Exactly my view from the field. When it went to 2-2 the feeling around the ground was annoyance because we let them in and we should have been winning the match at a canter. This is why both giving away possession close to our goal and giving away possession when the team structure isn’t right are equally dangerous. If the other team scores through constructive buildup play then fair play. But gifting them possession in the dangerous areas or while our shape is compromised is very frustrating because it’s entirely self-inflicted
5 Likes
Its not that I believe it or I don’t, its that the data (based on last season) indicates Bournemouth have the best chance right now of finishing 4th. It’s a spectrum though and they also can forecast as mid-table but currently they’e showing very high because Petrovic had an outstanding season in France and, even accounting for expected performance decline in the PL compared to Ligue Un, is forecasting as one of the best GKs in the league. Their attack hasn’t lost anything so they still have a good projection for goals scored and their defence hasn’t dropped off significantly.
I expect they’ll probably fall a little bit as the data from this season starts to feed into the data but prior to kick off yesterday they were showing as the favorites for 4th (albeit they were only 2 points ahead of Chelsea in 5th and 14 points behind Man City in 3rd).
6 Likes
Thank you, at last a victim of abuse pointing out the often overlooked fact that the vast, vast majority of football fans are decent people. Stop giving disproportionate airtime to scum buckets, and focus more on the good.
5 Likes
Then the data isn’t being correctly utilised. Sorry to sound like a bolshie git (I can’t help it!) and I congratulate you for putting effort in and getting this together and understand that it’s difficult to have a system that can be ‘manually overridden’ but I’m sure you understand that when data gives an inexplicable answer then it deserves to be re-assessed and or the model tweaked
I think all the players got bored of being told how tough it was to play in premier league but yesterday the penny finally dropped and now they know they will have to work 10 times harder to succeed. Wirtz probably never had such tight attention in the Bundesliege but now he knows that he will work like furry as he is a top class professional and Ekitike was totally spent having given his all and was visibly glad to come off.
Sure, but at the same token if you work with models you should have an acceptance of the mantra that “all models are wrong but some are useful.” I wouldn’t use this as a predictive of them finishing 4th but as a marker that they are much harder opposition than they get credit for, even after having given us a really tough time.
7 Likes
Absolutely fair point. I think I would refine my critique to say it should stick to what the data is and frame it in terms of player ratings or something rather than ‘expected to come fourth’ which is a bow that absolutely shouldn’t be drawn from the data.
3 Likes
He is a top manager and I was amazed none of the top teams snapped him up. I also found out he’d had quite a few offers and batted them all off, my guess he is waiting for an approach from la liga to come.
1 Like
From a data analytic point of view, I’d appreciate if there’s a GW to GW correlation.
I do suspect you’ll get a lot of results right. But would be interesting to discuss this match to match. If anything, might help in refining your algorithm.
I don’t doubt your basic premise.
1 Like
Fair enough and I didn’t mean to sound declaritive that they will finish 4th, only that my formula currently has them as the 4th placed team - but it also has them possibly as low as 12th. As @Limiescouse said all models are wrong and if I start changing it to suit my expectations its just a different kind of wrong.
Bournemouth’s high ranking comes from Petrovic having an adjusted 0.24 saved goals above xG (he was over 0.3 in France) so the model currently expects him to stop 24% of Bournemouth’s xG against which he earned by being excellent last season but is also likely to come down as the season progresses and I plug in the real world info.
For context Alisson is a 0.13 saved goals above xG.
Anyway point is Bournemouth are likely much better than most people expect.
9 Likes
Spare a thought for the poor steward.
4 Likes
If only the media would get it…
The clear agenda now appears to be Liverpool FC is racist.
Expect the FA/PL to come and bash us with some ridiculous fine.
Shine taken off an emotionally charged night, for one idiot
2 Likes
The media should speak to the supporters in the wheelchair area…(I think someone posted on X)..and how they ostracised this worm crawling specimen…In the first few minute’s Semenyo( I think) was taking a throw in…and was having a smile and a joke with the fans…The media should see this side aswell.
3 Likes
I think this is a flaw. Perhaps the adjusted xG save factor needs to be further adjusted when it comes from another club let alone outside the league? Not to mention that it a stat that skews in favour of a keeper playing in a leaky team, so a relegation GK moving to a top side heavily skews the stats. For instance rather than signing Wirtz, we would ‘look better’ on this model if we swapped Ali for a higher xG lower ranked goal keeper
1 Like
IMHO they create more racism by doing this and are happy to report on it. As they say, good news doesn’t sell.
1 Like
It’ll be interesting to see if that plays out as skewing towards the goalkeepers of low ranked teams. End of the day they have to make the saves and xG tends to expect slightly less than real life goals.
I initially set it up with save % compared to xShots against but that’s too loose - and I do think that favored low ranked teams who face a lot of low % chances. xG against compared to saves above xG seemed a lot “fairer”.
The high performers by this metric, outside of Petrovic, are people who we would generally consider to be some of the best in the league - Alisson, Ederson, Pickford, Vicario and Selz while the low performers are the ones generally thought of as poor goalkeepers (or at least they were last season) like Sa, Areola and Verbruggen.
Always interested in improving the metrics but based on publically available information I think its limited and sometimes you just get these quirks of the data. Petrovic dominated in France so now he gets a pre-season boost that inflates Bournemouth’s defensive metrics, after a few weeks of “real” data (which of course now includes conceding 4 goals vs 2.16 xG) then the data should settle into a more natural look.
3 Likes
Have you run your model on data at the start of previous seasons for comparison? Or is it built on those sort of stats?
Not yet but I plan on creating a retroactive one for last season (so using 23/24 data for 24/25) if I have time just to see what kind of results it would spit out.
3 Likes
I definitely buy the overall point that Bournemouth are better than most people think, and I’m also appreciative of the data led approach Sweeting gives (great to see you back posting!)
My take, based on nothing but my eyes and general knowledge of football and the Prem, is that with a fair wind Bournemouth might land themselves a Europa spot, but will probably finish 8th-10th if I had to predict it now. They will give some big teams some problems though, and there’s a lot to like about how they go about their business.
5 Likes