Why The Algorithm Doesn’t Believe in “Form”

Each season, the same debate resurfaces: should managers trust “form,” or is it an illusion? This week, The Algorithm recommended selling Haaland, and many people have called it crazy because he’s been scoring recently. But to understand why that recommendation makes sense, it’s important to look at what “form” really represents, and what it doesn’t.

What “Form” Actually Means

In football discussions, “form” is often used to describe a player who has produced strong returns over the last few games. It’s shorthand for saying someone is performing well. But statistically, that’s a description of past outcomes, not a signal of what’s likely to happen next.

When a player scores several goals in a short period, it’s tempting to believe they’ve entered a new state of performance or that they’re suddenly “hot.” In reality, these streaks are often just variance: normal fluctuations that occur in a low-scoring, high-variance sport like football.

Finishing is one of the noisiest parts of the game. A striker might score twice from two shots one week and blank from six the next. Those swings balance out over time. Once you control for the quality of chances (expected goals), team strength, and role, the apparent “form” adds little to no predictive value for what happens in future matches.

What the Data Says

Multiple academic studies and professional analytics teams have examined whether recent goals make a player more likely to score again. The consistent finding is that “form” has almost no independent predictive power once you adjust for xG, xA, and fixtures.

A player’s finishing streaks are mostly random noise. The only time “form” becomes meaningful is when it reflects an underlying change such as a tactical shift, a new role, or an improvement in the team’s attacking process. In those cases, the data already captures it: xG and shot volume rise, and the model naturally adjusts.

This is why predictive systems like The Algorithm are designed to focus on underlying indicators, not short-term outcomes. They care about what creates points, not what’s already been recorded.

Why The Algorithm Sold Haaland

Haaland remains one of the best forwards in world football. But even elite players follow the same statistical rules as everyone else.

If the data shows that:

• his upcoming fixtures are more difficult,

• his expected goal involvement is projected to decline relative to other players, or

• the same budget could deliver higher expected value across multiple positions,

then the optimal move from a probabilistic standpoint is to sell him.

That decision isn’t emotional, nor is it based on the outcome of his last few games. It’s based on expected value, the statistical forecast of what’s most likely to produce more points going forward.

This Doesn’t Mean Haaland Won’t Score

It’s important to be clear: selling Haaland doesn’t mean he’s suddenly a poor pick or that he won’t continue to score goals. He’s still an elite finisher in one of the most attacking teams in the world.

But Fantasy Premier League isn’t about judging players in isolation, it’s about how we allocate limited funds to maximise total points across fifteen slots.

The Algorithm doesn’t ask, “Will Haaland score?” It asks, “Given his price and projected output, is there a better way to spend £14.5m?”

Sometimes, the answer is yes. A collection of mid-priced midfielders with favourable fixtures might collectively produce a higher expected return than Haaland plus a budget option elsewhere. In that case, the data shows that spreading the funds creates more long-term value even if Haaland himself performs well.

This is one of the hardest ideas for managers to accept: you can make the right decision and still see the player you sold score. That doesn’t make the decision wrong; it means you played the probabilities correctly.

The goal isn’t to predict individual outcomes it’s to make decisions that are statistically optimal over the long run. That’s the entire philosophy behind The Algorithm.

Emotion vs. Expectation

It’s easy to understand why people react strongly to recent results. Goals are emotional and memorable; they make it feel like a player has momentum. But those moments aren’t reliable predictors of what happens next.

Human decision-making is heavily influenced by recency bias, the tendency to overvalue recent events and undervalue long-term data. The Algorithm is designed specifically to avoid that trap.

Where most managers react to outcomes, The Algorithm reacts to probabilities. It focuses on the variables that consistently drive future points, expected goals, expected assists, fixture difficulty, and minutes, rather than the emotional comfort of keeping a player who just scored.

The Bigger Picture

When people say, “You can’t sell Haaland, he’s in form,” what they usually mean is that it feels uncomfortable to go against recent results. But data shows that those runs don’t predict the future; they just describe the past.

“Form” feels real because humans are wired to find patterns, even in randomness. Predictive models like The Algorithm are built to separate signal from noise.

It might not always align with intuition in the short term, but that’s exactly why it works over time and why it beats most human managers.

In Summary

• “Form” describes recent outcomes, not future probabilities.

• Finishing streaks are mostly variance, not a change in ability.

• Predictive power comes from underlying stats like xG, not recent goals.

• The Algorithm bases decisions on expected value, not emotion.

• Selling a player in “good form” is often about optimising budget, not doubting their quality.

Join KnightManagers

If you want your FPL decisions to be driven by data, not emotion, you can join hundreds of managers already using The Algorithm. Each week it analyses your specific team, suggests transfer options, chip strategies, and captain choices — all based on expected value, not guesswork.

👉 Learn more and join The Algorithm here.