The data-driven XI: Who to select in your Fantasy Premier League team

To say that Fantasy Premier League (FPL) is all about data is stating the obvious – from goals scored to assists and clean sheets, anyone who plays the popular virtual football game is using statistics to some degree in order to make decisions about their team.

But some fantasy football players are taking things further. By digging deeper into the data, in theory it becomes easier to predict what will happen in the coming season or even in the next few gameweeks (GWs). We have a vast array of metrics at our disposal so it stands to reason that the closer you look, the more informed your decisions become.

The difficulty arises when you begin considering which statistics to factor into your team selections. Should you look at the number of shots a player has? What about shots on target? Then there are defenders to consider, what are the key metrics for a centre-back for example?

There’s no definitive answer and part of the beauty of the game is that everyone plays it differently. But broadly speaking it’s best to keep things simple. The key things to look at in terms of attacking returns (besides simply goals and assists from the previous season) are expected goals (xG) and expected assists (xA). When it comes to the defensive side of the game, it is better to look at the overall performance of the player in question’s team – primarily focusing on expected goals against (xGA).

Of course, there are a whole host of caveats and individual player-specific issues to consider, but xG, xA and team xGA seem like logical starting points from which to begin building a side. Each player’s record in these areas then needs to be weighed up against their FPL price.

At Twenty3 we’ve taken a deep dive into the stats so that you don’t have to, in order to pick out some potential value-for-money FPL stars that could help you shoot up the rankings and win your mini-leagues in 2020/21…


There’s little consensus on how to approach the goalkeeper position when it comes to fantasy football. While some managers prefer the reliability of a ‘big six’ keeper – someone who is likely to avoid any rotation and can remain in your team throughout the season – others are more keen to rotate budget options to save money for elsewhere in their side.

When picking a stopper we want to consider the number of goals his team are likely to concede (xGA), their actual number of goals against and their saves per 90 last season (for additional save points).

If you’re going to go for a ‘set and forget’ option then Ederson (Manchester City, £6.0m) seems like the best choice. He conceded the joint-lowest number of goals last season of any goalkeeper (0.76 per 90) and City’s xGA was also the best in the Premier League (34.21). You sacrifice save points by picking the Brazilian, but in return get a ‘nailed on’ defensive starter who is immune from Pep Guardiola’s dreaded rotation. After signing Nathan Aké from Bournemouth their backline should be even stronger next season, but you will need to wait until GW2 to bring him in with no Citizens fixture in GW1.

Ederson's shot map conceded for Manchester City in the 2019/20 Premier League.

For those shopping in a lower price bracket, Leeds United goalkeeper Illan Meslier (£4.5m) is an intriguing option. The Whites’ defensive record in the Championship was, frankly, incredible last season – their xGA was just 35.91, over seven goals better off than second place Brentford (43.42). While he wasn’t first choice for the whole season, Leeds conceded just 0.37 goals per 90 with Meslier in the team and he looks set to be the club’s No.1 for 2020/21.

Mathew Ryan (Brighton & Hove Albion, £4.5m) and particularly Alex McCarthy (Southampton, £4.5m) may be more tempting because of their favourable early fixtures and saves per 90 record in 2019/20 – 2.85 and 2.69 respectively – but Meslier is perhaps an underrated long-term option between the sticks.


Defenders clearly fall into two categories; those that can provide attacking returns significant enough to justify their high price point and budget ‘enablers’ who allow you to spend more of your funds in midfield and attack.

With regards to the former, it’s almost impossible to look past Liverpool full-back Trent Alexander-Arnold despite his rather expensive £7.5m price tag. It’s no surprise that the defender is in over 50 per cent of teams – his 13 assists were only beaten by one player last season (Kevin De Bruyne with 19), while Liverpool conceded the fewest goals in the division (33).

Trent Alexander-Arnold Passes Made Network for Liverpool in the 2019/20 Premier League season.

Neither Liverpool nor Alexander-Arnold were significantly outperforming their xGA or xA either, suggesting that they could hit similar heights this time around. The Reds were expected to concede marginally more (36.67), while their right-back had an impressive xA of 11.89.

Another player who catches the eye when it comes to attacking returns is Everton’s Lucas Digne. Again, if we look at xA from 2019/20 the Frenchman excels – 8.2 expected assists was the fourth most in the entire Premier League, which suggests he could be good value at £6m. With Everton’s new additions in the centre of the park (Abdoulaye Doucouré, James Rodríguez and Allan) you would expect that midfield to provide a better screen for the Toffees’ defence this season, potentially leading to a greater number of clean sheets too.

Clean sheets are really the only thing to be concerned about when it comes to budget defenders, with any extra points from assists, bonus points etc considered something of a pleasant surprise rather than an expectation.

Of the teams with £4.5m defenders listed on FPL, Burnley, Crystal Palace and Brighton & Albion are the best performers for goals conceded last season and also xGA, while Southampton’s xGA of 53.34 was significantly better than the 60 goals they actually conceded. Charlie Taylor (Burnley), Tariq Lamptey (Brighton) and Kyle Walker-Peters (Southampton) are all worthy of consideration, as are Crystal Palace’s £4.0m enablers Tyrick Mitchell or Nathan Ferguson – the former looks set to start the season but Ferguson could be the better long-term option.

Selecting centre-backs for goal potential from set pieces can be sensible, but all the above players are full-backs in the hope that they chip in with the odd assist or two over the course of the campaign.


Manchester United and Manchester City’s delayed start to the season makes selecting your midfield heavy-hitters pretty straightforward this season. What is going to be more difficult is finding a way of bringing in the likes of De Bruyne (£11.5m), Raheem Sterling (£11.5m) and Marcus Rashford (£9.5m) – three players who ranked particularly highly if we look at xG + xA in 2019/20.

In the meantime, Mohammed Salah (Liverpool, £12m) looks an essential pick after racking up 24 goals and assists, with his expected numbers around the same level. It’s a similar story when it comes to Pierre-Emerick Aubameyang (Arsenal, £12m), who has been reclassified as a midfielder and scored 22 league goals last season.

Mo Salah's Radar for Liverpool in the 2019/20 Premier League.

Intriguingly that was significantly higher than the Gabonese striker’s xG of 14.26, raising an interesting point about stats vs the ‘eye test’. On paper that difference should ring alarm bells, but anyone who watches Aubameyang on a regular basis will know that he is a sublime finisher who has the ability to find the back of the net from often improbable positions. In theory this should rule him out of a purely ‘data-driven XI’, but with fixtures against Fulham and West Ham United in GW1 and GW2 he serves as a useful placeholder until City and United are back in action.

A £4.5m rock-bottom bench-warmer will be needed to facilitate big money stars in midfield and up front (the likes of James McCarthy (Crystal Palace), Yves Bissouma (Brighton) or Harrison Reed (Fulham) are all much of a muchness), but there are a few interesting differential options available for £6m.

Southampton’s James Ward-Prowse may have only registered seven goals and assists last season, but his xG + xA was far higher at 13.65. With the Saints going from strength to strength under Ralph Hasenhüttl, it’s possible that the England international will get closer to that number in 2020/21. Burnley’s Dwight McNeil may not play until GW2 but he was third for xA in 2019/20 (9.53) so is another one to consider for similar reasons.

James Ward-Prowse Shot and Shot Assist Map for Southampton in the 2019/20 Premier League

Ryan Fraser didn’t enjoy the best of seasons on the South Coast last year, but will presumably be significantly more motivated now that his long-running contract impasse with Bournemouth is finally at an end. Reunited with his partner in crime Callum Wilson, it’s probably worth casting our minds back to 2018/19 when he registered the most assists in the league (13) and had an xA of 11.99. Is it worth gambling on a return to that kind of form on Tyneside?

Finally, West Bromwich Albion talisman Matheus Pereira is available for the same price – if he can get close to matching the 13 assists he managed in the Championship last season then he could represent excellent value. Certainly one to keep an eye on.


With most FPL players placing more of a focus on midfield assets this season, mid-priced forwards are very much the order of the day for 2020/21. That potentially rules out Jamie Vardy (Leicester City, £10m) for many people, despite his 23 goals across the previous campaign.

It may be worth finding the cash for the slightly lower priced Timo Werner (Chelsea, £9.5m) though. While it can be hard to ascertain how new arrivals to the Premier League will perform, Werner was fifth across all of Europe’s top five leagues for xG + xA with 26.88, only beaten by Robert Lewandowski, Lionel Messi, Cristiano Ronaldo and Ciro Immobile. That seems like a risk worth taking.

Timo Werner's shot placement map for RB Leipzig in the 2019/20 Bundesliga.

Danny Ings (Southampton, £8.5m) is a popular choice – currently in over 25% of teams – but it is worth noting that his tally of 23 goals last season was well above his xG (16.2). That may make Raúl Jiménez (Wolverhampton Wanderers) a better bet at the same price. Incredibly his xG was actually marginally higher than the 24 goals he scored at 24.45.

For those with a tighter budget, Everton’s Dominic Calvert-Lewin (£7m) was seventh in the Premier League for xG (16.85) hinting that he could benefit from the supply of chances that new signing Rodríguez may conjure up over the coming year. An even cheaper option is Aleksandar Mitrović (Fulham, £6m) who finished top of the Championship in 2019/20 for both goals (26) and xG (25.51), while also managing to score 11 and achieve an xG of 17.12 during his last spell in the top flight back in 2018/19.

The data-driven GW1 XI

(3-4-3): Illan Meslier (Leeds United); Trent Alexander-Arnold (Liverpool), Lucas Digne (Everton), Kyle Walker-Peters (Southampton); Mohammed Salah (Liverpool), Pierre-Emerick Aubameyang (Arsenal), James Ward-Prowse (Southampton), Ryan Fraser (Newcastle United); Timo Werner (Chelsea), Raúl Jiménez (Wolverhampton Wanderers), Aleksandar Mitrović (Fulham).

Substitutes: Ørjan Nyland (Aston Villa), Tyrick Mitchell (Crystal Palace), Yves Bissouma (Brighton & Hove Albion), Charlie Taylor (Burnley).

Our data-driven Fantasy Premier League XI for Gameweek 1.

All the graphics and visualisations in this article use Wyscout data and were produced in the Twenty3 Content Toolbox.

If you’d like to learn more about our products or services, and how they might be able to help you, don’t hesitate to get in touch.