palace vs Chelsea gave the correct match result
Everton vs Aston Villa was nearly spot on with the correct score instead favouring 2-0, correct result though.
Southampton vs Sunderland predicted 1-0, 2-0 or 3-0, the result was correct, the clean sheet was correct but it just doesn’t go up to 8!
QPR vs Liverpool the match result was correct but it didn’t take into account QPR helping Liverpool quite so much!
West Brom vs Manchester United would have been a correct score if it hadn’t been for a last minute distance goal from Blind. I’m very happy that its prediction was 2-1 because it shows how the model can fly (accurately) in the face of popular opinion.
The bad
Man City vs Tottenham was a bust, especially as a Spurs fan, not so much because I had Aguero in my ff team
Stoke vs Swansea it got it all wrong.
Arsenal vs Hull was another game where the second favourite score was almost there until Welbeck scored his last minute equaliser – urgh.
So what are this week’s predictions?
The games are all fairly self-explanatory, I especially like the prediction of a massive score of 2-3 for the big match of the weekend. I will again reiterate that there isn’t a huge amount of data making up these predictions. Additionally they focus on home and away averages so Southampton’s will be probably a little higher after last week’s score line than you can count on but Sunderland’s isn’t affected as they are at home.
Goalkeeper
This is still my favourite prediction tool, mainly because the bookmakers do most of the work for me! I especially like that it still has a return on investment of 120% and it picked two of the top four keepers last week including Howard! That’s a miracle this season itself! Here’s this week’s graph:
According to our previous rules this means that the top picks this week are Mignolet, Howard, Szczesny, Forster and Myhill. These selections range in value from 9.56 down to a lowly 4.27 so there should be something for everyone, especially those trying to keep Courtois in the background.
You can see that the low costs of Fabianski, Guzan, Heaton and McCarthy make them interesting options – you could do far worse but you will be taking a significant risk and all have a less than 31% chance of retaining a clean sheet (as per previous rules). Of course, the really brave of you might want to grab McCarthy as their substitute goalkeeper if he gets a decent run between the sticks.
Last Six Predictor
Those of you who read my column a few weeks back will know this prediction tool works on the last six results and their goal difference. In its first week it predicted 8 out of 10 results, last week it said that Southampton, Man City and Everton were the top three picks (that’s a total of 15 goals from those three teams) and managed 8 correct predictions once more – that’s twice it’s been scuppered by 2 draws!
Here’s this week’s predictions:
I’m a little concerned that Southampton are artificially inflated by last week’s numbers so please take heed. Personally I’m interested in how QPR vs AV and Burnley vs Everton goes considering that they both have serious negative goal differences for the last six.
I’ll be back on Thursday with the final Barn Door Bonanza column until January until then please do comment below with any adjustments you think are of interest – I have a number from last week.
