We get it, you miss FPL. Fill the void with Women's World Cup Fantasy.

How accurate are the bookies at predicting what’s going to happen?

Every week, FantasyBet (that’s us, by the way – we just think it sounds loftier to refer to ourselves in the third person sometimes) release an article that tells FPL managers what the bookies are predicting will happen in the upcoming round of Premier League fixtures. It tells us which teams are the most likely to keep a clean sheet and which players are the most likely to score in any given gameweek.

We call this column Bookies’ Advantage. It’s an ambitious title; one that almost accidentally commits itself to a rather large assumption: that looking at what the bookies have to say is indeed an advantage. But what if it isn’t? What if the bookies are getting it wrong?

Researching the bookies’ accuracy

When I began this study, I knew there was a very real chance that I might expose the Bookies’ Advantage article as a charlatan. Indeed, one of the head honchos at FantasyBet actually took me to one side and warned that there could be “dire consequences” if the numbers didn’t look good. Call me paranoid, but I’m pretty sure he was glancing at my kneecaps as he spoke, too.

But as we will soon discover, he needn’t have been so concerned. Fortunately for my kneecaps – and for the longevity of the Bookies’ Advantage column – the numbers are pretty good. Just how good are they?

To answer this question, I trawled through every Bookies’ Advantage article published across the course of last season – 36 articles to be precise (apparently the FantasyBet team take a short break over the Christmas period, the lazy bastards). My methodology was simple: see who the bookies were backing each week, and then check whether their predictions were accurate.

For each gameweek, I looked at the three teams that the bookies considered most likely to keep a clean sheet and the three players that the bookies considered most likely score. After some fierce discussion at FantasyBet HQ about inclusivity, we opted to use blue instead of green to indicate an accurate prediction. Sorry colour-blind community, but you’ve got no excuse not to share this article around now.

How many of the bookies predictions came true?

Here’s how the table looks.

GW 1st 2nd 3rd 1st 2nd 3rd
1 MUN* LIV* CHE Salah** Aguero Firmino
2 MCI** TOT* MUN Aguero** Kane** Salah
3 LIV* MCI* ARS Salah** Aguero** Aubameyang*
4 MCI** EVE* MUN* Aguero** Kane* Salah*
5 CHE* MCI* WOL Aguero** Morata* Hazard*
6 MCI** LIV* LEI Aguero** Salah** Aubameyang*
7 MCI** TOT WOL Aguero** Kane* Aubameyang*
8 MUN* TOT* BUR Kane** Aubameyang* Lukaku
9 MCI** LIV* WOL Aguero** Salah* Kane*
10 LIV** CHE* WAT Salah** Firmino* Mane
11 MCI* CHE* EVE Aguero** Hazard* Sterling*
12 LIV* LEI* MCI Salah** Mane* Aguero*
13 WOL* MUN* EVE Aguero** Salah* Kane*
14 MCI* CHE* CRY Aguero** Hazard* Morata*
15 LIV* MCI TOT Kane** Aguero* Salah*
16 ARS* MUN EVE Aubameyang* Lacazette* Kane*
17 TOT* CHE MCI Kane** Aguero** Salah*
18 MCI* CHE* ARS* Aguero** Aubameyang* Lacazette*
19 LIV* MUN* MCI Kane** Salah** Aguero*
20 TOT MCI LEI Kane** Salah* Aguero*
23 MCI** LIV* MUN* Salah* Aguero* Sterling*
24 MCI* LIV* MUN* Salah** Aubameyang* Lacazette*
25 MCI** CHE** TOT* Aguero*** Sterling*** Sane**
26 LIV* WOL SOU Salah** Aguero* Aubameyang
27 NEW ARS TOT Aubameyang* Kane* Lacazette*
28 MCI* LIV* WOL Aguero** Salah** Aubameyang*
29 WOL* BRI* MCI Aguero** Kane* Salah*
30 LIV** MCI* CRY Aguero** Salah** Vardy**
31 WHU* LIV* CHE Salah* Mane* Arnautovic*
32 MCI** CHE** MUN*
33 LIV* CHE LEI Vardy* Hazard* Higuain*
34 BRI** TOT** MCI* Murray*** Son** Aguero*
35 MCI** WOL** ARS** Aguero*** Aubameyang*** Sterling***
36 LIV** MCI* TOT Salah** Mane** Aguero**
37 MCI* LIV* MUN* Aguero** Sterling* Aubameyang*
38 MCI* LIV* SOU* Aguero** Sterling* Rashford*

If you’re looking at this and thinking “that’s an awful lot of red”, you’re not wrong. The bookies’ outright favourites for clean sheets were successful just 50% of the time, and they have an overall clean sheet prediction rate of just 43.5%. I can almost feel my kneecaps quaking in their proverbial boots.

The accuracy of anytime goalscorer predictions, whilst improved, doesn’t guarantee my safety either. The bookies were successful in 48.5% of their predictions here, which, for the mathematically challenged, is still less than half.

So what gives? Didn’t I promise decent numbers? Have I just destroyed a FantasyBet institution and, by extension, my chances of ever being able to walk again?

No, I haven’t. You’ll notice in the above table that many of the teams and players have stars next to their names. This indicates the likelihood the bookies gave them to achieve their desired outcome. 3 stars means 70+%. 2 stars means 60+%. 1 star means 50+%. No stars means the bookies gave them less than 50% of keeping a clean sheet or scoring.

Why is this significant? Because if Manchester United are the 3rd most likely to keep a clean sheet according to the bookies, but they’ve only been given a likelihood of 48%, the bookies aren’t actually backing them to keep a shutout. To the contrary, they’re actually telling us that Manchester United are more likely (52% likely, to be precise) to concede.

This changes everything. If the bookies give a team or player less than 50% to achieve their desired outcome and they fail to keep a clean sheet or score, the bookies are accurate. With this in mind, let’s take a look at the adapted table.

How accurate are the bookies’ odds?

GW 1st 2nd 3rd 1st 2nd 3rd
1 MUN* LIV* CHE Salah** Aguero Firmino
2 MCI** TOT* MUN Aguero** Kane** Salah
3 LIV* MCI* ARS Salah** Aguero** Aubameyang*
4 MCI** EVE* MUN* Aguero** Kane* Salah*
5 CHE* MCI* WOL Aguero** Morata* Hazard*
6 MCI** LIV* LEI Aguero** Salah** Aubameyang*
7 MCI** TOT WOL Aguero** Kane* Aubameyang*
8 MUN* TOT* BUR Kane** Aubameyang* Lukaku
9 MCI** LIV* WOL Aguero** Salah* Kane*
10 LIV** CHE* WAT Salah** Firmino* Mane
11 MCI* CHE* EVE Aguero** Hazard* Sterling*
12 LIV* LEI* MCI Salah** Mane* Aguero*
13 WOL* MUN* EVE Aguero** Salah* Kane*
13 WOL* MUN* EVE Aguero** Salah* Kane*
14 MCI* CHE* CRY Aguero** Hazard* Morata*
15 LIV* MCI TOT Kane** Aguero* Salah*
16 ARS* MUN EVE Aubameyang* Lacazette* Kane*
17 TOT* CHE MCI Kane** Aguero** Salah*
18 MCI* CHE* ARS* Aguero** Aubameyang* Lacazette*
19 LIV* MUN* MCI Kane** Salah** Aguero*
20 TOT MCI LEI Kane** Salah* Aguero*
23 MCI** LIV* MUN* Salah* Aguero* Sterling*
24 MCI* LIV* MUN* Salah** Aubameyang* Lacazette*
25 MCI** CHE** TOT* Aguero*** Sterling*** Sane**
26 LIV* WOL SOU Salah** Aguero* Aubameyang
27 NEW ARS TOT Aubameyang* Kane* Lacazette*
28 MCI* LIV* WOL Aguero** Salah** Aubameyang*
29 WOL* BRI* MCI Aguero** Kane* Salah*
30 LIV** MCI* CRY Aguero** Salah** Vardy**
31 WHU* LIV* CHE Salah* Mane* Arnautovic*
32 MCI** CHE** MUN*
33 LIV* CHE LEI Vardy* Hazard* Higuain*
34 BRI** TOT** MCI* Murray*** Son** Aguero*
35 MCI** WOL** ARS** Aguero*** Aubameyang*** Sterling***
36 LIV** MCI* TOT Salah** Mane** Aguero**
37 MCI* LIV* MUN* Aguero** Sterling* Aubameyang*
38 MCI* LIV* SOU* Aguero** Sterling* Rashford*

That’s more like it.

Suddenly, things are looking much better for the bookies. Their anytime goalscorer prediction accuracy is now the right side of 50%, coming in at a very respectable 53.3%. But it’s their clean sheet prophecies that really get the eyebrows twitching: they have shot up from 43.5% to a huge 65.7%.

With football being hard to predict at the best of times and near impossible to predict at the worst, these numbers are pretty impressive. In short, it means siding with the bookies will bring you more success than failure. I suppose that’s probably how they make their money.

I want to finish this article by addressing one final question: just how impressive are these numbers? How do they compare to other prediction tools?

Are the bookies more accurate than the Scout?

To answer that, I turned to the Fantasy Football Scout’s much-heralded recommended line-up. This, I believe, is one of the most ubiquitous guides for predicting what might happen in any given gameweek. But how accurate is the recommended line-up? I went through every single one of them from last season to find out.

My approach was thus: for every player who registered a return in the Scout’s line-up, I awarded an accurate prediction. Crucially, this included assists as well: a huge metric that the bookies don’t account for. As such, if the Scout picked Aguero and Aguero failed to score, he would still be considered an accurate prediction so long as he registered an assist. That applies to midfielders and defenders, too. If a player scored more than the default 2 or 3 points, it was good enough for me.

With this generosity in mind, I was surprised to learn that only 189 of the 418 players predicted to return by the Scout across the season actually did so. That’s an accuracy of just 45.2%.

The aim here is not to diss the Scout, but instead to illuminate the overarching point of this article: FPL is hard to predict. That’s why we like it. If it were easy to predict, it would be boring. So it’s a real testament to the bookies that they manage to achieve such impressive accuracy; if I’ve learned anything here, it’s that we should all be paying more attention to them next season.

And with that, it looks like my kneecaps will live to see another day.


James Copeland

A St Johnstone fan marooned in Norwich, James joined the FantasyBet team in April 2019. He currently co-hosts the Fantasy Weekly podcast and our Bookies' Advantage Youtube series.

BEST FPL FINISH: 2,343 (18/19)

More from James