Plettenberk suddenly spouting about Stuttgart boss Sebastien Hoeness having a €8m release clause today...
Changed my mind to sack him after the first 38 mins v Everton. This has been unacceptable with no sign of it ever changing, I will be staggered if we get a result out of this game as I type at the 40 mins mark.
I've finally voted yes. I really wanted it to work for him, and I think the real blame must be directed at Levy who is leading this club into the wilderness having sold us a vision of the promised land. But rarely have I seen a manager who makes it as hard to get behind him as Ange. He takes a difficult situation and makes it ten times harder with nonsense decisions and tactics.
If you give this manager under this board 100s of matches, you’ll be making a judgement from in the Championship.
After the Newcastle match I pointed out that, while the results were poor, the players hadn't downed tools Not only did today's first 45 minutes show that they had, but the tactics brought to mind the exact moment we knew Conte was done https://tenor.com/b2gtk.gif ...FFS, Tenor doesn't work on here either?
There are well proven statistical techniques that tell you how much data you need to be confident that it proves something. They are the opposite of plucking games out of thin air.
I'll try that again... please log in to view this image That was worth the time and effort, wasn't it...?
Nothing is more certain. Levy's not risking this gravy train derailing for anybody...certainly not one of the chumps. A fresh new chump installed, a couple of cheap loan deals and it's off to work out how much to charge customers to watch us play Ryan Reynolds FC in a Category A game next season.
Because that's how statistics work. The more data you have the more sure you become. I think his calculated number was 95% confidence so there is still a 1 in 20 chance of being wrong even with that many games.
I’m well aware of how statistics work. What I’m saying is why 100 games? Why not say 200, 500 or 1000 games?
Statistical robustness for a basic "unit measure" (such as manager win % ) can require at least 100 data items (changes in data value beyond that will by definition only cause at worst a 1% change in the measure value) . Statistical significance is the point at which whatever the values for the data items are, they cannot be deemed to have possibly occurred by chance. Notably you can have measures whose robustness stabiilises very quickly on a small data set, but are not yet statistically significant.
