FPL Differentials for the Final Gameweeks: Who to Pick by Ownership Tier
Differentials win you mini-leagues in the run-in. Here are the best low-owned picks from GW35 to GW38, broken down by ownership band and position.
Jump to section
Data current as of: 2 May 2026 Window covered: GW35 (locked) to GW38 Assumptions: Manchester City and Crystal Palace have a confirmed double in GW36. GW37 and GW38 are single gameweeks across the board. Salah is out for the season.
If you are chasing a green arrow in the final stretch of 2025/26, template players will not get you there. Salah is out. Haaland is owned by half the planet. The mini-league swing now comes from the players sitting between 5% and 30% ownership who happen to be projected for big returns in GW36 to GW38.
This is the differential map for the run-in. Three ownership bands, position by position, with the names that pass the test.
How to think about variance in the run-in
Differential value compounds in two ways at the end of the season. First, the City double in GW36 means owning a non-template City asset (anyone not named Haaland) gives you double the swing for half the ownership cost. Second, GW38 is the most-tweeted, lowest-rotation-protected gameweek of the season, which means floor scores diverge more than usual. A 2-pointer from a template captain and a 12-pointer from a 6% striker is a 20+ point swing on its own.
The trap is taking a differential just because it is low-owned. The right framing is highest projected points per percentage point of ownership. Use the table below as your sort.
Differentials by ownership tier
| Tier | Player | Pos | Price | GW36 proj | Why |
|---|---|---|---|---|---|
| Sub-5% | Hinshelwood (BHA) | MID | £5.1m | 4.44 | xG 0.41, 82% pStart, free hit fodder |
| Sub-5% | F.Kadıoğlu (BHA) | DEF | £4.4m | 4.91 | Cheap clean-sheet enabler at 84% pStart |
| Sub-5% | Truffert (BOU) | DEF | £4.7m | 4.09 | Bench upgrade for free hit and BB |
| 5 to 15% | Cherki (MCI) | MID | £6.5m | 7.93 | DGW36, xA 0.48, 11 assists this season |
| 5 to 15% | Sarr (CRY) | MID | £6.3m | 5.41 | xG 8.85 vs 7 goals, positive regression |
| 5 to 15% | Mateta (CRY) | FWD | £7.6m | 5.35 | Pen #1, xG 14.66 vs 10 goals, home only |
| 5 to 15% | Welbeck (BHA) | FWD | £6.3m | n/a | Pen #1, xG 12.20 vs 13 goals, no flag |
| 15 to 30% | Gibbs-White (NFO) | MID | £7.6m | 4.46 | Form 8.8, 168 pts, 13 goals from mid |
| 15 to 30% | Semenyo (MCI) | MID | £8.1m | 6.99 | DGW36 City asset under Haaland radar |
| 15 to 30% | Bernardo (MCI) | MID | £6.2m | 6.86 | DGW36, 84% pStart, sub-£7m City mid |
Numbers are pulled from the GW36 projections endpoint on 2 May 2026.
Sub-5% picks: free hit and bench-boost fodder
If you are activating Free Hit or Bench Boost in GW36 or GW38, the sub-5% bucket is where you build your bench and your one or two punts. Hinshelwood at £5.1m is the best example on the board: Brighton are a 84% pStart team for him, his xG of 0.41 for GW36 is comparable to mid-priced midfielders, and you can fit him alongside two premiums.
F.Kadıoğlu and Truffert are the bench-boost defenders most managers do not own. Both project for 4+ points at sub-£4.7m. If your bench currently has a £4.0m placeholder defender who plays 0 minutes, you are leaving 4 points on the table per chip activation.
5 to 15%: where the run-in is won
This is the sweet spot. High enough projection to justify a transfer, low enough ownership to actually swing your rank.
Cherki is the headline. He is on £6.5m, has 11 assists for the season, and is the most-projected non-Haaland City asset for GW36 at 7.93 points. That projection includes the City double. His ownership is rising fast (net +13,632 transfers in the latest snapshot per the price-change predictor), so this is a buy-now play.
Mateta and Welbeck are the contrarian forward picks. Both are penalty number ones, both have positive xG-to-goals deltas (Mateta xG 14.66 vs 10 goals, Welbeck xG 12.20 vs 13 goals), and both sit at sub-£8m price points. Mateta has a clear home bias (4.2 pts/game home vs 1.7 away), so check Crystal Palace’s GW37 and GW38 fixtures before pulling the trigger.
Sarr is the one to watch on positive regression alone. He has 8.85 xG against 7 goals scored, which is the kind of gap that resolves itself in a string of returns or never resolves at all. A risk worth 6.3m in a chip week.
15 to 30%: the not-quite-template
These are not “true” differentials, but they are the players the bottom half of mini-leagues do not own. Gibbs-White is the standout: 168 points on the season, form of 8.8, and 13 goals from midfield. If your mini-league rivals are stuck on Palmer (who is bleeding ownership, see our price changes piece), Gibbs-White is the upgrade that closes the gap.
Semenyo and Bernardo Silva are the two City assets you should consider alongside Cherki for GW36. Bernardo at £6.2m with an 84% pStart and 6.86 projected is one of the best price-to-points ratios in the entire game.
How many differentials should you actually own?
Three is the practical maximum if you want to keep template upside. The structure that works for a top-100k push:
- 8 template players (Haaland, Cherki, Bernardo or Semenyo, plus your existing core)
- 3 differentials from the 5-15% band
- 4 enablers and bench
If you are outside the top 1m and chasing a mini-league, push that to 4 or 5 differentials. The variance is your friend, and the season-end swing is short enough that one good week can recover months of mediocrity. We have a longer playbook for that scenario in our rank recovery guide for the final 4 gameweeks.
Chip interaction
If you are still holding a Bench Boost, the sub-5% defenders in the table above are exactly what you want on your bench in GW36 or GW38. The math is simple: a £4.4m defender projected for 4.91 points is worth two of a £4.0m bench placeholder. If you have not played a chip yet, the wildcard chip guide explains why GW35 to GW36 is the last sensible WC window.
What is the one differential to add this week?
Cherki. The City double is in GW36, his projection is the highest of any sub-£7m midfielder, ownership is still under 15%, and the price is rising. Every week you delay costs you 0.1m and a chunk of the swing.
Get differential suggestions filtered to your exact squad and ownership →
Projections, prices, and ownership data pulled from the FPL Pilot production data layer on 2 May 2026. xG and pStart figures reflect the GW36 model run. Verify team news before any chip activation.
PILOT NOTES
One note. Every week.
A single email before deadline. The transfers, captaincy, and chip timing that matter. Nothing else.