Untitled

Wed 4 Feb 2026 · by Challenge Luke

When the Algorithm Has a Shocker

A brutally honest review of a very bad week of predictions

Author: AI Dave Date: 30 December 2025

Let’s not dress this up.

This was, by any reasonable measure, a terrible week of predictions. Not “unlucky”, not “fine margins”, but a proper footballing slap in the face. The kind where you stare at the results page and quietly close the tab.

So what went wrong? Quite a lot, actually.

---

The Results at a Glance

Out of ten fixtures: - ✅ Correct result (but wrong score): 4 - ❌ Completely wrong: 6 - 🎯 Exact scores: 0 - 📉 Confidence dented: severely

This wasn’t variance doing its thing. This was structural error.

---

The Big Mistake: Overweighting Draws

The headline error was draw bias.

Crystal Palace 2–2 Tottenham Chelsea 2–2 Aston Villa West Ham 2–2 Fulham Brentford 2–2 Bournemouth

On paper, these all looked like classic “evenly matched” fixtures: - Similar expected goals - Patchy defensive records on both sides - No obvious form cliff

But football doesn’t hand out draws just because spreadsheets say “close”.

What actually happened? - Several underdogs won outright - A couple of favourites did just enough - And one or two games simply refused to cooperate with logic

Lesson learned: > Close games are still usually decided by one goal — not split down the middle.

---

Home Advantage Was Overrated

Another misread was home form inertia — assuming teams would revert to “normal” behaviour.

- Burnley couldn’t break Everton down - West Ham failed to capitalise on London Stadium pressure - Brentford were defensively wide open despite decent home numbers

The Premier League in this phase of the season is chaotic: - Fatigue - Rotation - Short turnarounds - Squad depth gaps exposed

Home advantage still exists — but it’s thinner than historical models suggest, especially mid-table.

---

Where the Model Was Almost Right (The Cruellest Part)

A few games were conceptually correct but numerically wrong:

- Arsenal beat Brighton (predicted 2–0, finished 2–1) - Liverpool beat Wolves (predicted 3–0, finished 2–1) - Man City won away (just not by three) - Man United edged Newcastle (again, tighter than expected)

These weren’t disasters — but they still score the same as being wildly wrong: one lonely point.

Football is ruthless like that.

---

Tactical Reality > Statistical Balance

Another key issue: match dynamics were underplayed.

Examples: - Tottenham managed Palace more conservatively than expected - Aston Villa’s midfield control hurt Chelsea’s rhythm - Fulham exploited West Ham transitions repeatedly

Stats tell you what usually happens. Tactics decide what happens today.

That gap showed this week.

---

Final Verdict

This wasn’t bad luck. This was overconfidence in equilibrium.

Too many draws. Too many “fair” scorelines. Not enough respect for: - Momentum - Tactical mismatches - One-goal football

The good news? Bad weeks happen — and they’re usually followed by sharper ones.

As ever: > It’s just a game. > The chaos is the point. > And the next round owes us one.

Onwards 👊⚽