Public trust center

Clear limits.
Honest evidence.

UFC Value is a personal, independent probability-research project. It compares one governed forecast model with market prices, records evidence, and makes uncertainty visible. It does not accept deposits, place bets, or promise profit.

Privacy & storage

Know where records live.

UFC Value is a workspace application, not a multi-user account service. Personal record features are intended for a private operator workspace. This notice describes the application’s own storage; a hosting or network provider may keep separate logs.

Your browser
On the private workspace address, a local browser copy can contain odds-display preferences plus actual and paper ledger rows. The public address does not load or save those records. Anyone with access to the private browser profile may be able to read its fallback copy.
Workspace database
When server persistence is available, the application synchronizes one workspace state to SQLite. Captured quotes, paper recommendations, actual-wager entries, settlements, model versions, and audit events are also stored there.
Page-load count
UFC Value’s built-in counter stores aggregate daily page-load totals. It does not store visitor identifiers, IP addresses, or tracking cookies in that counter.
What is not requested
UFC Value does not ask for sportsbook passwords, payment-card details, or bank login credentials. Do not enter those details into any free-text field.
Retention and corrections
The browser fallback remains until that browser’s site data is cleared. Workspace SQLite state has no automatic expiration. The workstation backup command keeps the ten most recent backups by default and removes older copies when another backup is created; NAS update archives remain until they are manually pruned or replaced by a separate NAS backup policy. Governed forecasts, quotes, recommendations, outcomes, settlements, and audit events are retained indefinitely as append-only research evidence. Corrections add a new record instead of replacing the original.
Your controls
The private workspace’s browser-facing actual and paper records can be exported as JSON; that export is not a copy of canonical SQLite evidence or backups. You can clear the browser copy through your browser’s site-data controls. This project does not create public user accounts, so it has no public account-deletion workflow. Clearing browser data does not remove private workspace SQLite records, append-only evidence, or backups; removing those files is a deliberate local maintenance action and may break research audit history.
Shared or public installation?

This build does not create separate UFC Value user accounts. Do not record a personal bankroll or actual-wager history unless the operator has confirmed appropriate access control, user isolation, and retention for that deployment.

Responsible use

Research is not a reason to chase.

Use UFC Value only if you are of legal gambling age and sports wagering is lawful where you are. A model cannot remove risk, recover a loss, or turn wagering into dependable income. The safest financial decision is always not to bet.

Set limits before the event

Decide a hard money and time limit in advance. Use only money you can afford to lose, never borrowed money or funds reserved for essentials.

Never treat a gap as certainty

Model–market differences can reflect model error, stale prices, missing information, or ordinary uncertainty. “Meets paper policy” is not a promise or instruction to bet.

Do not chase losses

Do not increase stakes to recover money. Pause if gambling affects sleep, work, relationships, mood, or your ability to meet financial commitments.

Keep decisions clear

Do not wager while distressed, impaired, or under pressure. Take regular breaks and review total spending across every sportsbook, not one app at a time.

Terms

Use the project for its stated purpose.

These plain-language terms apply to personal, non-commercial research use. UFC Value is a project, not a paid service. No purchase, subscription, fee, warranty, or promise of continuing service is offered. By using this deployment, you accept responsibility for your decisions and agree not to present research output as a guaranteed result or validated betting edge.

Personal research project

Access is provided without charge for lawful personal research and education. Nothing on this site is an offer to sell a product, subscription, betting service, investment, or access to a sportsbook.

Information, not advice

UFC Value provides probability research and record-keeping tools. It is not a sportsbook, fiduciary, financial adviser, or execution service, and it does not accept funds or place wagers on your behalf.

Eligibility and local law

You are responsible for meeting the legal-age requirement and following the laws, rules, and sportsbook terms that apply to your location. Do not use betting features where sports wagering is prohibited.

No guarantee of accuracy or availability

Forecasts, market references, event times, and provider data can be incomplete, delayed, or wrong. UFC Value may be unavailable or change as evidence and governance requirements evolve.

Use at your own risk

To the extent allowed by law, UFC Value is provided “as is” and “as available,” without warranties. You assume the risk of relying on its output. The project and its contributors are not responsible for wagering losses or indirect losses caused by that reliance. Nothing in these terms limits rights or liability that applicable law does not allow to be limited.

Lawful and respectful use

Do not misuse the project, bypass access controls, interfere with its operation, enter another person’s private information, submit passwords or financial-account details, or present UFC Value as an official UFC or sportsbook service.

External services and affiliation

External links are provided for context; those services control their own content, availability, privacy practices, and terms. UFC Value is independent and is not affiliated with UFC, any fighter, sportsbook, prediction market, NCPG, or Gambling Therapy. Names and marks belong to their respective owners. Do not copy, alter, or redistribute private deployment data without permission.

Effective date and future changes

These terms are effective August 28, 2026. They may be updated when project behavior, law, or deployment practices change; the effective date will change with them. This personal, non-commercial notice is not legal advice. Any future version involving payments, advertising, commercial access, public accounts, or multi-user personal records requires a separate legal and privacy review before launch.

Methodology

One forecast path. Explicit evidence gates.

The career-sequence Transformer is the sole official UFC Value probability source for the live interface, model–market comparison, and paper-policy sizing. Every other model is a comparison, shadow, research service, or audit control unless the governed registry explicitly promotes it.

01 Official forecast A frozen pre-fight probability from the registered career-sequence model.
02 Market comparison A captured or clearly labeled reference price; missing markets stay missing.
03 Paper policy Conservative research sizing after uncertainty, quality, and exposure limits.
04 Prospective evidence Forecasts and outcomes are evaluated chronologically without rewriting history.

Probability, not certainty

The displayed win chance is an estimate. Seed-model variation describes variation across model fits; it is not automatically a calibrated confidence interval.

Quote quality changes interpretation

A two-sided, timestamped sportsbook capture can support policy evaluation. A reference or one-sided price is comparison context only and must not inherit captured eligibility.

Chronology is mandatory

A market forecast must be captured before the exact event start. Timing failures are preserved for audit but quarantined from scoring rather than repaired after the fact.

Scores need identical cohorts

Model and market results are compared only on the same chronological observations. Log loss is primary; sample size, calibration, controls, and event-blocked uncertainty remain part of interpretation.

Promotion remains a governed decision, not a dashboard color.

An affirmative named-book review requires at least 100 eligible natural settlements first frozen in complete post-genesis V1.4 batches, event-blocked evidence of improvement, every required control passing, and an explicit model-registry decision. Until then, betting edge remains under prospective validation.

Evidence integrity and missing results
  • Forecasts, quotes, recommendations, outcomes, settlements, provider responses, and identity decisions are append-only evidence. Corrections preserve the earlier row.
  • A missing comparison or shadow result remains null with a typed reason. The official model probability is never copied into a missing channel.
  • Named-book evidence requires compatible, timezone-aware event timing across the selected quotes. Incomplete or late evidence remains visible to audit but non-scorable.
Paper-policy sizing in plain language

The research policy compares an uncertainty-adjusted model probability with the break-even probability of a captured price. It begins with one-eighth Kelly, reduces sizing for input quality and model error, and applies 1% per-fight and 6% per-event exposure caps. A zero size or “doesn’t meet paper policy” is a normal outcome. A positive size remains a paper-research output, not financial advice.

See the read-only Research section for current model evidence, backtests, experiments, and governance status.