The script - 2019/20 epl prediction

Discussion in 'The Beautiful Game' started by Sexy Beast, Jul 31, 2019.

  1. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    I've encountered interesting article today so i thought of sharing it here. Hopefully i am not spaming by creating the double thread.

    It's the collaboration between Opta, Squawka and Google to, based on data they have available, predict the outcome of the entire epl season, including goalscorers, assist makers and more. If i understood it correctly, its done entirely by machine learning method. Its the first attempt of the kind, that i am aware of and the results are super interesting imo. I am eager to see how good AI can become in years to come. From first glance, it will definitely need some polishing to do.

    The article: https://sport.bt.com/the-script/the...ue-and-europa-league-outcomes-S11364380558801

    If afterwards you feel inspired to make your own predictions, as i did :D , you can use this website: https://www.worldfootball.net/table_calculator/eng-premier-league/
     
  2. PDG1978

    PDG1978 Member+

    Mar 8, 2009
    Club:
    Nottingham Forest FC
    I nearly started to do that, but decided (for now at least) I didn't want to go through with it (I did them before in other seasons, and usually I'd end up 'predicting' more points for the teams I had near the top and less for ones near the bottom, than I would actually predict if not trying to come up with scores for each game).

    I'll have a go at a final table though, since what I had in mind last year didn't turn out too bad (compared to other times I did it in the past, probably I got a bit closer than normal - whether that's partly because it's now easier to do that?), so that can be a 'beat the stats guys' attempt maybe!
    1 - Man City
    2 - Liverpool
    3 - Tottenham
    4 - Man Utd
    5 - Chelsea
    6 - Arsenal
    7 - West Ham
    8 - Wolves
    9 - Leicester
    10 - Everton
    11 - Bournemouth
    12 - Southampton
    13 - Burnley
    14 - Watford
    15 - Newcastle
    16 - Aston Villa
    17 - Brighton
    18 - Sheff Utd
    19 - Crystal Palace
    20 - Norwich
     
  3. PDG1978

    PDG1978 Member+

    Mar 8, 2009
    Club:
    Nottingham Forest FC
    Because it's still in my mind, I can add my last season's attempt actually for info (maybe it doesn't seem quite as close as I hinted at!) - final actual placing in brackets)

    1 - Man City (1)
    2 - Liverpool (2)
    3 - Chelsea (3)
    4 - Tottenham (4)
    5 - Man Utd (6)
    6 - West Ham (10)
    7 - Everton (8)
    8 - Arsenal (5)
    9 - Wolves (7)
    10 - Fulham (19)
    11 - Leicester (9)
    12 - Newcastle (13)
    13 - Crystal Palace (12)
    14 - Burnley (15)
    15 - Bournemouth (14)
    16 - Brighton (17)
    17 - Watford (11)
    18 - Southampton (16)
    19 - Huddersfield (20)
    20 - Cardiff (18)

    So I can't be sure it's the closest I ever got (I did it a handful of other times but don't remember the predictions now), but maybe it was in terms of the total number of places out by. I'll not be surprised if I'm further out with what I typed on this thread for the new season anyway! Although I'd have some optimism at least at this stage that there's a chance to be slightly closer.

    It'll be interesting to see how the stats model fares indeed. But I'm sure it won't be especially close to 100% correct (it'll be interesting to see whether it gets even one game correct totally in terms of goalscorers/scorelines etc - that sort of prediction is really the 'impossible' I think!).
     
    EvanJ repped this.
  4. Bavarian14

    Bavarian14 Member

    Bayern München
    Jun 1, 2017
    The UCL predictions are hilarious to be honest. Zidane never loses in Champions League. Hattrick of hattricks.. LMFAO
     
  5. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    Yeah. Ive sort of ignored that part. It makes no sense trying to predict random events such as group stage and KO stage draws. And Real surely won't be that good.
    They would have a hard time winning this season even if they were in their 2017 squad, but they are not, they are much worse.
    I think that can be explained by a limited access to databases. Messis hattricks of hattricks feel quite inspired by his 2016/17 onwards group stage form.

    I would be happy with those predictions. Essentially, youve truely missed the placement of Fulham and Watford (and Arsenal, West Ham). Not to bad.

    I wish i could remember mine. I do remember few things tho. My top3 prediction that i do remember for last season were:
    1. City 94-96 (98)
    2. Liverpool 90-94 (97)
    3. Chelsea 80-ish (72)
    6. United
    *those numbers are points

    I feel quite satisfied with that.

    At first glance it seems a reasonable prediction.
    I am still not set on my predictions, i would still want to see how transfers go, but i would say you are underrating Leicester, overrating Wolves, Bournemouth, Southampton, Man United.
     
    PDG1978 repped this.
  6. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    My take on the script is that i only have few problems: that Pogba still plays for United, Martials stats, Salahs goal tally, Iwobi's assists tally, Chelsea's 1st place on Christmas (it sounds ridiculous really),.. it generally feels quite plausible, but i am sure looking at it year from now, it will look ridiculous.

    Predictions like those are impossible because you cant account for all the varibales that take place, but what i think will happen years from now is that some patterns will emerge that we have never seen before and they will influence the way we see football so i am excited about that
     
    ko242 repped this.
  7. EvanJ

    EvanJ Member+

    Manchester United
    United States
    Mar 30, 2004
    Nassau County, NY
    Club:
    Manchester United FC
    Nat'l Team:
    United States
    I think there would be some correct predictions. For nil-nil there are no goal scorers. Last season had 22 of those. If this season has 22 and the machine learning predicts 22, it will probably be correct at least once. I think there would also be correct predictions of a 1-0 or 2-0 win with all the goals being by a top scorer like Salah, Kane, Aguero, or Vardy.
     
    PDG1978 repped this.
  8. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    Ill share my predictions for gameweek 1 because why not:

    Liverpool 3-0 Norwich (4-1)
    West Ham 1-3 Manchester City
    Crystal Palace 0-2 Everton
    Burnley 1-0 Southampton
    Watford 2-1 Brighton
    Bournemouth 2-2 Sheffield
    Tottenham 4-0 Aston Villa
    Newcastle 1-2 Arsenal
    Leicester 0-0 Wolverhampton
    Manchester United 2-1 Chelsea

    *( [the real result] )
     
  9. PDG1978

    PDG1978 Member+

    Mar 8, 2009
    Club:
    Nottingham Forest FC
    I haven't done it myself before kick-off in game 2 (but while posting here and having the Leeds-Forest game on TV I've not become aware of anything that has happened in it yet - I believe it was kicking off at 12:30 though).

    So instead I'll copy and paste your predictions below and choose from most likely to come true IMO, to least likely (indicated by numbers in brackets from 1-9 with 1 most likely). I'll add a second set for the result (one team winning as you suggest, or the draw you select) as opposed to the scoreline.

    West Ham 1-3 Manchester City (1) (2)
    Crystal Palace 0-2 Everton (4) (5)
    Burnley 1-0 Southampton (8) (8)
    Watford 2-1 Brighton (3) (6)
    Bournemouth 2-2 Sheffield (7) (9)
    Tottenham 4-0 Aston Villa (5) (1)
    Newcastle 1-2 Arsenal (2) (3)
    Leicester 0-0 Wolverhampton (9) (4)
    Manchester United 2-1 Chelsea (6) (7)

    As you can see I think Leicester vs Wolves being a draw seems a very feasible call...but just not 0-0 for me!

    And Tottenham winning against Villa I would predict certainly, but precisely 4-0 might seem more of a long shot (not to say especially unlikely or anything).
     
    Sexy Beast repped this.
  10. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    West ham 0-5 City

    West ham players got demoralized towards end. At one point it could led to 1-3 result, but City was ruthless.
    Didnt expect such drop of level by West ham. Probably something that i should consider in possible future predictions. It could be just that they dropped moral due to being intimidated by the team that won 198 points in last 76 games. Its probably psychological thing.

    Btw, var is so awesome. It was funny to see offside goal for few milimiters (Pep certainly had flashbacks to tottenham game) and onside goal for even less.

    I feel like both leicester and wolves will be very careful in their approach. They know they are each others competition for the 7th place so its like the final for them. There will be few chances but they could be missed or saved. Patricio and Schmeichel are great goalkeepers. I originally had 1-1 but went for no goals
     
    PDG1978 repped this.
  11. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    I really dont know what to expect from United v Chelsea game. Does anyone have any idea?
    Both teams are a huge question mark for me. It will be interesting to see.
     
  12. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    Liverpool 3-0 Norwich (4-1)
    West Ham 1-3 Manchester City (0-5)
    Crystal Palace 0-2 Everton (0-0)
    Burnley 1-0 Southampton (3-0)
    Watford 2-1 Brighton (0-3)
    Bournemouth 2-2 Sheffield (1-1)
    Tottenham 4-0 Aston Villa (3-1)
    Newcastle 1-2 Arsenal
    Leicester 0-0 Wolverhampton
    Manchester United 2-1 Chelsea

    Red - i missed the outcome
    Green - i predicted the outcome
    Purple - i predicted the outcome and final score

    Pochettino is one of the best halftime manager in history. There is countless of examples where his team played badly at halftime and then he made some change and they dominated the second one. I dont think i have seen anything like that by any manager

    Kane had me on the edge. I took him in last minute for my fantasy team. It paid off at the end haha
     
    PDG1978 repped this.
  13. PuckVanHeel

    PuckVanHeel Member+

    Oct 4, 2011
    Club:
    Feyenoord


    Serial foulers on and off the pitch.
     
  14. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    Off the pitch?
     
  15. PDG1978

    PDG1978 Member+

    Mar 8, 2009
    Club:
    Nottingham Forest FC
    Hmm, Leicester vs Wolves being 0-0 was kind of inevitable after my comment lol (well, it wasn't really of course but it can happen like this at times).

    Maybe your logic was good indeed on that one then, although I just had a look at the BBC report so far.

    You're heading for a set of predictions a computer might find hard to beat anyway, even that might be the only correct score (Man Utd 2 Chelsea 1 can still happen, but obviously Man Utd might not win and that could knock you down to 7 correct results).
     
  16. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    Liverpool 3-0 Norwich (4-1)
    West Ham 1-3 Manchester City (0-5)
    Crystal Palace 0-2 Everton (0-0)
    Burnley 1-0 Southampton (3-0)
    Watford 2-1 Brighton (0-3)
    Bournemouth 2-2 Sheffield (1-1)
    Tottenham 4-0 Aston Villa (3-1)
    Newcastle 1-2 Arsenal (0-1)
    Leicester 0-0 Wolverhampton (0-0)
    Manchester United 2-1 Chelsea (4-0)

    Lol, this come out better than i expected. I might have guessed the outcomes but in some of them i am way off in the final score like United game.
    Here is the thing tho that makes this so damn hard. Notice that not a single team scored 2 goals. Either they didnt score, scored 1,3,4 or 5 goals.

    I predicted overall that there will be 27 goals scored in gameweek 1 and indeed 27 goals were scored, but the way they were spread is quite different, unpredictable. By far the hardest thing, and in practise, perhapse impossible to predict.

    The reason i think i went for 2 goals so often is because it's sort of a safe answer. Brain hardly thinks that extreme case like 4-0 for United is possible. It always goes for balancing things out.

    Because i was so successful with this i will probably share my predictions for gameweek 2 and will try to have that bias in mind.

    My biggest takeaway is that certain teams in certain scenarios simply give up and get destroyed. The score necessarly doesnt correspond to the relation in quality, but it is rather situational.
     
    PDG1978 repped this.
  17. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    Neither did i watch the game.
    But based on sofasore. 23 shots overal, 3 in target.
    One big chance in the game, missed by leicester. 70-30 possesion in favor of leicester.

    So it seems it wasnt particularly open game.

    Well logic behind it was that ive seen leicester under rodgers and wolves vs big 6 last season. Combining some understat stats that expected goals are low for leicester matches since rodgers took over with reasoning that i already mentioned, its the first game so they will be more careful, it felt like the right score.

    If it was a luck, and 0-0 score had nothing to do with my logic, i will fell into a serious confirmation bias, but what can i do..
     
  18. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    Mourinho gave his predictions for epl:
    "City, Tottenham, Liveprool, City B team."

    For those who know, Mourinho joined sky sports since he doesnt have a job. This will be hilarious:
     
    ko242 repped this.
  19. PuckVanHeel

    PuckVanHeel Member+

    Oct 4, 2011
    Club:
    Feyenoord
    He is right, normally City wins this again. Maybe not the cup competitions. At the same time, we have seen a few campaigns where the deepest squad doesn't win the league.

    One 'problem' is the pundits are hired for making provocative claims:




    “My children go to the same school as Savage's, which is a great guy, by the way. I once asked him why he usually judged me so hard. Robbie then said that it had to do with the competition between the different television stations in England. Robbie also said that he sometimes went over the edge in his critique, but that his attitude also had to do with the fact that he had to pay off a mortgage.
    [...]
    Van Persie therefore knows how it works in the English media, where viewing figures are often of decisive importance. In addition, well-paid analysts are expected to provide clear statements after the conclusion of an important competition."

    https://www.voetbalzone.nl/doc.asp?uid=356362


    Thus Mourinho might exaggerate a bit here....
     
  20. PuckVanHeel

    PuckVanHeel Member+

    Oct 4, 2011
    Club:
    Feyenoord
    #20 PuckVanHeel, Aug 13, 2019
    Last edited: Aug 13, 2019


    edit:

    Also telling:

    [​IMG]

    Pep the genius.
     
  21. PDG1978

    PDG1978 Member+

    Mar 8, 2009
    Club:
    Nottingham Forest FC
    I'm not sure about a computer, but a 'robot' (or robot impersonator) Peter Crouch has made an attempt that I saw in a paper at the weekend, so I'll see if I get closer than him.

    This was his list (and in brackets the change from my list above) - could be interesting for individual teams where we are a few apart to see which is the closer call (I wouldn't be mega confident to be closer at this point certainly!):
    1 - Man City (-)
    2 - Liverpool (-)
    3 - Tottenham (-)
    4 - Chelsea (+1)
    5 - Man Utd (-1)
    6 - Arsenal (-)
    7 - Wolves (+1)
    8 - Leicester (+1)
    9 - Everton (+1)
    10 - Watford (+4)
    11 - Burnley (+2)
    12 - West Ham (-5)
    13 - Southampton (-1)
    14 - Norwich (+6)
    15 - Crystal Palace (+4)
    16 - Bournemouth (-5)
    17 - Newcastle (-2)
    18 - Aston Villa (-2)
    19 - Brighton (-2)
    20 - Sheff Utd (-2)
     
  22. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    And still had to play with Delph as a left back and win 100 points.

    Why do you take in consideration spending that happened in 2009?

    In what kind of shape City were in 2016 when he took over?

    He did his spending for the future as well cuze he could. Give it a few years and i guarante you teams will catch up with City cuze they will be somewhat quiet in years to come.

    Plus he almost never gets what he wants, he goes for the alternative options. He wanted de jong, didnt get him, he wanted sanchez, didnt get him, he wanted maguire, didnt get him, he wanted felix, didnt get him, fred, etc.

    Who is the best player he bought? Laporte, Rodri, Gundogan? The amount he spent is based on quantity of players and City slightly overpaying everyone.

    His period in City is better than the one with Barcelona.

    I didnt expected such shalowness from you.
     
  23. PuckVanHeel

    PuckVanHeel Member+

    Oct 4, 2011
    Club:
    Feyenoord
    #23 PuckVanHeel, Aug 13, 2019
    Last edited: Aug 13, 2019

    Sorry for the brief oneliner. That was a bit unfair. I do not dispute 'Pep' has been the best manager of the last decade, or can have a strong case for that, merely that I'm sceptical about the overused word 'genius' and he made his teams so much better and over-perform (next to their recent history, spending, wages etc.).

    There is merit in the point spending in 2009 has diminished relevance to the situation in 2016, but the spending since 2016 is also the highest. The club has bought for 200+ million four full-backs.

    [​IMG]


    When 'Pep' took over in 2016 the team wasn't in the very best shape, but not too bad either. They had just reached the CL semis (and just like Bayern, Pep hasn't bettered that so far in three seasons) and yes merely 4th in the league. They were however 1st at the start of the season until the 14th matchday and two things cost them:

    1) An ageing team. Starting players and regulars as Sagna, Kolarov, Fernandinho, Yaya Toure, Navas, David Silva, Demichelis, Clichy, Zabaleta, Kompany were all at least 30 years old or reached that age during the season. It is not bad when you have a few of them but they had a bit too many, especially the full-backs.

    2) At least a half dozen injuries (2nd link, 3rd link). This really wrecked their campaign. Can we assign an improvement in this department to Guardiola? They wouldn't be 4th without all those injuries of De Bruyne, Silva, Toure, Kompany etc. Most likely 2nd or 1st really. The squad has only become deeper since.

    In 2016 their recent league record (2011 to 2016) was better as the other English teams and also with a better Champions League result (semi final) than the other underperforming English teams had managed to do. On both fronts they had done better than the other English sides.

    The point on him not always getting his desired players is relevant to the extent of domestic competitors getting those players (only Manchester United, basically). His domestic rivals - bar United - aren't getting those players either, with the exception of Van Dijk perhaps. The teams he beat don't get those players either.

    Of course, all the managers have their flaws. Alex Ferguson broke the domestic transfer record three times in the early 00s, had a much deeper (and better paid) squad than the competitors, yet underperformed domestically and in Europe.


    edit: I reckon of the big five leagues Bayern is the most vulnerable. More vulnerable than Barcelona, PSG, Juventus or City in their league.
     
    Sexy Beast repped this.
  24. Sexy Beast

    Sexy Beast Member

    Dinamo Zagreb
    Croatia
    Aug 11, 2016
    Zagreb
    Club:
    --other--
    Nat'l Team:
    Croatia
    First season for City Pellegrini won the league so what are the reasons behind drop in points and goals in his second, 2015 season? Injuries as well? Aguero was in and out all the time under all managers. I think last season is one of his rare seasons he actually stayed healthy.
    Ive always seen the drop in Citys game as a parallel to Yaya Toures drop in form. In 2014 he was unplayeable and it corresponds to the citys peak.

    Champions league is such a bad argument. It means something but hardly much.
    Depends who you get in KO you can relatively easily reach semi finals. They werent impressive at all in ucl 2016. Kyiv, PSG.

    Pep made some mistakes you can pinpoint but the level of play city showed in 2019 and 2018 in champions league is much higher. Very unlucky but thats the lottery nature of ucl. Monaco was a huge mismatch for them. Young, explosive team vs 30+ old slow defenders and its not like he is going to change his philosophy for 2 games. Was close nevertheless.

    I dont see any other manager that could do what Pep did for city in last 3 seasons and they dont seem to stop. All that without buying anyone above 100M or top 3 player in his position. Virtually anybody he bought was a successful buy except Nolito i guess.

    So yeah, Pep is genius. I dont see the need for sarcasm.
     
  25. PuckVanHeel

    PuckVanHeel Member+

    Oct 4, 2011
    Club:
    Feyenoord
    Jurgen Klopp achieved only one point less with a much weaker midfield and a thinner squad (while losing your most commercially valuable player to Barcelona). Is it really impossible to copy?
     

Share This Page