The enduring popularity of Fantasy Football is in part due to the fact that whilst the premise of the game is relatively simple, mastering the game requires the consideration of a huge number of different factors, and not to mention a little luck along the way. One of the most basic considerations to make is “Will this player start?” Given how few game actions result in negative points, a player who starts their match will almost always score more points than a player who doesn’t play, and therefore scores 0. Even if that player only chips in with 2 or 3 points, ensuring that you have 11 starting players from your squad of 15 each week can easily add up to more than 100 points over the course of the season.

However, that is rather easier said than done, as creating any form of prediction model for starting line ups is near impossible without insider information. Keeping yourself informed regarding injuries, suspensions and new signings obviously helps and some managers place a high value on continuity of selection, but I’d imagine almost every Fantasy Football manager has experienced the frustration of selecting a player as their captain, only to for that player to be rested and either score nothing, or pick up the odd point off the bench.

In some ways, you would expect game week 2 to be one of the easiest weeks to predict starting line ups. Suspensions are rarely a consideration as players cannot accumulate 5 yellow cards in a single game, with Morgan Schneiderlin the only fresh suspension following his dismissal against Leicester City for this game week. Fatigue is also largely not a factor at this stage in the season, with the possible exception of the Liverpool and Chelsea players who went through 120 minutes on Wednesday in the Super Cup.

However, the task of predicting a starting line up is made trickier early in the season due to the fact that, for a variety of different reasons, some players start pre-season at their club later than their colleagues. With the African Cup of Nations, Copa America and Gold Cup all taking place this Summer, many players started pre-season late due to their international commitments, whilst others transfered to their current club late in the transfer window, meaning that even if they had a full pre-season, their manager may prefer to slowly introduce them to the first team. When combined with the normal tactical and form considerations which impact selections throughout the season, what we saw in game week 2 was a series of selections which can only really be explained on a club-by-club basis, rather than through overall trends.

ClubGame Week 1 ResultChanges To Starting XI for GW2Players OutPlayers InGame Week 2 Result
Manchester CityWHU 0-5 MC4Stones, D. Silva, Mahrez, JesusOtamendi, Gundogan, B. Silva, AgueroMC 2-2 TH
Manchester UnitedMU 4-0 CHE1PereiraJamesWW 1-1 MU
LiverpoolLIV 4-1 NC5Alisson, Gomez, Fabinho, Henderson, OrigiAdrian, Matip, Oxlade-Chamberlain, Milner, ManeSOU 1-2 LIV
Brighton & Hove AlbionWAT 0-3 BHA1LocadiaTrossardBHA 1-1 WHU
BurnleyBUR 3-0 SOU0N/AN/AARS 2-1 BUR
Tottenham HotspurTH 3-1 AV1Lucas MouraEriksenMC 2-2 TH
ArsenalNU 0-1 ARS3Chambers, Xhaka, MkhitaryanDavid Luiz, Ceballos, LacazetteARS 2-1 BUR
BournemouthBOU 1-1 SU2Mepham, RicoDaniels, H. WilsonAV 1-2 BOU
Sheffield UnitedBOU 1-1 SU0N/AN/ASU 1-0 CP
Crystal PalaceCP 0-0 EVE1AyewZahaSU 1-0 CP
EvertonCP 0-0 EVE1SchneiderlinGbaminEVE 1-0 WAT
Leicester CityLC 0-0 WW1ChilwellFuchsCHE 1-1 LC
Wolverhampton WanderersLC 0-0 WW0N/AN/AWW 1-1 MU
Newcastle UnitedNU 0-1 ARS2Manquillo, LongstaffKi, KrafthNC 3-1 NU
Aston VillaTH 3-1 AV1HourihaneDouglas LuizAV 1-2 BOU
Norwich CityLIV 4-1 NC1McLeanLeitnerNC 3-1 NU
SouthamptonBUR 3-0 SOU2Stephens, IngsYoshida, HojbjergSOU 1-2 LIV
WatfordWAT 0-3 BHA1GrayPereyraEVE 1-0 WAT
ChelseaMU 4-0 CHE3Kovacic, Barkley, AbrahamKante, Pulisic, GiroudCHE 1-1 LC
West Ham UnitedWHU 0-5 MC5Balbuena, Cresswell, Antonio, F. Anderson, HallerOgbonna, Masuaku, Fornals, Snodgrass, HernandezBHA 1-1 WHU

The table above is arranged following the league table after game week 1, and so it might be reasonable to expect that those lower on the table would have made more ‘reactionary’ changes to their starting line up going into game week 2. That is somewhat evident with West Ham, who made 5 changes following their 0-5 defeat against Manchester City, the joint highest in the league and the most amongst the clubs who did not have midweek European fixtures to contend with. However, two places higher we find Watford, who experienced an equally disappointing 0-3 home defeat against Brighton, yet only made one change, whilst Manchester City made 4 changes, the third most in the league, despite topping the table after the first game week.

Ultimately, the ability to accurately predict a starting line up requires an understanding of the how a particular manager likes to use their squad. Nuno Espirito Santo named the same side for a record-breaking 9 consecutive games at the start of last season, and so it was no surprise that they named an unchanged lineup in game week 2. Predicting a Pep Guardiola starting line up is a somewhat more difficult task however.

If it is difficult form a prediction model for a starting line up, can the number of changes be used to predict how the team is likely to perform in the next match?

Number of changesImproved resultSame resultWorse result
0111
1224
2120
311
4

1
511

Again, sadly the answer is no. Whilst the teams who were either unchanged or made just 1 change generally achieved worse results than those who made 3 or more changes, the sample size is very small and relies on a rather clumsy interpretation of improved or worse results. For example, Tottenham might have picked up fewer points during game week 2 than game week 1, but in the grander scheme of things, is an away draw against Manchester City a worse result than a home victory against Aston Villa? The phrase ‘The Art of Selection’ tends to get used in Football a lot, and whilst art might seem like a somewhat grandoise term, it certainly isn’t a science.

Fantasy Football Tip – Dani Ceballos

Whilst these tips tend to look for the slightly more obscure choices, sometimes there is a player who is so obviously undervalued that you have to get him in your team as quickly as possible. Dani Ceballos certainly fits that description, having demonstrated his quality with two assists against Burnley. Currently available for just £5.6 million, an astonishing price when you consider Henrikh Mkhitaryan costs £7 million, Ceballos doesn’t even need to perform consistently to represent good value, and has the potential to become one of the highest scoring midfielders on the game if he can replicate this form.