Arsenal 3-0 Everton

Brentford 0-2 Newcastle

Brighton 1-1 Man United

Burnley 2-1 Nottingham Forest

Chelsea 2-3 Bournemouth

Crystal Palace 1-2 Aston Villa

Liverpool 2-0 Wolves

Luton Town 1-1 Fulham

Man City 3-0 West Ham

Sheffield United 0-3 Tottenham

From Steve:
"I've started on my games for this season recording expected shots (which is the formula without the /90 - fairly sure this is accurate) and actual chances (offside counts as a chance). Where tactics change I multiply by (time used)/90."

I think this is a good way to record the data. I think the right way to think about these games is a binomial distribution where each minute is an independent experiment unrelated to all other minutes with the yes/no being shot/no shot. So recording the shot chance (or expected shots which is equivalent by a constant), the actual shots and the minutes (number of experiments) is sufficient for any analysis you want. Happy to talk more about this (Andy at Motherwell would also be a good guide...we met studying physics in grad school before he went on to biology).