Scorecasting Economists

Beat the Boffins… at Reading FC

RED gets into the Reading FC matchday programme these days, providing analysis and forecasts for the day’s match. Today Reading entertain Cardiff in the FA Cup. It’s a tight match, if league form is anything to go by, with Reading in 15th and Cardiff in 13th. A 1-1 draw is the most likely outcome, but as is the case with scorelines, most of the probabilities are small – 12.5% for the 1-1 draw. As such, we as boffins don’t mind making predictions that are a little away from what is the most likely (12.5%), and in this case, in the matchday programme we opted for a 1-0 win for the Royals, which is 12% likely. Here’s our blurb in the programme:

“This may be a Cup match, but we can’t look beyond the numbers, and the league form of these two sides. The Royals remain in good spirits after some hard fought results in recent weeks, and we see this one going the same way. We think it’s a 1-0 win for Reading at 12%.”

Go buy your programme, and let us know if you think you can do better!

Spotlight on… Oldham Athletic

Since shipping four goals at home to today’s opponents, Salford, on Boxing Day, Oldham have failed to win, picking up just three points in draws. Prior to Boxing Day, they’d won two on the bounce. Salford haven’t really pushed on all that much since Boxing Day either, also only picking up three points. Nonetheless, RED (57%) and the bookies (46%) heavily favour them to win this afternoon. The most likely score is a narrow 1-0 win for Salford (12.8%), the 1-1 draw at just 11.3%. The table below contains some stats relevant to today’s game, comparing the sides. This is the first league meeting of the two sides at Salford.

RED calculates Elo ratings for teams. There’s a bit of an issue in a hierarchical system like football leagues, but the plot below emphasises the different fortunes of the two teams in recent years – Salford have been above Oldham since mid-2017, back when Oldham were still in League One but heading towards relegation.

League Two End of Season Outcome Probabilities

RED’s been hard at work producing end-of-season probabilities – who is going to do what? After the Premier League and the Championship, RED found a colour palette, and produced outcome probabilities for League Two:

The more warmer, the more likely. Surprisingly, league leaders Swindon aren’t nailed on to win the league – instead it’s second placed, but in-form Exeter City (47.1%).

At the bottom, Morecambe look in trouble, and have a 38.7% chance of falling out of the Football League. Stevenage, level on points with Morecambe after just their third win of the season last week, are at 22.9%.

Lower Leagues, 24-25 January

The lower leagues began on Friday night this weekend. RED wasn’t paying attention, instead it was producing end of season probabilities. Here are our forecasts including last night’s Sunderland v Doncaster snooze-fest, for the lower leagues. They incorporate Reuters codes, in case you wonder why the acronyms are a little different to normal.

In both divisions, some teams have (or had) better things to be doing that playing out league matches. Those are the matches without any bookmaker prices (final two columns).

  • Expected Goals: the forecast average number of goals the Model expects for Home and Away teams
  • Outcome Probs: the model predicted % chance of either a Home or Away win, with 100 minus these two numbers being the % chance of a draw.
  • Score Picks: the Most likely forecast scoreline outcome, as well as the most likely conditional on the most likely result outcome happening.
  • Home wins / Draws / Away wins: the predicted % chance of various potential scoreline outcomes of the match.
  • Mean odds: Estimates of online bookmakers’ average probabbility forecasts for the Home and Away teams to win, which could be compared with Outcome Probs.

New Codes for Teams

We’ve realised that, unlike Sajid Javid, adhering to common standards can be helpful. We produce, for example, forecasts alongside Jean-Louis Foulley, and we found ourselves disagreeing on the codes, or acronyms, we use for each team. Jean-Louis found that Reuters has a set of codes for teams all around the world, and so we’ve adopted these.

Here are the Premier League ones:

BOU AFC Bournemouth
ARS Arsenal
AVA Aston Villa
BRH Brighton & Hove Albion
BUR Burnley
CHE Chelsea
CRY Crystal Palace
EVE Everton
LEI Leicester City
LIV Liverpool
MCI Manchester City
MUN Manchester United
NEW Newcastle United
NOR Norwich City
SHU Sheffield United
SOU Southampton
TOT Tottenham Hotspur
WAT Watford
WHU West Ham United
WLV Wolverhampton Wanderers

More broadly, you can find out correspondence file here (which includes Soccerbase team names): https://www.dropbox.com/s/c05v92vktntuyyb/acronyms-file.csv?dl=0.

English Premier League Final Table Simulations, January 2019/20

The University of Reading’s Scorecasting Model has crunched the numbers to look at how this season’s Premier League is most likely to end. Based on 500 simulations of the Model to the end of the season, the below table shows the % chance of each team finishing in positions 1-20.

In every one of these simulations, Liverpool come top, giving them a forecast 100% chance of winning the league this season.

Champions League qualification (top 4): Manchester City have a 100% chance, Leicester 98%, Chelsea 50%, Wolves 17%, Manchester Utd 13%, Spurs 10%, Sheffield Utd 7% and Southampton 2%.

Relegation (bottom 3): Norwich have a 93% of going down, Bournemouth 55%, Watford 47%, Brighton 36%, West Ham 36%, Aston Villa 16%, Burnley 2% and Crystal Palace 2%.

Premier League Final table forecasts, 25/01/2020, RED

Based on all matches and results up to and including 23/1/2020

Championship Final Table Simulations

The Scorecasting Economists and their Model have been busy. Based on 500 simulations of the Model to the end of the season, the below table shows the % chance of each English Football League Championship team finishing in positions 1-24.

At the top, West Brom have a 65% chance of automatic promotion, Leeds have a 50% chance, Brentford 27%, Fulham 21%, Forest 17% and Millwall 8%.

At the bottom, Luton have a 93% chance of relegation, Wigan have a 76% chance, Barnsley 67%, Huddersfield 24% and Charlton 16%. *[Excluding Derby County’s potential points reduction]

Derby County are most likely to finish in 15th or 16th, but still have a 5% chance of at least making the playoffs and only a tiny chance of relegation, without any financial fair play punishment.

Reading FC have a 7% chance of at least making the playoffs and next to no chance of relegation.

2019/20 Championship end of season table forecasts, 24 Jan 2020

Based on all matches and results up to and including 23/1/2020

Lower Leagues, 21 January 2020

There’s a scattering of lower league matches tonight – though not at Cheltenham, where the local(ish) derby with Newport has bitten the dust (for reasons I cannot find any information about). Nonetheless back-on-track Wycombe travel to Peterborough in a match that the bookies think is a fairly safe home banker (44.8%), but RED thinks is in the balance (Peterborough at 38%).

Premier League, R24 (21-23 Jan): Rashford-less Utd have 55% chance of a win against Burnley, according to Scorecasting Model.

Manchester Utd are apparently in trouble, and face a tough end to the season without their star man Rashford. Tomorrow night Utd face Burnley at Old Trafford. The Model gives the hosts a modest 55% chance of the win, compared with the average odds among bookmakers implying 68%. This suggests that the Bookies don’t think the loss of Rashford will be too significant, given that they are more optimistic than the Model, and the Model does not factor in the Rashford injury. Utd are expected to score 1.8 goals, compared wtih Burnley’s 1.0. The most likely outcome though is still 1-1 (11%).

The table below gives the rest of the Model’s forecasts for Round 24 of the Premier League, along with forecast probabilities estimated from current online bookmaker odds.

PL forecasts, R24 2020, RED
  • Expected Goals: the forecast average number of goals the Model expects for Home and Away teams
  • Outcome Probs: the model predicted % chance of either a Home or Away win, with 100 minus these two numbers being the % chance of a draw.
  • Score Picks: the Most likely forecast scoreline outcome, as well as the most likely conditional on the most likely result outcome happening.
  • Home wins / Draws / Away wins: the predicted % chance of various potential scoreline outcomes of the match.
  • Mean odds: Estimates of online bookmakers’ average probabbility forecasts for the Home and Away teams to win, which could be compared with Outcome Probs.

Championship, R29 and 14 (21-22 Jan, 2020) — Model (just) favours Barnsley tonight against PNE.

The table below gives the Model’s forecasts for  some rearranged games from Rounds 29 and 14 of the English Championship, along with forecast probabilities estimated from current online bookmaker odds.

Champs forecasts, R29&14, 2020, RED
  • Expected Goals: the forecast average number of goals the Model expects for Home and Away teams
  • Outcome Probs: the model predicted % chance of either a Home or Away win, with 100 minus these two numbers being the % chance of a draw.
  • Score Picks: the Most likely forecast scoreline outcome, as well as the most likely conditional on the most likely result outcome happening.
  • Home wins / Draws / Away wins: the predicted % chance of various potential scoreline outcomes of the match.
  • Mean odds: Estimates of online bookmakers’ average probability forecasts for the Home and Away teams to win, which could be compared with Outcome Probs.