Almost never. Last season the model's best-projected defender outscored its best-projected midfielder or forward in 8 of 31 gameweeks (Gameweeks 8–38 of 2025/26). On average he scored 4.00 points to their 5.84. Captain the defender every week and you'd have given up 1.84 points a gameweek.
Why it looks closer than it is
Pick the top defender after the fact and he often wins. In 2025/26 the gameweek's highest-scoring defender beat its highest-scoring forward in 25 of 38 gameweeks. Against the highest-scoring midfielder or forward, it was 15 of 38.
That comparison can't help you pick a captain. FPL lists 217 defenders and 79 forwards, nearly three to one, so the best defender is bound to score big somewhere. And you only find out which one it was after the deadline.
The fair test is the one you face before the deadline: the model's best defender against its best attacker, projected in advance.
The fair test
| Season | Gameweeks | Defender outscored attacker | Tied | Defender projected higher |
|---|---|---|---|---|
| 2025/26 | 8–38 | 8 of 31 | 5 | 2 of 31 |
| 2026/27 | 3–5 | 2 of 3 | 0 | 0 of 3 |
The model projected a defender above its best attacker twice last season, and he lost both times. In Gameweek 26 it rated Gabriel 12.91 against Declan Rice's 11.59. Gabriel scored 7 and Rice scored 14. In Gameweek 37 it rated Gabriel 7.59 against Bruno Fernandes' 7.25. Gabriel scored 6 and Bruno scored 9.
This season is running the other way, so far. Gabriel scored 9 to Cole Palmer's 5 in Gameweek 4. James Tarkowski scored 14 to Erling Haaland's 6 in Gameweek 5. That's three gameweeks, and in none of them did the model rate the defender higher.
When it would make sense
The model says don't until a defender's projection passes your best attacker's. Captaincy doubles the gap, it doesn't close it. In the model's 22 September run, covering Gameweeks 6–13, that never happens. The closest is Gameweek 10: Gabriel on 7.22 against Bukayo Saka on 8.17.
Defensive contribution does give defenders a floor. Ten clearances, blocks, interceptions and tackles is worth 2 points, and Tarkowski has hit it in three of his five games this season. But a floor isn't a ceiling. The armband pays for ceilings.
The exception is chasing a mini-league lead late in the season, when you need a big swing and accept a worse average to get one. That's a bet against the odds, taken on purpose. It isn't a better captain.