With 5 games left in the 38-game Premier League season, the CURSD model projects how every team finishes if their underlying performance continues. The model blends 40% actual points-per-game with 60% expected points-per-game (xPPG), then breaks ties on projected goal difference, the same way the Premier League itself does. The chart below shows projected final points for all 20 teams. The full table follows underneath, with both projected goals scored and projected goals conceded.
The Title Race
Arsenal and Manchester City both finish on 80 points. The title is decided on projected goal difference, the league's first tiebreaker: Arsenal +42.1 vs Manchester City +41.6. Arsenal edges it by 0.6 goals (the table rounds both to 42, but the model carries the fractions to keep the tiebreak honest).
Why Arsenal? Their xPPG is 1.98 vs Manchester City's 1.85 — the underlying performance gap. The standings have them level, but the underlying chance creation and chance suppression don't. Arsenal sit at 63 goals scored on 56.6 xG and 26 conceded on 24.8 xGA — finishing and defense both tracking close to expectation. Manchester City sit at 66 scored on 61.7 xG and 29 conceded on 36.2 xGA — that 7.2-goal gap means their defense has been overperforming, conceding fewer than the chances allowed would predict. The model expects regression in the final 5 games. That is how the title separates by 0.6 goals on the projected GD tiebreaker.
Top 4 — Champions League
The model puts Arsenal, Manchester City, Manchester United, Aston Villa in the Champions League places. Tight race for 4th: Aston Villa edges out Liverpool by 2 projected points. The most luck-driven side in the top 6 by xPPG vs PPG gap: Aston Villa (running at +0.48 pts/game above expected). If finishing variance evens out, this is the side the model expects to fade fastest.
Movers in the Final Stretch
Climbing: Brentford (9 → 7, ↑2), Crystal Palace (13 → 11, ↑2).
Falling: Sunderland (11 → 13, ↓2), Bournemouth (7 → 8, ↓1), Chelsea (8 → 9, ↓1).
These moves come almost entirely from xPPG vs PPG divergence. A team currently outperforming the underlying numbers is expected to slow down; a team underperforming is expected to pick up. Five games is a small sample, so most table positions are locked in — but the ones that move tell you which sides have been riding luck.
Relegation
The drop zone projects to Tottenham, Burnley, Wolves. The 18th-place fight is alive: Tottenham (36) and West Ham (38) are within 2 points.
Full Projected Table
| # | Team | Played | Now | xPPG | Proj. Pts | Proj. GS | Proj. GA | Proj. GD | Move |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Arsenal | 33 | 70 | 1.98 | 80 | 72 | 30 | +42 | — |
| 2 | Manchester City | 33 | 70 | 1.85 | 80 | 76 | 34 | +42 | — |
| 3 | Manchester United | 33 | 58 | 1.57 | 66 | 66 | 52 | +15 | — |
| 4 | Aston Villa | 33 | 58 | 1.28 | 65 | 54 | 48 | +6 | — |
| 5 | Liverpool | 33 | 55 | 1.66 | 63 | 62 | 49 | +13 | — |
| 6 | Brighton | 34 | 50 | 1.50 | 56 | 54 | 44 | +10 | — |
| 7 | Brentford | 33 | 48 | 1.50 | 55 | 56 | 51 | +5 | ↑2 |
| 8 | Bournemouth | 34 | 49 | 1.50 | 55 | 58 | 58 | 0 | ↓1 |
| 9 | Chelsea | 34 | 48 | 1.61 | 54 | 60 | 50 | +10 | ↓1 |
| 10 | Everton | 33 | 47 | 1.20 | 53 | 46 | 46 | 0 | — |
| 11 | Crystal Palace | 32 | 43 | 1.50 | 52 | 43 | 43 | 0 | ↑2 |
| 12 | Fulham | 33 | 45 | 1.26 | 52 | 49 | 53 | -4 | — |
| 13 | Sunderland | 33 | 46 | 1.09 | 52 | 41 | 47 | -6 | ↓2 |
| 14 | Newcastle | 33 | 42 | 1.44 | 49 | 53 | 56 | -3 | — |
| 15 | Leeds | 34 | 40 | 1.44 | 45 | 50 | 57 | -7 | — |
| 16 | Nottingham Forest | 33 | 36 | 1.21 | 42 | 42 | 52 | -10 | — |
| 17 | West Ham | 33 | 33 | 1.16 | 38 | 46 | 65 | -19 | — |
| 18 | Tottenham | 33 | 31 | 1.13 | 36 | 48 | 61 | -13 | — |
| 19 | Burnley | 34 | 20 | 0.72 | 23 | 38 | 76 | -38 | — |
| 20 | Wolves | 33 | 17 | 0.98 | 21 | 28 | 69 | -41 | — |
How the Projection Works
The model is deliberately simple. For each team:
1. Compute actual points-per-game (PPG) and expected points-per-game (xPPG) from xPTS. 2. Blend them: 40% PPG + 60% xPPG. Heavier weight on xPPG because it strips out finishing variance, defensive luck, and close-game randomness — the things CURSD already measures and treats as luck. 3. Apply that blended rate to the remaining games to project final points. 4. Apply the same blend logic to goals scored (using xG) and goals against (using xGA) to project goal difference, the league's standard tiebreaker.
No simulations, no opponent-strength adjustments. Just current pace anchored to underlying performance. The point is to surface where each team should finish if luck stops mattering — not to pretend we know what happens in any individual match.
What This Isn't
This is not a prediction. It's a baseline. Five games leaves room for swings — a 9-point swing is statistically possible for any team. What this projection says is: here's what the data implies if performance continues. Where the projected position differs from the current position, the model is telling you the underlying numbers and the standings disagree. That gap is what CURSD measures every day across nine leagues.
For the live, daily-updated CURSD scores, see /league/epl. For FPL implications of the same finishing variance, see /fpl-watch.
