Chelsea have won the 2013-14 PL simulation according to a mathematical model run by an American company Bloomberg sports. After averaging 10,000 runs of the simulation Chelsea came out winners with City in second, united 3rd and Arsenal 4th. Liverpool were predicted to come 6th behind 5th placed Spurs with Everton in 7th. Liverpool also came out Champions in 1.2% of the sims compared to 35% for Chelsea. Full league table prediction available here: http://www.bsports.com/projectedtablesepl More detailed game by game analysis here: https://www.bloombergsports.com/en/football/app With that being said, I'm now definitely going to bet on us winning the League!
What a load of total crap and a complete waste of time. The squads that the sides will use for even the first part of the season are not yet fixes. The simulation cannot account for the managerial changes, etc. etc.
Personally I think people are really over rating Mourinho. Chelsea are different now to before and it is goig to take him a bit more time than no time at all.
Agree with this,the 'special one' wasn't so special last season with a stronger squad at his disposal.
look cheslea have 3 key things. 1. mourinho 2. no big names in the dressing room pulling them back 3. 4/5 really top players however 81 points to win the league... a bit low me thinks. now us in 6th sounds about right... tottenham are well back from arsenal mind and we are only 2 points off them so i think it's not quite right. Also just put 3 promoted sides int othe bottom 3 which suggest this was analysed by lawro frankly. If i were doing this analysis (and i might before kick off) I would put chelsea and city 1/2 but have not decided which way yet.... I would still put us 6th.
Despite the naysayers I've gotta admit I love stuff like this when properly done (mainly because I'm from a mathematical background). I'd guess the original odds were given by bookmakers and if so they have to be somewhat accurate enough considering the bookies are more often than not right. A game being simulated 1000 times will yield the probable result given well estimated odds, and so from a mathematical standpoint I can't argue with the basics of it. ...however this simulation does fail to take into account external factors happening throughout a season (key suspensions, loss of manager, injuries) and so the games simulated towards the end of the season will be much less reliably calculated. I'd say the results would probably be more accurate if it aimed to calculate who was top by Christmas. In any case it's probably a better, more well-thought out season analysis than Lawro could ever come up with.
Fair point but then Jose had to overhaul the 'invincibles' to win the title back then, this time it's a Fergie less United with Rooney unrest and no new signings so far...
Personally I back City for the title. Yes they have a new manager, but they have a great squad, have made some great signings and it is only a year since they last won it.
I don't understand how you can predict the end of the season right now. For a start the January transfer window could play a pivotal role in how the season finishes, Also, whose to say Suarez wont get banned again? Actually they've probably factored that in to their calculations
If they've done it right then is just like a weather forecast. A highly nonlinear system so prediction is impossible, but the forecast is the best objective guess of what will happen.
Theoretically and statistically speaking it's pretty sound for say, a season in FIFA 13 with all injuries/suspensions switched off. Which is why I'd say it could be used accurately to calculate the table by Christmas as there is obviously much less chance of any serious injuries affecting a team's form because there is simply less time for players to get injured. If you're referring to injuries and suspensions to players as the 'major variables' I'd have to disagree with you. If United, City or Chelsea were to take 3-4 major injuries in a season you'd still count on all 3 of them to finish in the top 3. Suggesting that injuries and suspensions aren't the major variables at play. The major variables are simply which two teams are playing. Followed by which team is at home as the third major variable (which is cancelled out over the course of a season). More often than not, the team tipped to win will win, and by repeating the simulation 1000 times the predicted table has confirmed that. I'm also arguing that it is mathematically sound because if you were asked to predict a table from a whim, it would probably look pretty darn similar to this which obviously offers a little more credibility to the method considering it was done using simple algorithms and simulations, rather than through personal bias and opinion. We'll have to see in May whether or not it is accurate really.
Inquiring minds wish to know: Did John Terry put his kit on and get his picture taken with the CGI of the cup?
Now, let's leave LFC aside for a moment. The programme has no way of interpreting what would happen if Rooney left United. It's more than 1 set of values replacing his values. You cannot measure the 'team' effect. Add on to that the effect of a new manager - and his former values do not apply in his new role. Then we have the fan/media pressure which cannot be truly accounted for. This whole thing smacks of a bastardised form of multi-dimensional scaling and factor analysis - both of which a help with low number analysis but find change hard to cope with (if only because the justification lies in repetition (law of inertia of large numbers)
Football has too many variables for a season long computer simulation to have any validity. Injuries, suspensions, player trades, manager swaps, weather conditions, flukes, random results, terrible luck, awful refereeing etc etc etc A clue to how difficult it is to predict, is that this season you can win £1m for predicting the correct scores in 1 weeks PL fixtures
I agree that maybe this should have been calculated after the transfer window shuts, there is obviously still time for teams near the top to change significantly (The future of Messrs. Suarez and Rooney could significantly alter the top 6). But I'm still going to argue that football is a largely predictable sport. I'm not trying to justify the validity of the method through large numbers, but I'm still going assert it lends a bit more credibility than if the simulation were run through just once or twice. I can't comment on low number analysis, mainly because I'm completely unaware of what it is. I'm also fairly unaware of multi-dimensional scaling, I shied away from most statistical areas on my course. But I just feel mathematics aside if you break down the method into small chunks, it still seems sensible: 1) Calculate the odds of each team winning each fixture in the calender using the methodology described in the link. Cons: Obviously it's difficult to assume the same variables of a fixture taking place in September to that of the return fixture possibly in May. January transfers and managerial changes see to this. 2) Simulate each fixture 10,000 times to find the average winner Cons: This inevitably leads to the 'favourite' coming out on top every time by simple probability theory. Leaves no room for 'shock' results. My counter to this is that for every 'shock' result against a team, there will be more or less a shock result in favour of the team throughout the season. Unlikely that these anomalies will drastically change a league table. 3) Collate the average results and look at the final table Not much wrong with this step. It just seems to me that despite the media painting football as a sport susceptible to shocks and surprises, more often than not the probable occurs while the improbable doesn't. Over the course of 10 years this is obviously false, nobody would have predicted Manchester City's rise to Champions, or even Chelsea's before them. But I still think that over the course of a season the league is predictable enough, especially the top 5/6 and bottom 3/5. Not even sure if I'm making sense anymore, I've forgotten how long I've been typing for. I need a job.
