ap04

This relates to the Siemens scandal, from the little I know the main reason these were aquitted is not them being found innocent but the case dragging on (20 years elapsing for the main charges, there was also other stuff such as admission of bribery from the Germans' part and an out-of-court settlement for the bribes) As far as I know and correct me if I am wrong this guy has been cleared of everything and there is nothing pending or "alleged". Also you are suggesting the court system may be rigged but ignoring there may be foul political/financial/sporting/media interests on the other side wanting him smeared or gone.

Nice to see you care more about others' claims and suspicions than what courts of law decide. And then use "alleged" to chicken out of passing it as your own opinion due to fear of libel and lack of proof. I presume if someone takes you to court and you are cleared you'll remain an (alleged) rapist/wife-beater/pedo for life (and they can happily slander you hiding behind a screen) PS No affiliation to said club or individual just find it really cheap.

I get your point that with a bigger spread results would be more predictable/gaps larger and the table more representative, but their example giving top 83 points and bottom 27 seems varied enough. And even if you changed it a bit it wouldn't mean 800 games but 600, still eons away from 46. I think the author just added this so that some readers don't dismiss it out of hand due to the 'all teams equal' assumption, the theory is the same. But this has made a mess of the topic which is about Warne.

I know the above wasn't meant to say this but in sport the opposite is true, the better you are the more unlucky you get. The two extremes in football: a perfect side that deserves to win every time can only get unlucky (draw/lose); an awful side that deserves to lose every time can only get lucky (draw/win).

If you -or anyone else- think so I strongly suggest you read something like this (not dismiss it, try and understand it)

That's the same as saying "because guys who win the lottery get rich 100% of the time them getting rich has nothing to do with luck". It just addresses the fact most points get promotion, not whether they were lucky to get these points. (Also if you think every league position is deserved ipso facto not sure why you'd use "not lucky" in your post to validate ours, top 2 would be enough in itself)

Not sure what your evidence is but I hope you trust mine as I've worked a bit on this. After 46 games the average percentage of lucky/unlucky games net for a team is precisely 9%. The maximum is over 20% -hard to believe I know. This is based on the standard +2 clear chances aggregate worthy of a win. The average discrepancy to the chances/goal average is 0.35, the maximum over 1. These can all be positive or negative. You understand what the above mean in terms of points, goals and league positions at the end of a season.

'Luck' is not the number of goals and wins, it's whether the goals and wins were proportionate to what happened. In fact the luckier we got the bigger our gd and number of wins would be. Fallacy of the single cause - high gd and number of wins means we were good therefore these were only a result of us being good and luck was not involved.

The opposite has been the case and by a long way, Wycombe was the 8th game we've scored less than we should have and we've scored more in 19(!) Creativity is the area we've lacked compared to others -maybe down to conservative tactics not quality- rather than taking the chances.

This may ignore the majority who haven't, that correlation doesn't mean causation for those who have, and also the longer term trend. You may be interested in this research that covered a large period, basically the impact of a change is minimal after an initial "bounce", and that bounce is just the preceding dip in results, quite often due to luck, that led to the change regressing to the mean. There has also been evidence showing the same would happen without a change.

Not sure how you define "good football", I define it as creating plenty and allowing little at the back. I repeat, first half was roughly +2.5 aggregate chances away to a mid-table side or better. Over 90 minutes that would be +5. Away effect say +6.5. To give you an idea last 10 years the best PL side averages about +5.5, in our league it's probably much less as it's closer. So we weren't just good, we were off the scale good. Yet we gave away a goal, an open goal, a 1v1 saved by Wildsmith's legs, and should have given away a pen, in 55 minutes (so defended relatively poorly) at the same time creating very little (so couldn't have attacked any worse). Again adjusting for home advantage and opponent that's almost relegation material.

Are you referring to this maybe? If so it was way behind him.

Yes, and yes. I already touched on this, the estimated difference between the two is 3 chances at best. Forgetting that at Portsmouth after our early-ish second goal we were equally 'awful', allowing 4 big moments and essentially creating fa. You reckon 3 or 4 chances to 1 (including on top of the obvious ones 2/3 killer corners in case you missed those), in one half alone, away, against a mid-table side, is not amazing and top of the league stuff? Not sure what you are expecting in order to be amazed then.

There is nothing perplexing, if anything yesterday was better. Portsmouth -2.5 chances agg, Wycombe +1.5, difference between Portsmouth and Wycombe give or take 3 chances. Both amazing first halves, poor second but decent overall and bang in line with what was expected. Everything else is kneejerk, bandwagon and outcome bias.

The Pythagorean is not used for football analysis. The lesser reason, because football has 3 points for a win and also draws so it needs a different exponent (and will over-predict every team near the top). The main reason, that it in baseball for which it was conceived -or other high-scoring sports such as basketball- hundreds of runs/points are scored/conceded per season which closely reflects performance, in football goals are rare and do not equal performance. xG was the evolution of Bill James's baseball methods for football, you could have put these instead of goals in the formula or -even better- simply used xPts or net xG instead. As for your conclusion that the team has "underperformed" this couldn't be further from the truth (and folk who have watched our games may attest to this empirically, we have been scoring without being too creative and winning even games on a regular basis). One xPts has us 7 points better off, I have us about 10.

The essence of motorsport is establishing who or what drives fast. The "racing" element is either practical inconvenience or entertainment.

Goals are randomly distirbuted (Poisson). The number of goals per near-goal events (the accurate measure of quality in an ideal model, NOT chances) is in the long run the same for every team, Derby or Man City, being off that in any given period is chance. The discrepancies you mention are mainly not due to the imperfections of the models (and they are imperfect) but the even bigger imperfections of the goal system as a metric (binary, most events ignored, no specificity). A simpler way to show this: Assume two equal teams, playing at exactly the same level all the time. If goals weren't random they would draw every single time. More often than not they won't.

I never look at xG when I've watched the game(s) as that would be idiotic so the above is simply not true (or if it is it's only a coincidence). But sorry if you think that's boring or off-putting, I'll try and post less - not that easy when everyone disagrees eh Except I never said that did I? I said it's not the perfect analysis tool because it will never do that. The luck element would be getting rewarded with 2 goals from only 2 clear chances (that would normally take 7.4), which then allowed you to sit back. But the win would be just about fair as long as the opposition created nothing afterwards. It all points to: Slightly worse this year (remaining schedule caveat) - lucky this year (with points/results not necessarily position) - unlucky last year - one maybe two stronger rivals last year. Another thing I haven't said (yet)

No, what I'm saying is that I can use the metric I want for Derby but have no time or interest to watch 11 other games a week so I have to go by the next best (models). It works both ways though, remember at Northampton everyone saying how we were a disgrace and deserved nothing from the game, guess what that was nonsense too.

I am not mixing anything up, just saying that this is one instance where xG is not a reliable performance indicator, as in a goal being the maximum in performance scales (perfect attacking/worst defending). Also not sure why you think it's either goals or chances (for evaluating not xG), ideally you want an analogue 0 (no threat) to 1 (goal), including execution and all non-shot events. In fact some later models use post-shot ball tracking or even goals as a 'boost'.

In other words that we have scored more and conceded fewer than would normally be due for our play (the opposite of last year). You call that skill, I call it sampling variability.

I am not dismissive of the club, I am dismissive of misguided opinions. 22-23 chances vs 3 best rivals Ipswich h 4-4 Ipswich a 0-4.5 Weds h 1.5-1 Weds a (4-6) red-adjusted 5.5-4.5 Posh h (6.5-3.5) red-adjusted 5.5-4.5 Posh a 4-5.5 23-24 chances vs 3 best rivals Portsmouth h 4.5-6.5 Portsmouth a 2.5-5 Posh h 4-8 Posh a 5.5-7 Bolton h 3-4.5 Bolton a (2.5-6) red-adjusted 4-4.5 22-23 total: 20.5 for 24 against (with Ipswich much tougher but Weds away dead rubber) 23-24 total: 23.5 for 35.5 against

I watched the actual game -as I assume did you- so have 100% of the info, why on earth would I downgrade to an estimate of what I just watched? Why use a pre-shot model when I have the post-shot version that actually happened? xG is an estimate for teams or games you haven't watched (and even then it's very hit-and-miss if not useless for individual games) A great example of the flaws of xG when used in isolation. It gives 0.15 or less for 3 of the goals (which should always measure 1), just 0.1 for their open goal (as close to a goal as it can get), 0.15 for the Nelson, Collins and Cbt openings which were all poor attempts (and 2 of which in the same sequence), and zero for their penalty claim which if given would be almost 1.

We were poor against all 3 rivals at home (4 if you include Stevenage).

In the game I saw we scored 2 goals and created nothing else of note in 90 minutes. Portsmouth on top of their 2 goals missed an open goal, had a 1v1 saved by Wildsmith's legs and the penalty incident at the end. Calling the above an even game and a fair draw is a bit of a stretch.