Policy

Sports Predictions and Risk Disclosure

What the records do and do not mean, and the statistical limits behind every number in the app.

Version
0.9-draft
Last updated
2026-11-16
Effective
not yet set

Draft policy

This is a working draft. It is published so members can see how OddsButter operates today. Business details and legal review are still required before it is final: Pending Legal Review.

Contents
  1. 1. What OddsButter is
  2. 2. How the records are calculated
  3. 3. Sample size
  4. 4. Data feeds can be delayed or incomplete
  5. 5. Interpretation limits
  6. 6. Backtests and historical studies
  7. 7. Groups, consensus and curated portfolios
  8. 8. Your decisions are your own
  9. 9. Legal compliance and eligibility
  10. 10. Responsible use

1. What OddsButter is

OddsButter is an analytics and record-keeping product for sports prediction markets. It studies historical trader activity, calculates directional accuracy from that activity and official event resolutions, lets you compare traders, build groups, watch agreement, and track your own selections.

OddsButter is not a sportsbook, betting exchange, tipster service, investment adviser or fund manager. It does not place trades, hold funds, or act on your behalf, and it does not guarantee any outcome.

2. How the records are calculated

A trader's directional accuracy is a derived metric. OddsButter reconstructs each trader's pregame position in a game's moneyline market from qualifying historical trades, then compares that position with the officially resolved outcome. The result is counted as a win, a loss, or excluded as ungraded.

Prediction-market trade activity may itself be independently verifiable on-chain. The win-loss record shown in OddsButter is not that ledger: it is our calculation, using our methodology, cutoffs, minimum-position rules and resolution sources. Figures can differ from other trackers for that reason.

Methodology details, including cutoff timing and minimum position size, are shown inside the app so the basis of a record can be inspected.

3. Sample size

A small number of graded picks produces a wide range of plausible true skill. A trader who is 4-1 has a record, but not a reliable estimate. OddsButter shows sample size and flags limited samples, and ranking uses a sample-size-aware measure rather than raw percentage alone.

A trader with a high accuracy over a large sample is still not guaranteed to repeat it. A trader with a modest edge over a large sample can be more informative than an extreme percentage over a handful of games.

4. Data feeds can be delayed or incomplete

Trade data, market metadata and event results come from third parties. Imports can be delayed, rate limited or partial, and a result can be corrected after publication.

Where an import is partial, the app's coverage indicators show it. A record not yet complete can change as more of the history is loaded.

5. Interpretation limits

Converting market activity into a single side per game requires judgement. Position sizing, averaging, hedging, market conversions and late trades before kickoff all affect the interpretation. Splits and some market conversions are not fully modelled.

A trader's displayed pick is our best reconstruction of their pregame position, not a statement of their intent, and not proof that they consider the selection a prediction.

6. Backtests and historical studies

Backtests and windows shown in the research tools are historical studies, not forecasts. They are exposed to selection bias when a rule is chosen after looking at the data, and to lookahead bias if any information was not genuinely available before the cutoff.

OddsButter's point-in-time research tools are designed to reduce lookahead bias, but no backtest can reproduce every real-world friction, such as availability of a selection at a usable price.

7. Groups, consensus and curated portfolios

Combining traders does not guarantee higher future accuracy. Agreement narrows the number of qualifying selections and concentrates outcome risk; a group that agrees on fewer games produces a smaller sample and a wider uncertainty range.

Raising the agreement threshold makes signals rarer. A higher threshold is not automatically a better threshold, and a lower threshold is not automatically a worse one.

Curated portfolio membership is reviewed on a regular schedule and can change. Past membership is kept for transparency, so the record you see for a portfolio reflects the roster that existed at the time.

8. Your decisions are your own

OddsButter provides information and tools. Every decision you make using them is yours. We do not guarantee profits, wins, a positive return, or that any pick or alert will be correct.

Nothing in the app is financial, legal or betting advice. Consult your own judgement and, where relevant, a qualified professional.

9. Legal compliance and eligibility

You are responsible for knowing and following the laws that apply to you, including laws about sports prediction, betting information, online subscriptions and age limits. OddsButter is intended for adults aged 18 and over.

OddsButter does not claim to be licensed, regulated or approved by any gaming, financial or sports authority, and no such status should be inferred from the product or its content.

10. Responsible use

If you find yourself using prediction information in a way that feels compulsive, or if it is affecting your finances, relationships or wellbeing, stop and seek support from a local gambling-harm or mental-health service.

Do not spend money you cannot afford to lose on any activity connected to this information. OddsButter is a research tool, not a source of income.

Questions, choices and requests

Back to OddsButter