Week 3 was tough. With eight teams playing on Sunday, the decisions made on Saturday involved trying to get into the heads of Wenger, Klopp, Conte, and others to figure out what sides they would select, and hoping you wouldn’t be forced to dip into the free agents when your star player is unexpectedly left out. Low and behold, Arsenal’s ever-popular manager decided to head to Anfield with record-signing Lacazette on the substitutes bench and a defence consisting of Holding and Monreal, rather than what would have seemed the more logical choice of Mustafi and Kolasinac.

On the other side, Klopp had made arguably an even more surprising call by swapping Mignolet for Karius, though from our fantasy football perspective this was more of a minor inconvenience than a major obstacle. Nevertheless, predicting the line ups at a weekend is a crucial step towards becoming a Togga champion, and unless you are able to absorb every bit of news coming in from all 20 Premier League clubs, you end up relying on “expert” sources for your information. But the big question is: how reliable are these sources? Which ones can be trusted and which ones should you avoid?

The Analysis

The predicted line ups given by 4 websites and 2 newspapers were compared for 17 matches across game weeks 2 and 3. The percentage accuracy of each of these is shown below. Websites were accessed on the Friday evening before each game week, whilst information from the newspapers were taken from the Saturday morning.

FantasyFootballScout.com has the highest predicted line up accuracy, at 89%. This is closely followed by the two newspapers: The Guardian (87.7) and The Daily Mail (87.4). Rotowire.com and 101greatgoals.com fair well with 85.6% and 84.2% respectively, but fantasy managers should steer clear of totalpremierleague.com, whose predicted accuracy was just 75.9%.

The differences between the top few sources are very small, and over the course of a game week, equate to less than 1 accurate player difference per match. That said, in week 2 I lost a match by 5.75 points whilst Michail Antonio’s 22 points sat mocking me on the bench. Having not had the time to check at 2pm on the Saturday, I had trusted my sources. And it cost me.

Final Thoughts

One thing that would have been interesting to analyse, and which I may do in the future, is to see which teams are most predictable and which are least predictable. In the current analysis I split the data according to matchups, rather than each team, so I do not know for sure. However, the average accuracy of all the sources for Liverpool’s matches were 71.2% (vs Crystal Palace) and 75.0% (vs Arsenal), which is considerably lower than expected and perhaps suggest extra caution should be taken when selecting your Liverpool players. Conversely, the average accuracy for Leicester’s matches were higher than normal (90.9% vs Brighton and 95.5% vs Man Utd), which likely indicates the Foxes to be a safe bet when preparing your team.

I think this may be a useful topic to revisit in the near future, so please let me know of any suggestions you have, be it other sources or different variables to analyse.