HereBeFoxes

O my heart and bowels.

Hmm

Are they seriously going to discuss Man City winning the league rather than the actual game?

Wooooo yes! But why the Fuchs can we do that but get rolled over by Newcastle, and 10 man Soton?

Ah yes I remember now, this is what it’s like being a Leicester supporter.

I too shall be in High Peak, albeit so far to the west that if I get drunk and fall over the wrong way I’ll be in Cheshire.

I want us to get 3rd at the moment, so all results preferences are based on maximising the chances of that. Ergo Man U win > draw > LFC win.

And well this is fun

I **** yeah!

Yup

So, deduct points from one of them, regulate them, and then vote to expel the rest. EFL can deal with expelling that one, or PL do it if/when they’re promoted again.

Do these people even watch us play? How many passes to youri’s supergoal in the last match?

Let’s overhaul them in the league now and take out Southampton in the semi - FA cup final 1st v 2nd - old school.

Well that’s amusing.

Yes - for the run those tables come from not entirely arbitrarily - they are based on the performance of each team over the last 8 matches - one can put any probabilities one wants, in, I'm just lazy and there are 90+ of them so I make an algorithm to look at the WLD probabilities for each team in a tie and calculate some liklihoods, quite simplistically I admit. An example, Villa-Spurs it calculates as 33% chance home win, 25% draw, 42% away win. Doesn't seem insane, but really who knows. If I could get the probabilities from somewhere else (although obviously they're all going to be arbitrary to some extent, as they are prediciting the future) I could look at the outcome set for that too. Someone suggested betfair as a source, but I don't know how to do that...

Sorry - Monte Carlo is a method of simulating and quantifying the possible range of outcomes from a series of probabilistic events; in this case each remaining match in the Premier League can either be a home win, draw, or away win. One assigns a probability to each of these outcomes, and then runs the remaining 92 matches (in this case) 10,000 times, with the result of each match randomly determined according to the assigned probabilities. You then get 10,000 final tables, each one of which is a real possible outcome. These numbers are the percentage of times in the 10,000 final tables that the listed number of points is enough to get 4th or 3rd place (at least); eg for 3rd - there was no final table where 61 points was enough to get 3rd or higher, whereas 70 points got 3rd or higher in 70% of the final tables. Edit to add: You can also look at it the other way round: in 30% of cases 70 points is not enough to get you 3rd place.

And the same again for 3rd: Points Frequency % (cumulative) 62 0.04% 63 0.5% 64 2% 65 6% 66 13% 67 24% 68 39% 69 56% 70 70% 71 83% 72 92% 73 96% 74 99% 75 99% 76 99.8% 77 99.96% 78 100.00%

Some more output from my (yes, flawed) monte carlo simulation; points needed for (at least) 4th. Eg 68 points will get you 4th (or higher) in 84% of the simulation runs. Remember - it's just a bit of fun (but these are real actual theoretical possibilies from possible combinations of real world results - ie it could happen that 58 points is enough for 4th. Don't lump on it though). Points Frequency % (cumulative) 58 0.04% 59 0.2% 60 1% 61 2% 62 6% 63 13% 64 25% 65 40% 66 56% 67 71% 68 84% 69 92% 70 96% 71 99% 72 99.8% 73 99.96% 74 99.99% 75 100.00%

And come on you Blues!!!

Hmm, I’d be interested to try using the effectively crowdsourced probabilities from Betfair, but as a betfair virgin a quick poke about on their site didn’t lead me to find any odds in the form you lost above - if someone with more knowledge than me can point me at where I could maybe scrape the data I’ll flex my old spider coding fingers and see what I can do...

In the scenarios detailed in that table it was assumed that all of our rivals suddenly found fantastic form, equivalant to them having won 8 and drawn 4 of the last 12, and us having a drop in form equivalent to having won 4, drawn 4 and lost 4 - hence the big swing to Man U; but notably we're still 3rd in 50% of iterations.

It is, in the words of Peter Snow, just a bit of fun, and has to start somewhere. What is point expectation anyway? If one decides in advance exactly how many points each team will get, or what the outcome of each match will be, there's no need to monte carlo, there's just one answer... So one has to assign probabilities to each of the three outcomes for each match, and then monte carlo the lot. You can do that subjectively, or using past results, or randomly, or whatever... I chose initially to use past 6 results as I had the data to hand, then did one where I manually gave various contenders very high probabilities of winning (though never 100% - football doesn't work like that IMO). If anyone wants to make a list for me their assessment of the probability of each of the 3 outcomes (home win, draw, away win) for each of the remaining 93 fixtures in a table form I'll be happy to run their scenario for them

They were in 8th at the time and their recent form was horrible, so the method was only going to have them going lower. Even putting in last night's result they've stll lost 4 of the last 6, so aren't (in this model) really a threat - there are only 7 interations out of 10,000 where they finish above us (them 4th, us 5th) (there was only 1 until they beat Wolves). In the real world of course recent form is not going to predict the future, so one can fiddle with the probabilities to compensate. Here for instance is one where Man City, Man Utd, Chelsea, West Ham, Arsenal, Spurs, Everton and Liverpool are all given championship-winning form until the end of the season (2.2 pts/game assuming playing the average team in the division - actual points gained will be affected by the quality of the opponents in the actual fixtures, and of course the random element that is inherent in the model, and I would argue in football too) and we're given poor to mediocre form (1.3 pts/game): Final Position Man City Man Utd Leicester Chelsea West Ham Everton Spurs Liverpool 1 99.8% 0.2% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 2 0.2% 74.6% 10.4% 13.7% 0.2% 0.4% 0.5% 0.0% 3 0.0% 17.2% 39.7% 32.7% 0.9% 3.5% 5.8% 0.3% 4 0.0% 5.9% 33.1% 29.5% 3.2% 10.7% 14.8% 1.2% 5 0.0% 1.3% 12.6% 13.4% 11.3% 22.8% 28.8% 6.1% 6 0.0% 0.3% 2.9% 6.4% 17.2% 27.1% 23.8% 13.8% 7 0.0% 0.1% 0.5% 3.0% 23.3% 19.3% 15.2% 23.5% 8 0.0% 0.0% 0.1% 0.6% 23.3% 11.1% 7.8% 28.9% 9 0.0% 0.0% 0.0% 0.1% 16.4% 4.5% 2.9% 21.2% 10 0.0% 0.0% 0.0% 0.0% 3.9% 0.5% 0.3% 4.4% 11 0.0% 0.0% 0.0% 0.0% 0.3% 0.0% 0.0% 0.6% 12 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.1% So I think we're in a good position. Based on this entirely theoretical model! You betcha.

Well hopefully that will give you faith in the predictions! (I meant top 3 - ie CL regardless of other European successes).