How the Mini-League Simulator Works: 100,000 Seasons, Explained

What "Monte Carlo" means

Some questions are too tangled to answer with a formula. "Who wins this league?" depends on every manager's skill, form, luck, remaining chips, and how similar their squads are - all interacting. The Monte Carlo approach sidesteps the formula entirely: simulate the whole thing with realistic randomness, thousands of times, and count the outcomes. If you finish first in 31,400 of 100,000 simulated seasons, your win probability is 31.4%. It's the same technique used in weather forecasting and election modelling.

Step 1: a scoring profile for every manager

For each manager in your league (up to 100), the simulator builds a profile from their actual gameweek history:

• Skill level - their average points per gameweek across the whole season. A manager averaging 58 doesn't suddenly become a 70-points-a-week manager in the simulations.
• Volatility - how much their scores swing week to week, weighted 70% towards their last six gameweeks and 30% towards the full season. A steady template-hugger gets a narrow distribution; a punty manager on a hot streak gets a wide one. Volatility is what turns "20 points behind" from hopeless into a comeback probability.

Each remaining gameweek score is then drawn from that profile - so a simulated season is a plausible sequence of scores for that specific manager, not a coin flip.

Step 2: managers with similar squads move together

This is the detail we're proudest of. If you and your rival both own Haaland, Salah and Saka, your scores aren't independent - a Salah masterclass lifts you both and changes nothing between you. Treating managers as independent wildly overstates how much shuffling is left in a league.

So the simulator splits each manager's weekly score into two parts: a shared component, driven by the league's template - the same draw for everyone in that simulated season - and an idiosyncratic component for their differentials. How the split is weighted comes from measuring how much of each manager's starting XI is owned across the rest of the league. Heavy template squads ride the shared draw; differential-heavy squads live and die by their own.

This is why the tool's advice often boils down to a classic FPL truth, now with maths attached: if you're behind, you cannot catch a rival by owning their team. The simulations literally can't produce a comeback without differentials somewhere.

Step 3: chips still in hand count

The simulator checks which chips every manager has already played and credits the remaining ones with a conservative expected bonus, each with realistic spread: a Free Hit is worth about 12 extra points on average in the simulations, a Bench Boost about 10, a Wildcard or Triple Captain about 8. (Deliberately modest - most managers play late-season chips imperfectly, and the occasional 25-point Bench Boost is balanced by plenty of duds.)

The effect: if you're 15 points clear but your closest rival holds a Free Hit and Bench Boost you've already spent, your lead is smaller than it looks - and your win probability will say so.

What the numbers mean

• Win % / Top 3 % / Top 5 % - the share of the 100,000 simulated seasons where you finished there. Bigger leagues also get a Top 10 %.
• Projected points - the median of your simulated final totals, shown with a 90% confidence range (5th to 95th percentile). The range is the honest part: early in the season it's wide, and it should be.
• Projected finish - your average finishing position across all simulations, plus the best and worst finish that appeared at all.
• Your persona - a strategic read on your position: a Guardian defends a likely win, a Predator is in the hunt, a Disruptor is near-certain of a podium but needs swings to win, a Climber has an outside shot, and a Wildcard has nothing to lose - which is its own kind of freedom.

What it doesn't know

Honest limits, so you can weigh the output properly: the simulator assumes managers keep being roughly who they've been - it can't see that your rival is about to rage-quit, or that you're planning a heroic triple-differential punt. It doesn't predict specific fixtures (that's what our Predicted Points model does); it models score distributions. And probabilities refresh every gameweek as real results replace simulated ones - a 60% chance in March is a different claim than a 60% chance in May.

FAQs

How does the FPL mini-league simulator work?

It plays out the rest of the season 100,000 times. In each simulated season, every manager's remaining gameweek scores are drawn from a distribution built on their season-long scoring average and their recent volatility, with squads that overlap the league template moving together and differentials adding independent variance. Remaining chips add an expected bonus. Counting how often each manager finishes 1st, top 3, and so on across all 100,000 seasons gives the probabilities you see.

What does a 70% chance of winning actually mean?

It means that in 70,000 of the 100,000 simulated seasons, you finished first. It does not mean you will win - 3 times in 10 you don't. Probabilities quantify uncertainty; they don't remove it. A 70% favourite losing is not the simulator being wrong, it's the 30% happening.

Why do my rival and I move together in the simulations?

Because your squads overlap. If you both own Haaland and Salah, a Haaland haul lifts you both and changes nothing between you. The simulator measures how much of each manager's starting XI is owned across the rest of the league and correlates their simulated scores accordingly - which is exactly why differentials, not template picks, are how you close a gap.

Does the simulator account for chips?

Yes. It reads which chips each manager has already played and credits the ones still in hand with a conservative expected bonus - a Free Hit averages around 12 extra points in the simulations, a Bench Boost around 10, a Wildcard or Triple Captain around 8, each with a realistic spread. A rival holding three chips when you have none is a real disadvantage the numbers will show.

How accurate are the projected final points?

The projection is shown with a 90% confidence range (the 5th to 95th percentile of your 100,000 simulated totals) rather than a single number, because a single number would be false precision. Early in the season the range is wide; as gameweeks run out it tightens. The probabilities refresh every gameweek as real results replace simulated ones.

Is the simulator free?

Every user gets one league simulation per gameweek free. Pulse Pro (£2.99/month or £24.99/year, 7-day free trial) unlocks unlimited simulations plus AI-generated advice on whether to attack or defend based on your specific position.