Crumpet, Firstly I really do hope that you get a job soon. I really don't know how this analytical tool has been developed and it's a guess that it employs elements of multi-dimensional scaling and factor analysis. But both techniques are used in research for the analysis of small number data as it looks for similarities rather than differences. So perhaps I was just trying to show -off
Maybe if I understood what you just said I wouldn't have got a Desmond! And I think I have some teaching experience lined up as a supply teacher/teaching assistant across various schools in the North West (so I can take on a teaching qualification next year) but over the last couple of months recruitment agencies have taught me always to temper my optimism. Fingers crossed anyway!
The majority of market research is directed towards the differences between 'targets' so that we can define that 75.4% of people prefer product X to product Y. The 2 systems that I identified are focused upon looking at the similarities between things. There, we can identify that 75.4% of people think product X and product Y are the same - even though we can't tell why! Never taught in schools only in Business Schools but some undergrads behave like children anyway!!! But from the little that I know, it get's easier to find work when you've got your foot in the door - good luck.
Okay I'm starting to see where you're coming from. For some reason it's reminding me of a common misinterpretation of Bayes' Theorem in the way that it may seem logical to deduce information from seemingly correct assumptions when in actual fact the logic is flawed (or I'm talking nonsense again). And yeah that's the hope, I'm lucky in that I'm hoping to teach Maths and there doesn't seem to be an abundance of good Maths teachers floating about. With a bit of experience I fancy my chances of getting my qualifications done and finding a school round here. And I know exactly how children are these days don't worry! The amount of times I've been asked about my experience working with secondary school children despite the fact I was one 5 years ago is pretty ridiculous. In the current jobmarket it seems that to gain experience you need to have experience. Not only that, but pretty much every 1 in 2 grad jobs are in recruitment which I find just so completely and utterly baffling I can't even begin to fathom the logic in any of it. The mind boggles.
Even if it's properly done, a full season is simply too big to simulate using that kind of Monte Carlo method. A single game being simulated 1,000 times will probably produce a fair degree of accuracy. But there are 380 games in the season, so even if you can simulate each game with 99% accuracy, that will only give you a 2.2% chance of getting all 380 games right. Obviously you don't need to get all 380 games right to come out with the right table, but then the longer the season goes on, the less accurate a simulation made at this stage will become. A single change to the initial conditions, such as a club making a big signing or losing a player, and all the predictions become much less accurate. It also ignores the fundamental conceptual issue that they have simulated the season tens of thousands of times, but it will only happen once. Monte Carlo sims are very poor at accounting for low probability events, such as a long term injury or set of freak results. And as well all know, those low probability events tend to be virtually inevitable in football. It's also quite amusing that their own stats shoot themselves in the foot. At the very bottom of the page it tells you the chance that all three promoted teams face relegation is 6.6%. Guess which three teams face relegation in their predicted table?
Find it hard to disagree with most of your points. I think I said in a previous post though that I think it's more likely to be accurate for the top 6 and bottom 4/5 purely because even if a large number of variables are altered these teams will usually perform to a similar standard set by the two previous seasons (where most of the data was collated from I believe). Even when United have an injury-stricken season you still expect them to beat Stoke at home or Aston Villa. That's why I think it's fair to discount a lot of these small variables if your only aim is to calculate the probable top 4 and probable relegation candidates. Anything from 15th to 7th is almost impossible to predict usually. The fortunes (and misfortunes) of teams like Newcastle, Everton, Aston Villa and West Brom over the last few seasons have often been massively unpredictable and so I agree that the method is highly unreliable in placing most of the midtable sides (not suggesting Everton are a midtable side before anyone sees this as a WUM, just suggesting they're one of those teams seemingly impossible to predict).
I think it will probably be accurate for placing the top 4/5, but that's largely because the top clubs are predictable by nature. It's been Utd, Chelsea, City, Spurs and Arsenal for about the last four seasons, and past results and financial power alone indicate it will be Utd, Chelsea and City in the top three this season. But you hardly need a massive sim to figure that out! Bottom 4/5 I completely disagree - the promoted clubs are often the hardest to pick out, as are the relegation candidates. If you asked everyone on the board their predictions for the top 3/4, I think most people would get it right. But I think almost everyone would pick the bottom 4/5 wrong - how many people at this stage of last season would have predicted Newcastle to be in the bottom five, QPR to be last, and Norwich to be almost in the top half? Their own analysis shows that the three promoted clubs are only 6.6% likely to go down, and yet that's what they've predicted as they simply have no basis on which to make accurate predictions.
What? These sort of things are always completely wrong. I bet they haven't even took in to account that this is our year.
They appear to believe there is a 1.2% chance than next year will be your year. I'd get down the bookies and stick the bet on now
No need to play this season then if we're gunna win the league! Shall we just skip to the following season and put blue ribbons on the trophy?
Undoubtedly the table will be wrong! but it gives the most probable outcome. So its more probable that taken individually the promoted teams will be relegated hence why they are all at the bottom, however it also says at the bottom league table that the chance of all three promoted teams to be relegated is only 5%. If you look at the Liverpool United game it says Liverpool are more likely to win, but a 1-1 score is the most likely result!
You mean you could be more than 1.2% sure that Utd won't win the league this year dave? Brave prediction Tho' I do recall you making a similar prediction last year. Remind me how that one worked out again?
