Matchweek 28
🧠 This Week’s Best Bets
TL;DR: Week 10 of the £100→£1000 Challenge. We had another green week last week but still below where we started. Brighton delivered at 3.53 and the West Ham/Bournemouth BTTS No landed in a goalless draw. This week the headline is Crystal Palace Win @ Man Utd (+96.1% EV) - the highest EV figure we’ve logged in several weeks - where Carrick’s fragile bounce meets a Palace side that the model still respects. We’re also backing Everton to beat Newcastle (+23.2% EV) and running the Liverpool double chance with West Ham win and draw covering 40% of outcomes. A sixteenth-Kelly flyer on Sunderland at Bournemouth rounds out the portfolio.
💰 Total Portfolio EV: +36.2% (weighted average) 📊 Bankroll Stake: ~9.1% ⚡
See for yourself: Odds Calculator
Consider following me on X for live-play updates and betting thoughts
📊 Matchweek 27 Review
Another green week. Not a fireworks display - but the bankroll ticked up and the process delivered.
| Bet | Odds | Stake | Result | Outcome |
|---|---|---|---|---|
| 🔥 Brighton Win @ Brentford | 3.53 | £2.50 | Won 2-0 | ✅ WIN |
| 🔥 Everton Win vs Man Utd | 3.56 | £2.50 | Lost 1-0 | ❌ LOSS |
| 💎 Villa Win vs Leeds | 1.83 | £2.00 | Drew 1-1 | ❌ LOSS |
| 💎 West Ham/Bournemouth BTTS No | 2.58 | £1.50 | Won 0-0 | ✅ WIN |
Result: 2/4 hits, +£1.85 profit
The Brighton pick was our model at its purest. Band 1 fixture, 27-point ELO gap, away team systematically underpriced. Brentford 0-2 Brighton. Job done. The BTTS No at West Ham delivered exactly the kind of cagey Nuno-managed 0-0 we expected - West Ham registered 20 shots and an xG of 2.82 and somehow couldn’t score. The market will keep overpricing their attacking output until the results prove otherwise.
The losses are worth unpacking. Man Utd 1-0 Everton - Sesko’s goal on 71 minutes decided it, but Everton had their moments. The Carrick bounce is real, the underlying model was right on the probability, and we’ll be going back to Goodison this week in a different fixture. Villa 1-1 Leeds hurt. Anton Stach’s free-kick from 35 yards was genuinely extraordinary, and Villa needed an Abraham equaliser to salvage a point. Emery’s side dropped 2 points; the model said they should win 60% of the time. They were on the wrong side of 40%.
A brief note of genuine frustration: Man City 2-1 Newcastle would have been a perfect Under 2.5. I had it in the analysis, decided against placing it - and watched three goals go in. The model was right. It goes like that sometimes.
New Bankroll: £95.45 (+£4.70 on the last two weeks, creeping back toward breakeven)
📈 The Bigger Picture
| Week | Starting | Bets | Wins | Staked | P/L | Ending |
|---|---|---|---|---|---|---|
| MW18 | £100.00 | 6 | 2 | £16.00 | +£3.55 | £103.55 |
| MW19 | £103.55 | 6 | 1 | £10.00 | -£6.18 | £97.37 |
| MW20 | £97.37 | 4 | 1 | £11.00 | -£6.62 | £90.75 |
| MW21 | £90.75 | 5 | 2 | £9.72 | +£2.00 | £92.75 |
| MW22 | £92.75 | 6 | 2 | £10.50 | -£0.80 | £91.95 |
| MW23 | £91.95 | 5 | 2 | £9.00 | +£0.50 | £92.45 |
| MW24 | £92.45 | 7 | 2 | £11.86 | -£2.17 | £92.85 |
| MW25 | £92.85 | 5 | 2 | £11.50 | +£3.60 | £96.45 |
| MW26 | £96.45 | 5 | 1 | £7.00 | -£2.85 | £93.60 |
| MW27 | £93.60 | 5 | 2 | £10.00 | +£1.85 | £95.45 |
| Cumulative | - | 54 | 17 | ~£106 | -£4.55 | £95.45 |
Ten weeks in. The drawdown from our £100 starting bankroll has narrowed to just £4.55. We’ve been as high as £103.55 and as low as £90.75. The last four weeks have produced three green results. The process is working; the volume just needs to keep building.
🔥 PREMIUM PICK: Crystal Palace Win @ Man Utd
The Bet: Crystal Palace to win @ 5.36 (18.7% implied) Model Probability: 36.6% (fair odds: 2.73) Expected Value: +96.1% Stake: 1.0% (Sixteenth-Kelly - bankroll cap applied) Confidence: Medium-High
Let’s address the number in the room: +96.1% EV. That is not a typo. This is the highest expected value figure we have calculated across any market in the last several matchweeks, and it demands an explanation.
This is a Band 1 fixture. Man Utd and Crystal Palace are separated by a relatively small ELO gap, meaning this is, by historical standards, a close contest. The model gives Palace a 36.6% chance of winning. The bookmakers are pricing them at 5.36 - implying only an 18.7% probability. That’s a gap of nearly 18 percentage points. For context, our standard threshold for placing a bet is a 5% edge. This is nearly four times that.
How does this happen? Carrick’s bounce. Three wins from three since taking charge - beating Man City, Arsenal away, and Fulham - has the market pricing United as though Fergie just walked back through the door. Our model disagrees. United’s form rating is 4.4/10 with those wins built on outcomes that have outperformed underlying performance. Carrick’s system is possession-based and Fernandes-dependent, but the squad’s defensive resilience when pressed hasn’t been tested by a low-block side yet.
Palace at 5.36 is extraordinary value. We should flag that Palace’s “stable” form reading may flatter their most recent games, which is a fair point - the model may be giving them slightly more credit than a pure eye test would. But even if we apply a generous form discount and say Palace’s true probability is 30% rather than 36.6%, at 5.36 odds that’s still +60% EV. The market is wrong here, and significantly so.
We’re sizing at just 1.0% - well below what the pure Kelly formula would suggest - because the 12% weekly bankroll cap is already doing useful work here, and at 5.36 odds a small stake still provides meaningful upside. A losing bet at 5.36 is expected to happen 63% of the time.
💎 VALUE PICK: Crystal Palace / Man Utd Draw
The Bet: Draw @ 4.15 (24.1% implied) Model Probability: 26.4% (fair odds: 3.79) Expected Value: +9.4% Stake: 1.5% (Half-Kelly) Confidence: Medium
Combined with the Palace Win bet, we now cover 63% of outcomes in this fixture. The only scenario where we lose both bets is a Man Utd win, which the model says happens 37% of the time. That’s the risk, and it’s priced in.
The draw at 9.4% EV is a supporting play, not the main event. It’s there because Band 1 fixtures draw 26% of the time historically, and 4.15 offers thin but genuine edge. If Palace score late and United equalise, we still cash. If it’s a cagey tactical stalemate - highly plausible when a counter-attacking side (Glasner’s Palace) meets a possession team (Carrick’s United) - the draw is the natural outcome.
🔥 PREMIUM PICK: Everton Win vs Newcastle
The Bet: Everton to win @ 4.47 (22.4% implied) Model Probability: 27.5% (fair odds: 3.63) Expected Value: +23.2% Stake: 2.5% (Half-Kelly) Confidence: Medium-High
We backed Everton at home last week to beat Man Utd and lost by a single Sesko goal. The model said 34% then; it says 27.5% now against a different opponent. We’re going again.
Newcastle lost 2-1 to Man City in MW27, which continues a pattern of dropping points at home and away in fixtures where they should be competitive. Their form has been uneven - the 4-1 hammering at Liverpool earlier in the season still haunts the ELO data, and away form has been poor. Goodison Park under the lights in a relegation battle is not where you want to be travelling.
Everton’s 6/10 form is the highest of any team in the bottom half, perhaps more significant than the headline ELO position suggests. Moyes has this squad organised and difficult to beat at home. The argument that Everton’s performance against Man Utd added another 0.1-0.3 form points that aren’t yet reflected in the model is compelling qualitative context - the kind of thing that tips a 23.2% EV bet from “interesting” to “back it.”
At 4.47, we’re getting well paid for a probability the model puts at 27.5%. Half-Kelly applies.
💎 VALUE PICK: Liverpool vs West Ham - Draw & West Ham Win
| West Ham Win: @ 6.04 (16.6% implied) | Model probability: ~19.1% | EV: +15.1% | Stake: 1.5% (Quarter-Kelly) |
| Draw: @ 3.76 (26.6% implied) | Model probability: ~32.2% | EV: +21.1% | Stake: 2.0% (Half-Kelly) |
40% combined probability that either lands, and we’re covering both with separate bets. Liverpool’s form has been uneven - they needed a 97th-minute Mac Allister winner to beat Nottingham Forest in MW27, and Slot acknowledged his side “really struggled in the first half.” West Ham under Nuno sit deep, absorb pressure, and threaten on the counter. The 20-shot, 2.82-xG performance against Bournemouth that somehow produced zero goals tells you the attacking threat is there even if the finishing isn’t.
Liverpool at home are dangerous, but they’re not invincible. We’re not backing West Ham to dominate. We’re backing two outcomes - a cagey draw or a classic Nuno low-block win - that represent 40% of possibilities at combined odds that give us genuine edge on each leg separately.
This is a DIY Double Chance with two individually +EV bets. The correlated risk is explicit and manageable: combined 3.5% exposure across two separate markets.
🎰 LONGSHOT FLYER: Sunderland Win @ Bournemouth
The Bet: Sunderland to win @ 4.23 (23.6% implied) Model Probability: 36.6% (fair odds: 2.73) Expected Value: +54.7% Stake: 0.6% (Sixteenth-Kelly) Confidence: Low
The highest EV figure in the portfolio after Palace - and the one we trust the least. That tension is the whole story with Sunderland right now.
Bournemouth and Sunderland are separated by just 20 ELO points. In pure model terms, this is as close to a coin flip as football gets, and at 4.23 the market is substantially underpricing Sunderland’s chances. That’s the maths. The football tells a different story. Sunderland’s form rating has fallen to 2.8/10 - a declining trend that has seen them lose five games in nine so far in 2026. The young squad that started the season with such belief appears to be hitting the wall that most promoted sides eventually face.
We’re marking this one in the tracker spreadsheet for a reason: this is exactly the kind of data point we want to accumulate over time. Band 1 fixture, high EV, poor form. Does backing the mathematical edge in these situations remain profitable when form divergence is this severe? We don’t have enough data yet. Until we do, the methodology says bet it - but the stake says acknowledge the uncertainty. Sixteenth-Kelly (0.6%) is our way of saying: “the model sees value, but we’re watching you, Sunderland.”
The EV of 54.7% sounds spectacular. The true EV, once form is properly incorporated, is probably considerably lower. Consider this a token of faith in the methodology while we build the dataset to improve it.
❌ Fixtures We’re Passing On
Wolves vs Aston Villa - Villa are the Band 6 stronger team away from home, and the model gives them 60% with +12.2% EV at 1.87. The numbers are sound. But this is a genuinely interesting fixture with compelling qualitative arguments on both sides, and the return on 1.87 doesn’t warrant the allocation when we already have higher-EV plays elsewhere. We’ll mention it without backing it.
Arsenal vs Chelsea - Chelsea at 5.20 with +20.1% EV is mathematically credible. But Arsenal just dismantled Spurs 4-1 with Gyokeres in devastating form, and the Emirate’s after a performance like that is a fortress. Chelsea drew 1-1 with Burnley in MW27. The EV exists; the football context makes it a pass this week.
Fulham vs Spurs - Igor Tudor’s Spurs are in tactical transition following Thomas Frank’s sacking. The first game under any new manager is genuinely difficult to model. We pass on all Spurs markets until the shape stabilises. Besides, as your notes so eloquently put it: “lol.”
Burnley, Leeds, Brighton/Forest - No material +EV opportunities identified across these fixtures worth the allocation.
📋 Portfolio Summary
| Bet | Odds | Model % | EV% | Kelly Stake | Type |
|---|---|---|---|---|---|
| 🔥 Palace Win @ Man Utd | 5.36 | 36.6% | +96.1% | 1.0% (Sixteenth-Kelly) | PREMIUM |
| 🔥 Everton Win vs Newcastle | 4.47 | 27.5% | +23.2% | 2.5% (Half-Kelly) | PREMIUM |
| 💎 Draw @ Man Utd/Palace | 4.15 | 26.4% | +9.4% | 1.5% (Half-Kelly) | VALUE |
| 💎 Liverpool/West Ham Draw | 3.76 | 32.2% | +21.1% | 2.0% (Half-Kelly) | VALUE |
| 💰 West Ham Win @ Liverpool | 6.04 | 19.1% | +15.1% | 1.5% (Quarter-Kelly) | SOLID |
| 🎰 Sunderland Win @ Bournemouth | 4.23 | 36.6% | +54.7% | 0.6% (Sixteenth-Kelly) | LONGSHOT |
| TOTAL | - | - | +36.2% avg | ~9.1% | - |
Bankroll Allocation: ~9.1% on £95.45 (≈ £8.69 at risk) Correlated Risk: Man Utd/Palace has 2.5% combined exposure (Win + Draw). Liverpool/West Ham has 3.5% combined. All other bets are in separate matches.
Final Thoughts
The theme this week is familiar because the market keeps making the same mistake: it prices narrative, not probability. Carrick wins three games and suddenly Crystal Palace are 18.7% shots at Old Trafford - a ground where the model says they should win 36.6% of the time. Everton are priced at 22.4% implied at Goodison against a Newcastle side in inconsistent form, when the model says 27.5%. Liverpool are priced to win comfortably against West Ham, ignoring Nuno’s defensive setup and the fact that they barely scraped past Forest in the 97th minute last week.
We’re not predicting upset after upset for the sake of being contrarian. We’re following the mathematics. The calculator doesn’t watch Carrick’s press conference. It doesn’t factor in Goodison’s atmosphere at 7:45pm on a Tuesday. What it does is look at 2,000+ matches of data and say: “these probabilities are wrong, and by this much.” Our job is to exploit that gap with discipline.
The Palace bet at +96.1% EV will lose most of the time. Over 100 identical bets at these odds and probabilities, you retire. We just need to trust the process and let variance do its thing.
Total Bankroll Exposure: ~9.1% (£8.69 at risk)
Expected Profit: +£3.15 at these edge levels
All probabilities calculated using 5+ seasons of Premier League data (2,000+ matches). ELO bands updated weekly.
Don’t be an idiot. In case you’re feeling like being an idiot today, here is a page wherein I tell you not to gamble your mortgage away. Hey, that rhymes.
Odds correct at time of writing