Shark Snip Logo

Marketplace Rules

What can and cannot be sold on the Sharksnip marketplace.

Marketplace Rules

Last updated: May 6, 2026

These rules govern what may and may not be published on the Sharksnip Marketplace. Violations may result in content removal, account suspension, tier demotion, or termination.

What's allowed

Trained machine learning models

Models you have trained on data you have rights to. Includes:

  • Spread predictors, totals predictors, moneyline models
  • Player prop models (rushing, passing, receiving, scoring, rebounds, etc.)
  • Game flow / pace / situational models
  • Ensemble combinations of multiple model types

Components

  • Feature engineering pipelines
  • Training scripts and hyperparameter recipes
  • Backtesting harnesses
  • Ensemble strategy templates
  • Data preprocessing utilities

Datasets (opt-in publishing only)

  • Datasets you have rights to publish (your own scraped/compiled data, or licensed data with redistribution rights)
  • Synthetic / simulated data you generated
  • Aggregated/derived datasets that don't redistribute proprietary source data

Educational content

  • Tutorials, walkthroughs, courses about modeling techniques
  • Strategy guides for using the platform
  • Methodology explanations and case studies

What's prohibited

Picks, tipster content, "lock of the day"

"I think the Chiefs will cover" is not a marketplace product. Selling individual game predictions or "guaranteed winners" is prohibited because:

  • Multiple states regulate paid pick services / sports betting consultants
  • It conflicts with our positioning as a tools-and-models platform
  • It exposes Sharksnip to regulatory risk affecting all users

You can publish a model that produces predictions; you cannot publish today's predictions as a product.

Copyrighted or unlicensed data

  • Live odds feeds scraped from sportsbooks (use The Odds API or similar licensed source)
  • Proprietary stat services' data (e.g., licensed PFF/Sportradar/Statcast feeds without permission)
  • Player likenesses, team logos, league trademarks for commercial product imagery
  • Anything you don't have explicit rights to redistribute

Insider information

  • Models trained on leaked NGS data, unpublished injury reports from sources, or other non-public information
  • Anything obtained through misrepresentation, scraping in violation of ToS where actively litigated, or hacking

"Guaranteed winners" / misleading performance claims

  • "GUARANTEED 80% PICKS" — guaranteed anything is prohibited
  • Cherry-picked backtests presented as full track records
  • Hidden performance windows (only showing best months)
  • Stat fudging or fake calibration metrics

If you make a performance claim, it must be substantiated by your verified live track record. We reserve the right to remove listings with materially false claims regardless of stated disclaimers.

Malicious code

  • Backdoors, keyloggers, miners, or any code that performs unauthorized actions
  • Code that exfiltrates data from buyers
  • Code that calls home with anything beyond what the model needs to run inference
  • Deliberate misperformance (model designed to underperform when sold to others)

Harassment, abuse, illegal content

  • Content directed at harassing other users
  • Content promoting illegal gambling in jurisdictions where prohibited
  • Content involving minors
  • Hate speech, threats, doxxing
  • Sexually explicit content
  • Content advocating violence

Multi-account / sock puppeting

  • Creating multiple accounts to game leaderboards, run review brigades, or manipulate forks
  • Creating fake buyer accounts to inflate sales numbers
  • Coordinating with other users to manipulate Consensus weighting

Performance verification requirements

Every listed model has a public live track record. To list a model, you agree:

  • Live predictions are tracked (every inference is logged with the prediction, the eventual outcome, and metadata)
  • Track records are public on your model's listing page (Brier score, ROI, sample size, recent results)
  • You may not delete listed predictions to manipulate track records
  • Calibration and performance metrics are computed by Sharksnip from raw data, not self-reported

If a buyer-facing track record diverges materially from your published claims, listings may be flagged or removed.

Naming and metadata

Model names

  • Must be original and not impersonate other creators or established brands
  • May not include trademarks (NFL, MLB, team names, sportsbook names) in product titles
  • May not include misleading numbers ("96%", "Guaranteed Winner")

Descriptions

  • Should accurately describe what the model does
  • Should specify what data the model was trained on (in general terms — you don't need to disclose proprietary feature engineering)
  • Should disclose any known limitations (e.g., "trained on 2018-2024 data; may need re-tuning for recent rule changes")
  • May not contain affiliate links to third-party services

Profile information

  • May not impersonate other creators or sports figures
  • May not include contact info for off-platform paid pick services

Pricing rules

You set your own prices, subject to platform-suggested ranges and these constraints:

  • Minimum: 5 credits ($0.50) per access window — anything below isn't economically meaningful
  • Maximum: 1,000 credits ($100) per single access window — anything above triggers regulatory concerns about luxury-tier pack equivalence; tiers may be unlocked for Signature creators in some categories
  • Price changes: allowed up to once per week per item; existing buyers retain their unlock at the original rate
  • Discounts: can be offered (free trials, bundles); cannot be used to manipulate Consensus weighting or buyer counts

Forking

By default, all published models are fork-eligible. You may opt out, with these caveats:

  • Opting out forfeits royalty income from forks of your work
  • Opting out limits your reach (forking is a primary discovery mechanism)
  • Opt-out applies prospectively only — existing forks remain valid

Fork royalties (per Creator Agreement):

  • Direct child fork: 10% royalty to parent
  • Grandchild: 5% to parent, 2% to grandparent
  • Great-grandchild: 5% / 2% / 1% chain
  • Total ancestor royalty cap: 20% of any model's earnings

Removal and appeals

We may remove listings for:

  • Violation of these rules
  • Valid DMCA or other IP takedown notices
  • Documented buyer-side performance disputes (model performs significantly different from listed claims)
  • Activity inconsistent with these rules (manipulation, fraud, etc.)

When we remove a listing, we will:

  • Notify you via email with the reason
  • Provide an opportunity to respond and appeal
  • Issue refunds to buyers in their unlock window if appropriate (creator earnings may be clawed back)

Appeals: email creators@sharksnip.com within 30 days of removal with your case.

Repeated violations

A pattern of marketplace violations may result in:

  • Tier demotion (Signature → Established → Beginner)
  • Marketplace seller status revoked (account remains, but cannot list new items)
  • Account suspension or termination

Three or more material violations within 12 months will trigger marketplace seller revocation regardless of intent.

Questions?