sim. i gave each team a % to win/draw/lose each game according to their o/u.

it treats each result as completely independent (simming the actual fixtures would be quite a lot better but would take ages so **** that). it's a pretty crude simulation, obviously. theoretically every team could win every game in a particular season, but these things tend to come out in the wash when dealing with large samples. this is 500 seasons

what the thing looks like. each column is a season of results



results (teams are in order of o/u). bournemouth best, hud worst. chelsea ran pretty bad. city not as awful as i expected



winning score in each sim. +8 seems to be the magic number, if your team hits +8 you'll likely be in good shape. +7 not bad either

watford were responsible for the bonkers +18 season



notes: looks like there's a goldilocks zone of ****ness - you want to be just the right amount of ****: not too good, not too ****



hud didn't run bad in terms of total points, they averaged out to about what you'd expect (as did ev1 else). but their results are too clustered around 0 to win many seasons. thinking about it this makes some sense: they cant do worse than -8, their long-term average tends to 0 like ev1 else: fewer huge negative results may suggest fewer huge positive results to compensate

imagine a completely **** team with an o/u of 1 (derby 2008), they'd avg 0 like everyone else and it would consist of a million results of -1, 0, 1, 2, 3, very occasionally more. they'd barely ever win a contest.

there's def a limit to how **** you want to be, and it looks like hud (and brighton, and swansea) might be the wrong side of the line


--

ofc there's an opposite problem at the top end of the table, where your result is capped on the positive side. a few +8s and +6s in the city results, enough to win 16 contests out of 500. they were winners on some +5s and +4s also



--

goldilocks ****ness. at first glance looks similar to hud but there's way more 7s 8s, 9s and 10s in there, which = wins



strangest season was one where 6 teams chopped at +3

You want a team with the highest variance. The highest binomial variance is that of a coinflip (50-50, as opposed to 60-40, 90-10...). Similarly the highest team variance is that of a midranking team (slightly complicated because the games aren't IID). Not sure why Bournemouth & CP score differently - do they have different o/u or did you not do many sims?

We appear to be playing with three wing backs tonight, interesting choice

Pulis plan: Park the bus.

Pardew pla: Park the taxi.

Just seen the Leicester line up, if 1/2 is still available as it was this morning then that's free money but I guess it isn't

That’s pretty funny and pretty epic, sure beats the old lift the mini and spin it.

That West Brom thing is awesome. It is good to see the senior pros setting an example for the younger lads.

I wonder if Livermore was supplying the gak to keep morale up?

Pulling out the big guns for this it seems.

And we're very much not and mostly bus parking, gl but I doubt you need it

This big French lad up top for Chelsea looks handy.

Wenger out

Hull getting bummed

Conte in

Nananana

anyone know if var is in play tonight, commentators don't seem to be mentioning it?

It is yep.

clearly Giroud & Batsman leading the line in a 4-4-****ing-2 would be unstoppable

Brooks freed

That was fun while it lasted

Michy looks legit over there in Germany

City looks bad because they will drop infinite points once they win the title.
How is Burnley injury situation? Swansea City looks sexy to me iyam.

If I was in charge of the game, it would have been whoever concedes the last goal in GMT time in any competion wins. Would have made for some good sweats.
Now its going to require math and stuff to figure out who is winning.

Turns out Pardew had his phone and wallet stolen the night before Taxi-Gate.

Slow pony. I proposed precisely this. I like this version too though.