As you'll already know from the previous updates. FootySimulator recorded a handsome profit over the season. There were an average of 40 players a week and 40 weeks in total. Therefore a par score would have been one win and one runners up performance. We cashed 7 times (3 x 1st and 4 x 2nd) along with several top 5 finishes. This included a hot spell of three cashes in succession. Unsurprisingly the international form was even better. Three competitions saw one win with the highest score of the season and 2 more 4th place finishes.

In the overall competition, we finished third on 620 points. That was 6 behind second place and 17 above fourth. The winner, somewhat conveniently, was the bloke who marked the entries.

So the exercise was an unqualified success financially. But let's address the question of how much more accurate it was compared to the average performance. Of course we could take part in a competition against 40 idiots to make our scores look good. But although I do not know many of the opponents personally, it is reasonable to assume they were a pretty typical cross section of players, and all were prepared to put their money where their mouths were.

Our weekly average was exactly 15.5 points. Fifth place was 15.0. The median score was 13.55. So the software represents an improvement of 2 points a week on the average performance. Put another way, it improves the score of an average player by 14%. Now, 2 points per week and 14% performance improvement may not sound that much overall, but this was enough to more than triple our initial investment. And as many a professional gambler would testify, it's all about the margins. This season's efforts have shown that FootySimulator can deliver a significant edge.

This experiment has shown how FootySimulator can make money out of medium sized predictions competitions. In 2009/10, we'll be looking for a new test for the software. If I can't think of a better idea in the next couple of months, we'll look to do a correct score Lucky 15 based on the 4 most confident correct score predictions of each week.
Updates have been a bit thin on the ground recently, partly because I have been busy on an exciting new venture (of which more soon) and partly because by week 37 most outcomes have happened several times before. Once the season is over, I will write up a summary of how the money situation has turned out overall. So in what will most likely be my last normal update there follows news of the last few weeks. Although no further money has accrued, we have achieved seriously high scores of 22 and 21 in the last fortnight. In the majority of weeks, both of these efforts would have been good enough to take down a prize. But for the last 2 weeks, our monster score has only been good enough for joint 4th place out of 38. Have the matches been easier to predict recently ? Is Club FootySimulator performing better now we have more results data to work with ? Has there been a breakthrough in human evolution leading to an improvement in the ability of people to predict football scores ? Have we simply been unlucky not to win money the last 2 weeks ? Or is it a combination of all of the above ?

In the overall competition, a top 3 finish is all but certain and that would take our finances from the current triple-up scenario to a quad-up. A finish higher than third is possible.
The last few weeks have seen a fairly typical set of performances. Another second place 3 weeks back confirmed that our minimum return would be greater than three times our original investment. The recent international week saw a result of joint 4th out of 40 - incidentally this was our WORST performance of any international week so far ! In the overall competition, we've been rattling around between first and fifth place and currently lie third.
For the third week in a row we came second, this time out of 40. As a result the initial investment has very nearly been trebled - even if we fail to win any more money this season.
On the overall front, we now sit in joint second just behind the leader. The five man breakaway at the top of the table has become three so we are in a very strong position to receive even more money at the end of the season.
A fairly nondescript Christmas and New Year period with performances dotted around mid-table. However the last two weeks have witnessed successive runners-up finishes out of 38 and, fortunately, the competition pays out two places (with ties). So the significant mathematical news is that these two results mean that even if we win no more money this season, we are guaranteed to have doubled our investment with around a third of the season to go.

On the overall front, things are very close between the top 5 and we are currently in third. It is extremely likely that we will be getting further prize money as this competition pays out down to five places.
Joint first this week out of 40 entrants. This made a third victory of the season and further enhanced our profit margin - even if there are no more victories to come. However there was a cruel twist to this tale. Going into the Monday match we were outright leaders but vulnerable to a few scenarios in the Charlton Derby game. A correct score of 1-1 or 0-1 would have relegated us to second or even third as the FootySimulator prediction for the game was a 2-1 victory for Charlton. And with 94 minutes on the clock that score prevailed and we were all set for a triumph by a margin of four points. One second later, Derby had equalized and first prize was shared. Gutted. This goal also prevented us from regaining top spot in the season long predictions league. Of course last minute goals happen all the time but it is particularly difficult to take when it comes 28 hours after every other game has finished.
November 29th was poor and the FootySimulator entry came about three quarters of the way down the field of 40. This was followed up by a strong performance on December 6th where we finished well inside the top ten and just one result short of the money. The main conclusion I draw here is that December 6th was a pretty standard set of results short of major surprises whereas November 29th was quite the opposite. If we run the simulations just once, virtually any result is possible. However I run them best of 499 to generate the scores that are entered into the predictions competition. So it's natural that this approach mitigates against the surprise outcome. So when there is a lot of freak results in a week, we will do disproportionately badly. The reverse also applies and in the more 'normal' weeks our performances have generally been very strong.
In the overall season-long competition, the FootySimulator entry is rattling between first and third place from week to week. Again, this is out of 40+ entries.
Average scores for the first two weeks saw me first regain and then lose top place in the overall competition. This was followed by a nice surprise in the low scoring 22nd November round. Although my score was fairly unimpressive, it was enough to put me in contention for second place for the week and another payday. A correct score in the Wigan-Everton game will result in a payout, but in any case it's been a relatively strong performance.
Two relatively unexciting weeks: the first saw a lower mid table result, the second saw a result of joint 9th out of 40. Overall, I slipped to second in the first week but have clawed the deficit back to a single point following the November 1st fixtures. Game on.
It was always going to be difficult to come anywhere near last week's stupendous 32 point effort. Nevertheless 20 points and a share of 5th place in a 41 runner field was more than respectable. And this result, coupled with the big one last week has catapulted the FootySimulator entry back to the top in the overall predictions competition. Our overall position in the last few weeks has gone 1-2-3-7-8-2-1. This demonstrates that a very strong week can counter the effects of a handful of mediocre ones. So back at the top, hopefully to stay. But with around 6 other players in close proximity nothing is certain.
I'm always happier predicting international results. I think they are generally easier to forecast, yet a lot of people are out of their comfort zone when trying to predict them. Personally I think it will always be easier to predict Sweden versus San Marino than West Ham versus Wigan. For the predictions competition, I am religiously following the FootySimulator computer predictions, and the international prediction software has been tried and tested for years whereas the club software and data is relatively new.

And I'm pleased to say that my confidence this week was not misplaced. The competition covered 22 matches from the first eight European groups. FootySimulator correctly predicted 20 out of 22 results. The offending pair were Austria failing to beat the Faroes and the Czech Republic failing to draw with Poland. Everything else was correct including the draws in Romania, Bulgaria and Georgia. 6 of these results were three-pointer correct scores which was well above the statistical expectation.

This came to a grand total of 32 points which meant a second victory in the predictions competition. As a result, I can enter this competition without success from now until May and still be guaranteed a profit. After just 2/9ths of the season I have achieved my first long term objective - making more money on the competition than I will spend in entry fees. Secondly, this catapults me back into serious contention for the overall season-long competition. Thirdly, after steaming for a week, I have recaptured the high score record which I lost last week. Admittedly, it was a high-scoring (i.e. easy) week but records are records. Nobody complains that Bob Beamon's 8.90m long jump was achieved at altitude !

The weekend's betting was topped off by a nice treble with Germany, Switzerland and Holland who were the 3 great value bets according to FootySimulator. And the website's Big Game prediction came up too as Germany won 2-1. Happy days !!!
More of the same for these two weeks, only getting slightly more points and finishing slightly higher up the table. Hopefully this will be the start of a sustained improvement. One point of note was that my week 3 record of 25 points has finally been cracked and this is now the second highest score out of 320 odd entries this season. The new record holder achieved 28 points - albeit in a higher scoring (and therefore easier) week. All the same that's an highly impressive performance. Back to home territory (i.e. internationals) next week.
After several weeks of being either near the top or the bottom of the predictions competition, there followed two practically identical weeks remarkable only for their unremarkableness. There were no outstandingly high or laughably low scores from any of the 39 entrants and the FootySimulator entry finished very much in the peloton on both occasions. In the overall competition, we've slipped from first to third but there's not a massive amount in it between the top 7. I hope to be able to give a more interesting update next time !
After a few weeks in the experimental world of Club FootySimulator, I am very much back on familiar territory for international week. Strangely, the Scotland game is not one of this week's 22 selected matches. This turns out to be no bad thing as FootySimulator has it down as a 1-1 draw.

As it happens, I get 4 correct scores and 12 other correct results totalling 24 points. Two of the correct scores (the games featuring both Irelands) came in thanks to last minute goals. Conversely, Glen Johnson's disallowed goal in Barcelona cost me another correct score. Most impressive was the tip for Italy to win 2-1 in Cyprus as predicting the hosts to score against Italy raised a few eyebrows.

As it happened, the top score this week was 25 points. This was a 3 way tie out of 38 entrants. The FootySimulator picks came in equal 4th. If you'd bet a unit stake on each correct score selection, you would have made a small profit (something like �28 from a �22 stake depending on your bookmaker).

On a separate note, one of FootySimulator's big value bets for the weekend was for Fiji to beat Vanuatu. This duly came up and it was refreshing to find a UK bookmaker - Sportingbet - that was giving odds on such a remote encounter.
21 English games plus the Old Firm Derby this week. After last weekend's freak results, Club FootySimulator came up trumps big time with 6 correct scores & 7 other results. My score of 25 beat 39 opponents and second place scored just 18.

If each correct score was priced at 6/1, you would break even with three correct out of 21. So this was a remarkable performance and just the boost I was looking for.
Poor but not a total disaster. FootySimulator is very good at working out the most likely outcomes so when you get a bunch of shocks at the same time, it's not likely to be a good week. Stoke won, Hull drew away and Fulham defeated Arsenal. Enough said.

A single correct score and a bottom half finish. But the consolation was that anybody who'd predicted that lot would have been certifiably insane.
Fifth overall but joint top on Premier League matches only. 19 points including 4 correct scores. Assuming that a correct score is around 6/1, you will break even with three correct scores out of 22 games.
At this point I was wondering why the Championship predictions were not as accurate as the Premier League ones. I now believe this was purely due to chance as the victory of 30th August was largely down to Championship predictions.

Homepage