FPL 2026/27 · captain record · graded from GW5

Onside Captain Pick Accuracy

Every captain recommendation tracked and scored. Did Onside call it right?
GWs graded
6
Beat template
0%
0 of 6 GWs
Top-1 hit rate
0%
Picked the actual best captain
Avg pts edge
-2.7
vs template captain per GW

2026/27 season · graded from GW5

Onside cumulative captain pts42doubled across all scored GWs
Template captain pts74most-owned captain doubled

Gameweek-by-Gameweek History

GWRecommendedxPActual ptsResult
GW11Haaland(MCI)
6.2 xP
—
Pending
GW10Haaland(MCI)Sealed before the deadline
5.2 xP
2
= Held the templateBest: Alysson (16 pts)
GW9Haaland(MCI)Sealed before the deadline
5.5 xP
2
= Held the templateBest: McBurnie (12 pts)
GW8Rogers(CHE)
4.7 xP
1
✗ Went against it · lostBest: N.Jackson (13 pts)
GW7Haaland(MCI)RevisedRevised to Haaland after the GW7 deadline — the original pick, Semenyo, came from a ranking that over-weighted one week and was then flagged doubtful. Shown, not graded.Originally Semenyo.——Not graded
GW6Schade(BRE)
8.7 xP
1
✗ Went against it · lostBest: Henderson (18 pts)
GW5Palmer(CHE)
8.4 xP
2
✗ Went against it · lostBest: Semenyo (17 pts)
GW4HaalandCalibration—9Not graded
GW3HaalandCalibration—9Not graded
GW2HaalandCalibration—13Not graded
GW1HaalandCalibration—2Not graded

"Actual pts" = raw points scored by the recommended player (not doubled). Scoring runs every Tuesday morning after the previous GW is finalised. Gameweeks 1–4 ran while the engine was being calibrated week to week; they show that week's template captain and are not in the season figures.

How accuracy is measured

A captain pick is marked “Beat template” if the recommended player outscored FPL's most-captained player that gameweek (the armband the field actually wore, from FPL's own gameweek data; until 1 Nov 2026 this used the most-owned attacker, which named João Pedro for GW3 and GW4 while Haaland was the most-captained both weeks). Each week is labelled as the call it was: held the template, or went against it and won or lost. This measures edge over the consensus — beating the average manager's choice — rather than simply predicting the single top scorer. The “Top-1 hit rate” tracks the harder challenge: did Onside pick the actual highest-scoring captain that week?

Methodology follows prequential evaluation (Dawid 1984): every prediction is logged before the outcome is known, and scores are applied only after the GW is finalised by FPL. No historical predictions are fabricated. The record starts from the moment the tracker was deployed.

Frequently Asked Questions

How does Onside pick the FPL captain each week?▾

The engine blends nine independent signals into an expected-points score. The top captain is selected by highest median captain EV — doubling expected value — from a Monte Carlo simulation. From Gameweek 9 the tracked pick is sealed about an hour before each deadline, after the latest team news; earlier weeks were sealed on the Tuesday.

How accurate is Onside's captain recommendation?▾

The track record is shown above. The engine targets consistent outperformance of the average FPL manager's captain choice for that gameweek. This means beating the consensus, not just predicting the top scorer — a higher bar that actually moves the needle on your rank.

Where can I see this week's captain pick?▾

The dashboard Captain card shows the current gameweek recommendation. You can also visit the /captain/[GW] pages for a full breakdown of each week — top pick, alternatives, risk profile, and confidence margins.

What data does Onside use for captain predictions?▾

Nine independent signals spanning form, fixtures, market and availability. How each is built and weighted is proprietary — see /about-the-engine.

Explore more