Championship, R4 (21-22 August, 2018)

There are just ten midweek Championship fixtures, with two from this round having already been played in the opening week of the season.

The standout game is at the Liberty Stadium, where Bielsa’s Leeds look to keep their perfect start to the season going. However, the model favours Swansea to end this run, with a 2-1 home win (10%).

After a last minute winner in their opening game, Derby have failed to match our predictions and have struggled. The model gives them a 58% chance of victory against Ipswich at Pride Park.

Unbeaten Bolton are the surprise package of the Championship so far. The Trotters welcome Birmingham to the University of Bolton Stadium. This is a match where the most likely result is a home win, but our forecast scoreline is a score draw (1-1; 13%)

As discussed before (here & here), we continue to tinker with our forecasting method. Our aim has always been to provide a single scoreline forecast to hang our hats on. But it is challenging to balance both plausible result and score forecasts simultaneously, which will perform well against some metric. Therefore, we are now making “Fuzzy Conditional” forecasts, discussed here.

The table below contains:

  • The Forecast scoreline, with the % chance of that happening
  • The % chance of a win by the Home team, P(H), or Away team, P(A) (with one hundred minus those two numbers giving the % chance of a draw
Forecast Win (%)
Score Pr (%) P(H) P(A)
Ch Derby Ipswich 1-0 10 58 18
Ch QPR Bristol City 1-2 9 27 46
Ch Swansea Leeds 2-1 10 60 17
Ch Rotherham Hull 1-2 10 19 59
Ch Blackburn Reading 1-1 13 37 34
Ch Aston Villa Brentford 1-0 13 61 16
Ch Norwich Preston 1-1 14 36 34
Ch Sheffield Wed Millwall 1-0 14 42 29
Ch Stoke Wigan 1-0 14 64 14
Ch Bolton Birmingham 1-1 13 39 32

 

Lower Leagues, R4 (21-22 August)

Midweek is a busy week in the Football League, with a full set of fixtures in Leagues One and Two. We continue to refine our conditional forecasts, also. There’s a fuzzy area around where all three probabilities (home win, away win and a draw) are similar,  though usually with the draw a little less likely than the home or away win.

But if the draw is at 28%, home win at 32% and away win at 30%, it seems most likely that a draw scoreline is the better forecast than either team to win. It’s an empirical question as to at what point a draw is the right scoreline to predict, relative to a win for either side, and one we will look at in time.

In the meantime, we make use of the measure of entropy thanks to Claude Shannon, a measure which is highest the more “undecided” is a market – i.e. the closer to 33.333% is each probability in a football match. If the measure of entropy is above 1.09 (corresponding to probabilities between 28% and 39%), we conclude the most likely outcome is a draw, and provide a draw scoreline as our prediction. These are our conditional forecasts.

In the next two tables, we present our forecasts constructed this way. In League One, the strong starts from Sunderland and Barnsley seem set to continue (2-1 away wins, both with probability 9%, but both with win probabilities above 40%). In League Two, Lincoln and MK Dons may end up stealing a march on Exeter and Stevenage in the early four-way tie at the summit of the league.

Most likely Win (%)
Score Pr (%) P(H) P(A)
Lincoln Bury 1-0 10% 40% 34%
Macclesfield Cheltenham 1-0 11% 40% 32%
Colchester Crewe 2-1 9% 48% 29%
Cambridge U Exeter 1-0 11% 40% 33%
MK Dons Grimsby 1-0 13% 59% 18%
Tranmere Mansfield 1-1 13% 37% 35%
Morecambe Northampton 1-2 9% 29% 47%
Newport Co Notts Co 1-0 11% 43% 31%
Yeovil Oldham 0-1 13% 30% 40%
Carlisle Port Vale 1-0 11% 46% 28%
Forest Green Stevenage 2-1 9% 39% 37%
Crawley Swindon 1-2 9% 31% 44%
Most likely Win (%)
Score Pr (%) P(H) P(A)
Oxford Accrington 1-0 11% 51% 23%
Rochdale Barnsley 1-2 9% 35% 40%
Bradford Burton 1-1 13% 39% 30%
Blackpool Coventry 1-0 14% 47% 24%
Charlton Peterborough 2-1 9% 43% 31%
Bristol R Portsmouth 1-0 9% 40% 34%
Doncaster Shrewsbury 1-0 11% 48% 26%
Luton Southend 1-0 9% 40% 33%
AFC W’don Walsall 1-0 12% 42% 29%
Plymouth Wycombe 1-0 14% 53% 20%
Scunthorpe Fleetwood 1-0 14% 53% 21%
Gillingham Sunderland 1-2 9% 34% 42%

 

 

Reality Check

Last weekend the Premier League forecast performance was exceptional. This weekend, it looks no better than what people who thought Gay Meadow was just the name of a funfair could achieve.

But with 40 matches played so far in the Football League, the model is still at a par with last weekend in terms of correct scorelines predicted (4/40; 10%), and could improve with 6 games left to play. Overall, this hit rate is a little lower than what we would expect the model to achieve, but we are still tinkering with our exact method.

Model vs the Experts

Last week was a great week for us. How will this week turn out?

Mark Lawrenson and Paul Merson have both revealed their forecasts – see the table below, where we compare them to our conditional forecasts.

Overall, quite a lot of agreement. Only in one match is there disagreement over which team will win – our model suggests a win for Wolves at Leicester, whereas both experts pick Leicester to win.

Model Lawro Merson
Chelsea Arsenal 1-0 2-0 2-1
West Ham Bournemouth 2-1 1-1 2-1
Tottenham Fulham 1-0 2-0 3-1
Cardiff Newcastle 1-0 1-1 1-1
Everton Southampton 1-0 2-0 1-0
Burnley Watford 1-0 1-0 1-0
Leicester Wolves 0-1 2-1 2-1
Man City Huddersfield 3-0 3-0 5-0
Brighton Man Utd 0-1 1-2 1-2
C Palace Liverpool 1-2 1-1 1-2

Lower Leagues, R3 (17-19 August)

We continue our lower league forecasts. We report conditional forecasts (see here and here for more on these), the most likely scoreline conditional on what is the most likely result. In League One, we see some examples of tight matches – most of all, Wycombe vs Bristol Rovers, where Wycombe’s probability of winning is 38%, and Bristol Rovers’s is 35%.

Clearly there is little between these two teams, yet the probability of a draw is 27%. As such, we report as our conditional scoreline prediction 1-1, reflecting how evenly matched the teams are.

Conditional Win (%)
Score Prob (%) Home Away
Barnsley AFC W’don 1-0 12% 62% 16%
Shrewsbury Blackpool 1-0 13% 41% 29%
Southend Bradford 1-0 12% 43% 30%
Wycombe Bristol R 1-1 13% 38% 35%
Accrington Charlton 1-1 13% 40% 32%
Burton Doncaster 1-0 14% 51% 22%
Walsall Gillingham 1-0 10% 45% 29%
Peterboro Luton 2-1 10% 54% 24%
Portsmouth Oxford 1-0 12% 48% 26%
Coventry Plymouth 1-1 13% 38% 33%
Fleetwood Rochdale 1-1 12% 43% 31%
Sunderland Scunthorpe 2-1 9% 48% 28%

We report the same for League Two below. The Grimsby vs Lincoln match is about the tightest looking match of the entire weekend (and a local derby to boot), and so we predict a 1-1 scoreline. Most of the match-ups this weekend in League Two, perhaps unusually (this is, after all, an empirical question), are between a stronger home team and weaker away team. Will we see more home wins than away wins, as a result?

Conditional Win (%)
Score Prob (%) Home Away
Notts Co Yeovil 1-0 10% 52% 23%
N’hampton Cambridge U 1-0 11% 50% 25%
Cheltenham Carlisle 1-1 12% 40% 35%
Mansfield Colchester 1-0 11% 53% 23%
Port Vale Crawley 1-0 11% 53% 23%
Bury Forest Green 2-1 10% 58% 20%
Grimsby Lincoln 1-1 13% 37% 36%
Oldham Macclesfield 1-0 16% 53% 20%
Crewe MK Dons 1-1 11% 34% 41%
Stevenage Morecambe 2-1 10% 54% 23%
Exeter Newport Co 1-0 12% 58% 19%
Swindon Tranmere 1-0 11% 47% 27%

Premiership, R2 (18-20 August)

The table below gives the model’s forecasts for Round 2 of the Premier League.

We still predict the most likely scoreline in the match, but read here on how and why we have expended our range of forecasts coming from the model.

[Table Key below]

Most likely Win (%) Conditional
Score Pr (%) P(H) P(A) Score C.Pr (%)
PL Chelsea Arsenal 1-1 12 49 26 1-0 10
PL W Ham B’mouth 2-1 10 54 24 2-1 10
PL Spurs Fulham 1-0 15 64 13 1-0 15
PL Cardiff N’castle 1-0 21 62 12 1-0 21
PL Everton So’ton 1-1 11 57 21 1-0 10
PL Burnley Watford 1-0 15 44 26 1-0 15
PL Leicester Wolves 0-1 14 24 48 0-1 14
PL Man C Hudd 3-0 14 92 1 3-0 14
PL Brighton Man U 0-1 17 10 69 0-1 17
PL C Palace L’pool 1-2 8 13 71 1-2 8

Key:

Most likely – gives the model’s forecast of the most likely (unconditional) outcome of the match.

Win (%) – gives the model predicted chance of either the Home team, P(H), or the Away team, P(A), coming away from the match with a win (one hundred minus the sum of these two values gives the chance of a draw).

Conditional – gives the most likely scoreline which is consistent with the most likely result in the match.

C.Pr (%) – gives the chances of the conditional scoreline forecast happening among all other possible scorelines in the match.