How scoring works.

The Brier Score is how we measure the accuracy of your predictions. Think of it as a report card for each forecast you make. It measures how far your prediction was from what actually happened.

Here's the simple version: you give a prediction between 0 and 100. That number gets converted to a decimal (so 75 becomes 0.75). After the match, the outcome is either YES (1.0) or NO (0.0). Your Brier Score is the squared distance between your prediction and the outcome.

Example:

The key thing to remember: the Brier Score measures error, so lower is better. A score of 0 means you were perfect. A score of 1 means you were as wrong as possible.

It's natural to think 'higher = better' since that's how most scoring works. But the Brier Score measures how far off you were — it's an error score. Think of it like golf: the fewer strokes (errors), the better. A Brier Score of 0.06 means you were barely wrong. A Brier Score of 0.56 means you were way off.

You don't earn points from your raw Brier Score. Instead, you earn Relative Brier Points by being more accurate than the other participants. Here's how it works:

After each question, we calculate the field average Brier Score — the average of everyone's Brier Scores. Then we subtract your Brier Score from that average. If your score is lower (better) than the average, the difference is positive and you earn points. If your score is higher (worse), you lose points.

Your cumulative total of these points across all questions determines your rank on the leaderboard.

Short answer: because if you're wrong, the penalty is catastrophic, and even if you're right, you barely benefit versus predicting, say, 90.

Let's do the math. Say you predict 100 (= 1.0):

Now compare that to predicting 90 (= 0.90):

By predicting 90 instead of 100, you give up almost nothing when you're right (0.01 vs 0.00), but you save yourself significantly when you're wrong (0.81 vs 1.00). The scoring system mathematically rewards you for expressing your actual confidence level, not for exaggerating it. This is called being 'well-calibrated.'

The bottom line: predict 100 only if you are literally certain — like, the match has already been played and you know the result. Otherwise, always leave yourself some room.

No. The Brier Score is what mathematicians call a 'proper scoring rule.' This is a technical term that means the scoring system is specifically designed so that your best strategy is always to report your true belief. Any attempt to be clever — predicting more extreme numbers to 'win bigger,' hedging toward 50 to play it safe, or copying other people's predictions — will, on average, give you a worse score than just being honest.

Here's why copying doesn't work: if you match the field average on every question, your edge is always zero. You'd never earn a single point. You have to be different from the crowd AND more accurate to gain an advantage.

And hedging toward 50 on everything is equally bad — you'd be saying 'I have no idea' on every question, and anyone with actual knowledge will consistently outperform you.

The only winning strategy is genuine accuracy. Do your research, form an honest opinion, and submit it.

This is one of the most important strategic decisions in the contest, and the answer is nuanced.

The case for being selective: if you only answer questions where you have a genuine edge (strong knowledge, good research, a clear opinion), your average points per question will be high. You avoid the risk of losing points on questions where you'd just be guessing.

The case for answering everything: every question is an opportunity to earn points. If you skip a question where you would have beaten the field average, that's points left on the table. And more questions means more chances for your skill to show through — one bad prediction gets diluted across many.

Here's a practical rule of thumb: if you have any opinion at all, submit it. The field average often reflects a sort of consensus — if you have information or insight that others don't, even a slightly better prediction earns you points. The only time to skip a question is if you truly have zero knowledge and would literally be guessing at random.

Remember: the leaderboard ranks by cumulative points, not average points. A player with +20 points from 30 questions beats a player with +18 points from 5 questions, even though the second player has a higher average.

Predicting 50 on everything is like saying 'I have absolutely no idea' for every question. It's the safest possible strategy — but it won't win you the contest.

If the field average is close to 50, your edge will be near zero on every question. You'll hover around the middle of the leaderboard, never gaining or losing much. But if the field is generally better calibrated than a coin flip (and they likely will be), you'll slowly bleed points.

They beat you by a wide margin.

The 50-everywhere strategy guarantees mediocrity. To win, you need to have conviction when you have information.

As the World Cup progresses, the points you earn (or lose) are multiplied:

• Group Stage: 1× (base value) • Elimination Rounds: 2× • Final: 3×

This means a prediction in the Final is worth three times as much as the same prediction in the Group Stage. If your base edge would earn you +5.0 points, the 3× multiplier makes it +15.0.

Critical: multipliers apply equally to losses. A base loss of −5.0 becomes −15.0 in the Final. The stakes genuinely rise as the tournament progresses. This is by design — it keeps the contest exciting and means that late-stage predictions can dramatically reshape the leaderboard.

Yes, absolutely. If your Brier Score on a question is worse than the field average, your edge is negative and you lose points. This is a fundamental part of the contest — it's not enough to make predictions, you have to make predictions that are better than the crowd's.

For example:

This is why blind guessing is risky. If you submit a prediction on a topic you know nothing about, there's a good chance the crowd (which includes some knowledgeable people) will outperform you, and you'll lose points.

These are two different things that people sometimes confuse:

Brier Score: how accurate your individual prediction was, independent of anyone else. It ranges from 0 (perfect) to 1 (worst possible). This is calculated per question.

Relative Brier Points: how your accuracy compared to the field average, converted to a points format. This is what appears on the leaderboard. Positive means you beat the crowd; negative means the crowd beat you.

Think of it this way: a Brier Score of 0.15 might be great or terrible depending on the question. If everyone else scored 0.40, your 0.15 is excellent (+25 points). But if everyone else scored 0.05, your 0.15 is actually poor (−10 points). The Relative Brier Points capture this context.

Relative scoring solves a fairness problem. Some questions are inherently easier to predict than others. If a match is between the world's #1 team and a huge underdog, most people will get it roughly right, and the raw Brier Scores will all be low. If we just ranked raw scores, easy questions and hard questions would carry the same weight.

Relative scoring means you only earn points by being better than the crowd on that specific question. On an easy question where everyone gets it right, there are few points available. On a hard question where the crowd is divided, a confident correct prediction earns big points. This rewards genuine insight over picking obvious outcomes.

Yes! You can update your prediction as many times as you want, right up until the match starts. Only your final prediction (the last one you submitted before the deadline) counts for scoring. There is no penalty for changing your mind.

In fact, updating is encouraged. If new information comes in — a key player gets injured, weather conditions change, you find a better statistical model — you should absolutely adjust your prediction. The best forecasters constantly update their estimates as they learn new things.

Your prediction locks in at the match's scheduled start time. Once the match kicks off, your last submitted prediction becomes final. You cannot change it after that point, regardless of what happens during the match.

Tip: don't wait until the last second. Submit your initial prediction early, then update it if new information arrives. This way, if you forget to come back or have technical issues, at least you have a prediction on record.

No. Update as often as you like. Only your final submission before the deadline is scored. The system does not track or penalize the number of updates. Frequent updating based on new information is a sign of good forecasting practice, not indecision.

Absolutely not. The contest is designed so that comebacks are always possible, especially with the escalating multipliers. Points earned in the Elimination round are worth 2×, and points in the Final are worth 3×. A strong finish can completely erase a weak start.

Also, this contest doubles as a talent search. Jump Trading isn't just looking at who finishes first — they're looking for sharp thinking and forecasting skill across the board. A player who starts poorly but demonstrates strong predictions later shows resilience and learning ability. That gets noticed.

Every single prediction is an opportunity to demonstrate your skill. Keep going.

If you have truly zero knowledge about a match — you've never heard of either team, you have no information whatsoever — then it's okay to skip that question. A random guess is just as likely to hurt you as help you.

But be honest with yourself: do you really know nothing? Even basic knowledge (one team is historically strong, the match is a home game, it's a group stage match vs. a final) can inform a prediction that's better than the field average. Most people underestimate how much their general knowledge is worth.

Rule of thumb: if you can articulate any reason to lean one direction, submit a prediction. If you literally cannot, skip it.

The scoring system punishes being confidently wrong much more than it rewards being cautiously right. This is because of the squaring in the Brier Score — errors grow exponentially.

A prediction of 60 is 'cautious' — it says you lean toward YES but aren't very sure. If you're right, your Brier is reasonable (0.16). If you're wrong, it's not catastrophic (0.36). Compare that to predicting 90: if you're right, your Brier is great (0.01), but if you're wrong, it's terrible (0.81).

The takeaway: be confident when your information supports it, but don't overstate your certainty. The squaring in the formula means the penalty for overconfidence grows faster than the reward.

No. You need to be a good judge of probability — which is a skill most people use every day without realizing it. When you say 'I think it'll probably rain tomorrow,' you're making a probability estimate. When you decide whether to bring an umbrella, you're acting on that estimate.

The contest is about the same skill applied to World Cup matches. You don't need to know what 'Brier Score' means mathematically — you just need to be good at saying 'I think this is about 70% likely' and being right more often than the crowd.

Domain knowledge (following the teams, understanding match dynamics, tracking injuries) is far more valuable than statistical expertise. The math just keeps score.

The winner receives a live $1,000,000 USD portfolio to trade on real prediction markets. This is real capital — not a simulation. Profits from the trading are split between you and Jump Trading. It's a genuine opportunity to prove your forecasting edge and profit from it.

Beyond first place, Jump Trading is watching for forecasting talent throughout the contest. Strong performance — even if you don't finish first — puts you on their radar.