The BIG 3 - Goal Forecast.

The BIG 3 - Goal Forecast.

Hello Football Fans, I started this article with the intention of trying to work out if there are any stats out there to help me to select players from the Top 3 teams. Personally I no longer think there is a Top 4, Arsenal, Everton, Newcastle & Spurs will fight it out for that 4th Champions league spot this year. So with this in mind for this article I have decided to try predict how many goals the Top 3 will score (and let in,) to help with decision around who to pick from which team. The idea here being to fill in the data in the red section (Cells P2:R5.)

So yes another set of colorful charts, but what do we have here? If only Utd, City or Chelsea have a chance of winning the league, I decided to compare them with past league winners at the same stage of season, 11 games gone. The Top Table is this year so far, and the second table is all of the the past champions, 11 games into there respective seasons (Bar last season where I have included both Utd and City, as they were only split by goal difference.) Note that only 3 out of 12 past winners were actually outside the top 3 (Column C.)

In Mancini's interview this week he said his Strikers are '5 goal short of where they should be,' even with their multi-million pound strike force the stats show they are 4.62 goals (=24.62-20, Cells I4 - I22) behind a past champions mean average 'Goals For' at this point in the season. Mancini had obviously been checking the stats!!

Looking back at the Tables above, the Blue Columns (B:L) are Standard League Table Data, the following 3 White columns (M:O) are Linear Average Forecasts for end of season 'Goals For & Against.' As we are only 11 games into the season it is simple to linearly calculate how many how many goal would occur (at either end) by a simple extrapolation I.e. Lets take the 2010-11 season for example, 11 games gone  means 11 games / 38 total = 29% of the season gone so far. In that season there were 24 goals for the Champions Man Utd (Cell I10.) If we extrapolate 24 goals for 29% of the season to 100% this would mean that there is 83 goals (=24*100%/29%, Cell M10.) Now we know no this can't be correct, but this gives us linear calculation that we can use to compare to what actually did happened at the end of the season (Cells P7:R20.) The percentage difference can then be seen in cells (S7:T20.) So for Goals For Cell M10 Forecasts 83 Goals, there was actuall 78 Goals Cell P10, this is percentage difference of -6% Cell S10.

Now further to this, to calculate this seasons Forecast end goals (P2:R5) this is a two step process. The first step is to compare the data tables to select similar data to infer comparable percentage differences (Estimates, Cells S3:T5.) The Second step is then to use percentage differences to calculate the forecast end goal. Let me take you through an example.

In the above table I have highlighted values in columns J & T and ended up with a value for Q3, the Forecast End Goals Against. This all starts from the goals that Man Utd have let in this season (Cell J3) if we look at cell J23 we notice that the total of 16 goal's let in this year, is three more than the maximum of any team that has won the title after 11 games. 16 goals conceded is also in the second standard deviation when comparing sigma from the mean (11.51=8.38+3.13, Cells J22 +J25), hence taking into consideration all the GA's which fall into the second standard deviation i.e. GA's >11.51, I have highlighted J9, J10 & J19. These values correspond to the 'GA percentage Difference,' cells T9,T10 & T19, we can see that these values are roughly the same, implying that the 'GA percentage Difference' would be similar. Hence by taking the mean average of T9,T10 & T19, we gain the value in cell T3, the 'Estimated GA percentage difference.' 

This then allows us to work backwards to calculate the Forecast End Goals Against, simply by taking the 'Forecast Average End Goals Against' & using this as a base to add the 'Estimated GA percentage difference' too. I.e. 45 = 55+(55*-19%), Cells (N3+N3*T3).

This methodology can then be applied to the rest of the cells, colour matching in the below table (columns I,J,S & T) through the same methodology shows how the forecasts were calculated.

The red section (P2:R5) is the Forecast. I think Man City are a little undervalued and will most probably do much better come May, the stats show that Mancini's new formation is effecting the forwards, at least he has identified this already.

From the Big 3, last year's Champions are a rotation nightmare for Fantasy Football Managers. With more than one star player for every outfield position, it becomes difficult to select players without the fear of a revolving door looming. Tevez, Kun Aguero & Yaya Toure are naturally the players who have been least rotated (when fit.) However Tevez hasn't scored in his last 3 games and Dzeko, Ballotelli are knocking at the door for his place. I transferred Tevez out this week just for the fear of rotation & with City lacking goals so far. The model was the last straw, I cross my fingers he gets rested. I think this model could actually help city identify that they need to revert back to the things they were doing right last season.

Further to this, the model made my decision on whether it was worth while getting another Manchester United player for my Fantasy side quite easy. I already have RVP & Rafeal, United have far too many Midfielders to avoid rotation & can't afford to blow money on Rooney as well. The Model makes it clear that I should not pick another defender, United look set to smash the record for goal conceded and with potential to still win the title, thanks to the amount of goals they are going to score. Perhaps Sir Alex will buy some defenders in Jan. After all they can't be sure they will beat Champions League teams in the latter rounds 4-3 or 3-2.

Chelsea's stats are nicely placed, they could do with scoring a few more goals, rumors of Falco in the window might be the catalyst for a successful season. Due to missing a game in gameweek 15 and hard fixtures I have moved away from Chelsea players at the moment but will be coming back to them, after all the model indicates that as long the scoring rate improves a little, then they look good for a title challenge.