Of tonight’s games, Norwich v Aston Villa (1-1 draw most likely at 12.8% as most likely score) looks closer but Man City v Leeds is also interesting for the neutral (2-0 win for City predicted but our model gives City a 15% less likely chance of winning than bookmakers Mean Odds).
Leeds actually beat Man City in April so there is a slim chance (11.9% our model, 4.8%(!) with bookmakers Mean Odds) they could repeat that but their form has been patchy whilst Man City last lost a league game on 30 October against Palace and have won every game since.
On Wednesday, Arsenal v West Ham looks good, with West Ham flying higher than pretty much anyone expected and Arsenal occasionally looking good and sometimes not (although apparently arguing with their now ex-captain so how that will affect the game, I don’t know).
Our model is very close to bookmakers Mean Odds in terms of likely winners, with just over 1% between the likely odds of an Arsenal win and just 0.5% for a West Ham win. A 1-1 draw seems a fair outcome though.
On Thursday whilst Chelsea and Liverpool are playing, both should win against Everton and Newcastle (both are predicted as 2-0 home wins in our model) – Leicester v Tottenham actually looks closer and perhaps more interesting, and although a 1-1 is most likely, our model favours Leicester at 48% with bookmakers just a bit less at 45.6%.
The Premier League weekend kicks off with a London derby and also a top four clash. West Ham are fourth but have recently lost to Wolves, Man City and then a draw with Brighton, whereas Chelsea are top and have lost one game all season. Our model has a 1-1 draw as the most likely score and this exact score as 12.6% likely.
When compared to Bookmakers Mean Odds (we compare to these as a ‘standard’) however, we find that although our model favours Chelsea at just over 38.6% as winners, bookmakers have them on 53.2%. This is likely to be home advantage helping West Ham’s chances in our model. It’s also probably helped by West Ham’s decent enough expected goals (1.26) and West Ham’s ELO rating is probably better than it’s been for many years.
I did wonder if Lawro at the BBC is just handed some predictions from a similar model to ours but it turns out he thinks it will be a 2-0 Chelsea win (something our model thinks is 6.7% likely). I’d be inclined to agree with him but I expect he is also underestimating West Ham.
Most likely score according to our model: 1-1
Elsewhere in the league, unlike in the Championship, which had 10 draws predicted, we have 5 wins as most likely outcomes.
There are wins for Leeds, Man Utd, Tottenham, Man City and Liverpool on the cards.
Despite this, again when compared to Bookmakers Mean Odds, we find that Man City and Liverpool’s likely wins are underscored as a probability in our model. It seems pretty reasonable to expect Liverpool to beat Wolves, but our model has this only 55% likely, although, whilst a 1-1 draw is the 2nd most likely result at 11.4%, a 1-0 away win in 11.8% and a 2-0 win is 10.1%.
Most of this weekend’s games look pretty close, with all but two games coming out with a 1-1 draw as the most likely score from our model.
This is in part because the two top teams are playing each other rather than someone much lower down the league.
Fulham v Bournemouth
Which leads me to this. These two teams are already 7 to 8 points clear of the rest of the league and are in the automatic promotion slots, although a slip up from either would obviously threaten that slightly.
Bournemouth have slightly the better of it, having won 13 games to Fulham’s 11.
In recent weeks, both teams have drawn twice against lower placed opposition so very recent form is perhaps difficult to give pointers as to what will happen tonight. Our model predicts a draw but actually favours Fulham by the second highest percentage for all teams we favour (52.5%). Fulham also have the highest expected goals as you’d expect.
Most likely score according to our model: 1-1 (11.1% is the percentage likelihood for an exact 1-1 draw – see table below for Score Picks and Draws).
Reading v Hull City
Towards the other end of the table this is quite an intriguing game, Reading won their last game (with Andy Carroll, scoring) and Hull have managed to win their last 4, Reading are scoring but their defence is the third worst in the league.
Our model favours Reading at 46% with Bookmakers Mean Odds just 5% less, and favouring Hull by 5% more than our model, so it should be close.
Most likely score according to our model: 1-1 (this exact score is 12.% likely).
Midweek Premier League games are a fairly rare thing this side of Christmas, unless there is a game postponed.
Usually they are reserved for whatever the League Cup is now called or for European games. This week though we get a full-set, unless it snows again.
Our model (see table above, Outcome Probs) predicts a 1-1 draw for all but three of the games as the most likely score. On the fact of it, this looks like it shows the closeness of the league.
There are three games that aren’t showing as a likely, 1-1. The first is, Villa v Man City, where we favour Man City at 56.6%, with bookmakers mean odds giving City a much higher percentage chance of a win at 70.7%. Steven Gerrard may have done a bit to help Villa since joining but apparently, Man City are in an ’emergency’ according to the BBC. None of this really seems to have affected the likelihood of Man City winning.
Most likely score: 0-1
The other two are, Watford v Chelsea (we favour Chelsea at 62%, bookmakers mean odds 71.5%) and West Ham at home to Brighton, where our model favours West Ham at 58.6% but bookmakers have them just below 50%.
Most likely scores:
0-2 (Watford v Chelsea)
1-0 (West Ham v Brighton)
In tonight’s games our model and bookmakers mean odds each produce almost exactly the same percentage for a Newcastle win at 48.2 to 48% against Norwich. Leeds v Crystal Palace has a bigger gap between our model and bookmakers odds, with our model more heavily favouring Leeds at 47.1%. I expect we are slightly overestimating Leeds here who are 17th compared to Palace in 11th.
I suspect we are also overestimating Everton‘s chances of a win at 21.6% against Liverpool in the Merseyside derby (bookmakers have them at just 12.7%) and personally I think a 1-1 draw is the most likely result for Man Utd v Arsenal (our model agrees but favours Man Utd at 44.4%).
Our table below doesn’t yet include the 3 UEFA League/Conference teams as we update based on immediate past results and haven’t collected these yet (*updated model now shows below the original one*)
Today, Arsenal play Newcastle after a 4-0 defeat against Liverpool last weekend. Newcastle on the other hand have Eddie Howe as their new manager after a negative Covid test (he was technically manager last weekend but was unable to attend due to testing positive at the time). It doesn’t look good for Newcastle, as Arsenal have most of their squad fit.
Arsenal are 5th, Newcastle are currently in last place and our likely winners are Arsenal at 63.7% (see table below in Outcome Probs). Bookmakers Mean Odds are only 1.5% higher, so it looks pretty likely for an Arsenal win. A 2-0 exact score is 11.2% likely with a 1-0 11.1% likely. Personally I’ll be quite surprised if Arsenal don’t win this.
It doesn’t exactly improve Man Utd’s chances though – again our model and bookmakers mean odds have just 1.5% between them in terms of picking the likely winners. Chelsea (between 61.5-59.9%). Both teams won in the Champions League in midweek but Man Utd were slow to get going against Villareal, I thought.
Most likely score: 1-0
Elsewhere the model’s likely winners odds are also pretty close to bookmakers mean odds, except for Brentford v Everton where we have the two teams closer (between 3% on likely winners). Bookmakers favour Brentford at 42.5% though.
More football tonight in the Championship with West Brom v Nottingham Forest at 8pm. I’ve been at a conference all day and it finished late, so am typing this at the point this match is now at half-time.
This is one of those Championship fixtures that I look at and feel slightly nostalgic as when I was growing up in the early 80’s it would have been in the old Division One (now the Premier League).
The most likely score though is a 1-0 home win for West Brom (see model below). We have West Brom at 54% to win with bookmakers mean odds agreeing with us (only 2% higher).
So, instead of nostalgia, looking back slightly less far back to midweek, I just thought I would have a quick look at how we did in terms of predictions.
The big one we got wrong was Fulham v Derby which ended 0-0. It did feel as though there was a better chance of an upset than either our odds or bookmakers means odds gave but it still came as a surprise. Fulham were between 76-73% to win, one of the highest chances to win we’ve had so far this season. Another one the model could hardly predict was Blackburn winning 4-0 against Preston. That said it did give just under 50% chance of a Blackburn win, with bookmakers having them at 56% to win.
One we did get right was the 1-1 between Millwall v Bournemouth where we had an exact 1-1 as 12.6% likely to occur.
So what will happen this weekend?
Logically you’d think if Preston shipped four goals to Blackburn, they won’t fancy their chances against Fulham who were thwarted by Derby in mid-week. Well… Our model predicts a 1-1 draw with Preston having a 22% chance to win. a 1-1 is 11.2% likely as an exact score, with a 0-1 away win for Fulham at 10.4% exact
Speaking of Derby, for some reason the Championship is spread over 4 days this weekend and they play yet another in-form team, Queens Park Rangers on Monday night. After a win and draw against past league leaders, they might just fancy their chances against 4th placed QPR. Our model sort of agrees, in that it gives them a decent chance of a draw. We favour QPR in our model as likely winners at 38% but Derby are only just behind on 34%. Bookmakers mean odds have a slightly bigger margin between the two teams but not much (7% instead of 4%).
Is there any chance for Derby against high-scoring, table-topping Fulham?
Well yes, actually but probably less than 10%, as we give them an 8.8% chance of an away win, with bookmakers mean odds coming in at 9.2%. Fulham are on an expected goals of 2.55 compared to Derby’s 0.75.
One thing you can say though, is that if it wasn’t for their points deductions they would certainly not be bottom of the league. In fact you have to go all the way up the table to 9th placed Millwall to find a similarly mean defence.
Derby haven’t exactly been scoring much (16 in 18 games) but have only let in 18. They beat Bournemouth at the weekend and our model gave them just over a 20% chance of doing that so they may just beat today’s odds too.
Most likely score: 2-0
Millwall v Bournemouth
After Derby beat Bournemouth at the weekend, Millwall must also fancy their chances. Trouble is, despite losing to Derby, Bournemouth have the best defence in the league currently. Unlike Derby, Millwall are in with a shout, according to our model – giving them over 30% change of a win (compared to 27.5 with bookmakers mean odds).
Most likely score: 1-1
Elsewhere, our model matches to within a few percentage point of bookmakers mean odds, except for Blackpool v West Brom. The outcome isn’t exactly different as such, both our model and bookmakers favour West Brom, it is the extent of the difference in the odds.
There is a 10% increase in the likelihood of West Brom winning with bookmakers mean odds, but Blackpool also drop by 10% as likely winners. This could be due to our model including home advantage, it may also be that their expected goals are both very similar (just 0.05% difference).
The Premier League returns this weekend with Leicester v Chelsea as the lunchtime kick-off. Leicester haven’t quite been consistent in the league so far this season so the home game against Chelsea looks fairly tough, especially as Chelsea are still top of the league.
We have this game with the most likely score of 1-1. However, bookmakers mean odds (see table below) have a Chelsea win at 53.6% and our model still favours them at 48.6%. There are a lot of 1-1 draws predicted this weekend, which is more usual for the Championship rather than Premier League.
Most likely score: 1-1
Newcastle v Brentford will be probably of more interest than it would have been a month ago. This is Eddie Howe’s first game as manager of Newcastle. Our model favours Newcastle at 42.3% with bookmakers having just 1% between the two teams (36.3-35.3%). New manager bounce? Well not much is expected, at least, unless a draw counts!
I see Mark Lawrenson on the BBC thinks Newcastle will win 2-0 but personally I can’t see a drastic turnaround in Newcastle’s fortunes without better players.
Most likely score: 1-1
A game to watch will be Liverpool v Arsenal. Liverpool lost their last game before the international break, whereas Arsenal have now won their last three matches. Will this be enough to stop Liverpool?
Our model favours Liverpool at 56.7% and although it predicts a 1-1 draw as the most likely score, there really isn’t much between that and a 1-0 home win with both exact scores being 10.8% likely. Bookmaker mean odds do favour Liverpool even more than us though at 64.7%.
Championship football returns after what feels like quite a long international break.
QPR v Luton
QPR are 6th so are currently in the play-off positions. Luton are 11th but only 2 points behind QPR. We have this game with 1-1 as the most likely score but both our model and bookmakers mean odds favour QPR (see table below).
In terms of exact scores a 1-1 draw is 12.5% likely but a 1-0 home win is 12.2% likely according to our model, clearly showing how close it is between a draw and a home win.
Elsewhere in the league, Reading‘s point deduction takes them perilously close to the relegation places so a win against Nottingham Forest might help. With our model it is a similar tale to the QPR v Luton game in terms of odds, and most likely scores.
It is however, one of the matches where our predictions do not match bookmakers mean odds, with bookmakers favouring Forest instead of Reading, by nearly 40%.
Derby have also had additional points deducted and their misery is quite likely to be added to, as they play top of the table Bournemouth. With home advantage factored into the model, it gives Derby 22.5% to win, with them just creeping into 1 on expected goals. The most likely score however is still a 0-1 away win for Bournemouth. Bookmakers mean odds for Bournemouth winning have just 0.2% between them and our model’s predicted likely winner.
Fulham v Barnsley has Fulham winning 2-0 and there is under 2% between our model and bookmakers mean odds in terms of how likely this is – we have Fulham at 72.8% to win, with bookmakers on 74. This level of certainty has really not happened much in Championship games so far this season. Fulham are also on just below 2.5 expected goals. The way Fulham are outscoring everyone, if they can maintain that you’d surely expect them to be promoted.
This is the first game to catch my eye this evening. It has the highest odds in our model for a likely winner and also is the only game with a 2-1 away win as the most likely score. This is in part down to Fulham’s amazing scoring record so far this season. They won 7-0 in mid-week and have 43 goals so far this season, 16 more than anyone else. Peterborough have the worst defensive record in the league, letting in 31 goals so far so this doesn’t look at all good for them. A 2-1 away exact score is 9.9% likely.
We have them at a 63.7% chance of winning with bookmakers mean odds just over 1% less than that. If Fulham can maintain this form and momentum you’d have to expect them to return to the Premier League.
Compared to earlier in the season, the differences in our odds for likely winners and losers are now generally closer to bookmakers mean odds. The differences are more often than not in the percentage confidence for a winning team rather than a completely different outcome showing. There is just 5% between our model and bookmakers for West Brom to win and 8% for Barnsley at home to Hull for instance.
Blackpool v QPR
Blackpool v QPR is one game where there is an actual difference in the team favoured but still by only 8% difference. We favour Blackpool at 41.5% with bookmakers on 33.2%, whereas we have QPR at 32% and their mean odds show 38.5%.