Why a good week proves very little
Results in fantasy contests are generated by events nobody controls: a reviewed catch, a touchdown that goes to the other receiver, a defensive score, an injury in the first quarter. Over a single week that variation is larger than almost any edge, which makes results a poor instrument for judging a process.
Slot 01Where the noise comes from
A lineup score is a sum of many small, weakly related events, and each one carries its own randomness. The total therefore varies far more than the underlying quality of the lineup does, which is the technical reason a good process can lose and a careless entry can win.
- Event variance
- Whether the ball arrives, whether the referee’s call stands, whether the shot goes in. Unrelated across players and mostly unpredictable.
- Role variance
- Whether the workload assigned to a player is what was expected. Coaching decisions, blowouts and injuries all change a role within a game.
- Payout variance
- The format amplifies or dampens the effect of the score. A flat payout compresses it; a top-heavy payout multiplies it.
Slot 02How long a sample actually is
Statements about a process need a sample of outcomes, not a handful. In flat formats a break-even rate sits above half the field by roughly the size of the fee, and distinguishing a rate slightly above break-even from a rate slightly below it takes a very large number of contests, because each individual result is close to a coin flip.
Two implications follow, and both are uncomfortable. A run of wins is almost no evidence of skill, and a run of losses is almost no evidence of the absence of it. If a break-even rate is only a few percentage points away from even, the number of contests needed to detect the difference is in the hundreds or thousands, not the dozens.
The arithmetic of thin edges
When results are close to a coin flip, the number of contests required to be confident about a small difference grows quickly. That is a property of the measurement, not a comment on any entrant.
Slot 03Result-based thinking and its two failures
Judging a decision by its outcome is the most common way an entrant learns the wrong lesson. When the result is mostly noise, the same good decision teaches a different lesson on a different day, and the entrant updates towards whatever happened rather than towards what is repeatable.
A tactic that worked once is adopted permanently, and it carries no information about future weeks.
A sound approach that lost is abandoned, and the abandons happen exactly when the noise is loudest.
Write down the decision and the reason before the lock, then review the reason, not just the score.
Bankroll management (covered in its own cycle of this series) and tilt (also covered separately) are the two practical consequences: the first decides how long a process is allowed to run, and the second decides whether a losing week changes the process. Neither is a fantasy-specific problem, and both are more important than the quality of any single lineup.
Slot 04A plan, and what it can and cannot do
A plan cannot make a raked game positive for the average entrant, and no page should pretend otherwise. What a plan can do is make the decisions consistent, keep the size of each entry within a limit set in advance, and produce a record honest enough to learn from.
- Set the entry size before the week A fixed amount per contest, decided when you are calm.
- Fix the number of entries An upper limit, so a bad week cannot be chased with a larger one.
- Write the reason for each lineup Before the lock. It is the only thing that can be reviewed afterwards.
- Review the reason, not the score The score is mostly noise; the reason is what can be improved.
- Stop on the limit A limit that is exceeded is not a limit.
Skill and variance in one line
- Weekly results are dominated by variation, so a small sample teaches almost nothing.
- Distinguishing a thin edge from none takes hundreds of contests.
- Judge the reason, not the score, because the score is mostly noise.
- A plan manages the size of the bet and the record; it does not remove the fee.
The score is mostly noise
The decisions that can be reviewed are the ones made before the lock. Everything after it is a result, and results in this game are noisy by construction.
Affiliate disclosure and risk warning
Every affiliate link on this page and in the header is a sponsored link to a partner operator, and we may be paid if you open an account through it, at no extra cost to you. That link pays us; it does not make any contest cheaper to enter, it does not improve any lineup, and it is never a recommendation to play. Nothing on this page is legal, financial or betting advice. 18+ only. Gambling is a real risk of real loss. Every contest format described here charges a fee out of the entry pool before prizes are paid, so the average entrant is behind before a single result is known, and most entrants lose money over a season. Whether a fantasy contest is legally skill or chance, who may enter, what must be disclosed and how prizes are taxed all differ between countries, states and provinces, and they change; a general explanation of how these products are put together is not advice on your own position or your own eligibility. Never stake money you cannot afford to lose, never borrow to enter, and never chase a losing week with a larger entry. Contest results are driven by events nobody controls, and a short run of results tells you almost nothing about the quality of your process. Gambling can cause serious financial harm, including debt and damage to relationships and mental health. Free and confidential support is available in most countries through national gambling-harm helplines, for players and for the people around them.