How SmartPlay picks a captain — and why your mini-league changes the answer
Most captain advice hands everyone the same name. The right pick depends on expected points and where you sit in your mini-league. Here's how the model reasons about both.
Every gameweek, the same thing happens on FPL Twitter: a dozen accounts post the same captain. Sometimes they're right. But "who has the highest ceiling this week" is only half the question. The other half is who you're actually trying to beat — and that half is the one nobody else answers.
Here's how SmartPlay thinks about it.
The short version
- The highest-xPts pick is the baseline captain, not automatically the right one
- Your mini-league position — protect or chase — reshapes the call
- League context can break a tie between close options; it can't turn a weak pick into a good one
What "expected points" actually means
The model assigns every player an expected points (xPts) value for the upcoming gameweek — its best estimate of how many FPL points they'll score, built from minutes, shot and chance quality, fixture strength, and appearance likelihood. The current model was trained on four completed seasons, from 2020-21 through 2023-24. The 2024-25 season was used for validation, while 2025-26 remained held out for evaluation.
xPts is a ranking and risk signal, not a crystal ball. It won't tell you a striker will score exactly 9 points on Saturday. It will tell you, reliably, that one option out-ranks another — and by how much.
The captain shortlist
Left alone, the model produces a simple shortlist: the players with the highest xPts for the week, adjusted for how nailed-on their minutes are. The first name is the highest-xPts baseline, not automatically the right choice for every objective.
If your objective is overall rank, that baseline keeps the decision focused on expected points rather than rival ownership.
The top bar is the highest-xPts baseline. How far clear it is tells you how much a differential armband is really costing you.
Why your mini-league flips the call
But you're usually not playing the overall ranks — you're playing the twelve people in your office league. And that changes everything, because a captain choice is really a bet about variance relative to your rivals.
Protecting a lead
- Favour the popular, well-supported captain
- Matching rivals removes variance
- A widely shared haul doesn't cost you ground
Chasing from behind
- A lower-owned captain with model support
- More separation when it lands
- More downside if the popular pick hauls
Same gameweek, same xPts, different risk trade-offs. League position may change the recommendation when the candidates are close; it does not turn a weak option into a good one.
Putting it together
SmartPlay does three things at once: ranks every option by xPts, checks who your specific rivals own, and then reshapes the recommendation around your chosen mode — Protect, Balanced, or Chase. You still make the call; the model just makes sure you're making it with the full picture instead of a timeline consensus.
If you want the mechanics — how xPts, fixture ratings, and nailedness are computed — it's all documented, no black box. Read how the model works.
The strongest captain for your league context is not always the most popular name on the timeline. Pick for the objective you're actually playing.
Keep reading
Expected points without the false precision
xPts is most useful as a comparison tool, not a promise. Here is how to read the number, the gap between players, and the uncertainty underneath.
When is a transfer actually worth four points?
A hit is not simply incoming xPts minus four. The real comparison includes the player you sell, the planning horizon, captaincy, and the value of waiting.
Template, differential, or trap?
Ownership changes the shape of your rank swings, not the quality of a player. Use it after expected points and minutes, never in place of them.