Should you always captain the top-projected player?

Rank the candidates by expected points and the actual points fall almost step by step: the top-ranked captain averaged 7.3 a week, the 5th 4.8, the 10th 3.8. Picking at random from the 20 would have averaged 4.4.

Against the second-ranked player the top pick gained 0.9 points a week, about 33 a season. Four seasons can’t pin that down: allowing for chance, the true edge could be anything from -14 to 80 points a season, so it could even be nothing. In 2023/24 the second choice beat the first by 32 points.

The top-ranked captain was the week’s best choice only 20% of the time. He blanked in 28% of weeks and hauled (10 or more) in 29%. Hindsight captains averaged 14.4, twice what any rule managed, so almost every week someone you passed over scores more.

That isn’t evidence the pick was wrong. One player’s score in one match is mostly noise: a goal is worth 4 to 6 points, and even the best forwards score in under half their games.

Two captains can score the same over a season in very different ways. A reliable one returns most weeks; an explosive one blanks often and makes it up with the odd huge score. Explosive points only reach you if your armband was on him that week.

This season shows the difference. Through gameweek 7, Haaland has scored 2, 13, 9, 9, 6, 6, 13: 58 points, one blank. Bruno Fernandes has scored 2, 23, 2, 2, 2, 1, 1: 33 points, 70% of them in one week. Captaining Fernandes every week would have earned 66 captain points against 116 for Haaland, and any week you picked him except that one earned you 4 or fewer.

Over a full season, though, the lumps even out. Among the most-owned premium options of the last four seasons, the best two weeks were only 12% to 19% of a season’s points. Fernandes himself blanked in under a quarter of his games in 2025/26. What separates the good captains is how often they blank, not how big their best week is.

Do blanks come in runs? Barely. Players priced £9m or more who played an hour blanked 35% of the time. After a blank it was 38% (192 cases); after two in a row 40%, from only 55 cases; after a return, 33%. A run of blanks is a reason to check his minutes and chances, not a forecast by itself.

How far the top pick was ahead decided whether ranking him first meant much. With less than a point between the top two, the top pick won the head-to-head about as often as he lost it. With two points or more between them he won 62% and scored 2.6 more on average.

The top-ranked player was also the most-owned only 33% of the time. In the other weeks he scored 1.0 more on average, and over the whole period about 24 points a season more than always captaining the most-owned player.

Captaining the crowd’s player protects your rank when he hauls; it doesn’t score more. And the margins are small next to the noise: in 65 of the 141 weeks the top two were less than a point apart, while our simulations put a whole team’s chance variation at about 15 points a week.