Identification of skill in an online game

A game built on uncertainty

Every fantasy result begins with another contest. A defender can lose a clean sheet to a late deflection. A captain can leave injured. Weather, form, selection, and chance all enter the score. That uncertainty is part of the appeal, but it makes skill hard to see in a single gameweek.

Fantasy Premier League adds its own layer of decisions. Managers begin with a fixed budget, select fifteen players, choose a starting eleven, name a captain, and manage transfers across thirty-eight gameweeks. Player prices move. Fixtures are postponed. A few weeks contain fewer matches, while others contain twice the opportunity.

The right question is not whether luck exists. It plainly does. The question is whether good decisions leave a stable trace once the study looks across enough managers, weeks, and seasons.

Skill appears in repeat performance

The clearest evidence comes from history. Managers who score well in one season are more likely to score well in another. For the 2017/18 and 2018/19 seasons, the paper finds a correlation of 0.42 across roughly three million returning managers. The same pattern remains visible in comparisons stretching back more than a decade.

Experience matters too. A regression on 2018/19 performance associates each previous season played with 22.1 additional points. The model explains only part of the final score, so experience is not destiny. Still, the direction is consistent: practiced managers tend to do better.

This is what skill looks like inside a noisy game. It does not remove bad weeks. It shifts the distribution of outcomes over time.

Better decisions, repeated weekly

To understand where the advantage comes from, the researchers divide managers into ranked tiers. The strongest tier does not win through one extraordinary week. It outperforms the lower tiers throughout the season, including the opening gameweek, when preparation matters most.

Transfers and captaincy

After a transfer, top managers earn more from the player they buy relative to the one they sell. When each choice is compared with the affordable alternatives, higher-ranked managers are also more likely to have selected one of the better options. Their captain choices produce more points across the season as well.

Team value

Strong managers build squad value earlier. That matters because a larger budget expands the set of future transfers. At gameweek 19, each extra £1 million in team value is associated with 21.8 additional points by the end of the season. The relationship is useful but incomplete. Team value supports good decisions; it does not replace them.

Planning creates the moment

Blank and double gameweeks turn the fixture calendar into a strategic problem. In a blank week, some clubs do not play. In a double week, some play twice. Managers can use single-use chips to benefit from these disruptions, but only if their squads are ready.

The Bench Boost result makes the difference visible. In double gameweek 35, 79.4% of top-10,000 managers used the chip, compared with 28.9% of the rest of the dataset. The top group also earned more from it: 23.2 points on average, against 13.8.

Waiting was only the final action. The larger advantage came from planning transfers, preserving the chip, and assembling fifteen useful players before the opportunity arrived.

The template team

Better managers do not always succeed by being different. Their squads are, on average, more similar to one another than those of lower-ranked managers. A small group of widely owned players forms a temporary template, then expands or dissolves as fixtures, prices, injuries, and form change.

At its largest, the three core clusters contain 32 of the 624 available players. The similarity between teams rises and falls across the season, which suggests active convergence rather than a fixed set-and-forget squad.

Herding can be careless, but consensus can also be informed. Skilled managers often agree on the strongest core. Their remaining edge comes from timing, captaincy, and knowing when the shared answer has changed.

Skill inside luck

The paper does not claim that FPL is predictable. Its stronger conclusion is that good management survives uncertainty. Rank carries across seasons. Higher tiers gain more from recurring choices. Their preparation becomes most visible when the calendar creates a scarce opportunity.

The data also gives a more useful definition of fantasy skill. It is not the ability to name every scorer. It is the ability to make sound decisions with incomplete information, preserve flexibility, and repeat that process long enough for a small edge to matter.