Frequently Asked Questions

TL;DR for the degens

• Free forever to build + backtest. Live picks are the only paywall.
• $5 Slate Pass · $10/mo Grinder · $30/mo God Mode. Top up extra Sharp calls at $1 each.
• Sharp calls = one-click ML helpers (Optimize Weights, Diagnose, Auto-tune). Free tier gets 1/week.
• Marketplace = creators keep 50% cash, paid via Stripe. No tier ladder.
• We are not a sportsbook, don't sell signals to books, and don't allow paid picks ("the Chiefs cover" is not a product here).

Every paid feature on Sharksnip rests on real-data dashboards like this one — pulled from local CSVs, no marketing fluff.

Getting started

What is Sharksnip?

A sports betting analytics platform with three integrated products: a model-building environment (Tinker), a marketplace where creators sell trained models, and Shark Snips Consensus — an aggregate signal from hundreds of opted-in marketplace models. We're built around the idea that the best sports analytics happens when many people independently build models, not when one company guesses.

Is this gambling?

Sharksnip is not a sportsbook. We don't accept wagers, hold funds for wagering, or facilitate betting. We provide analytics tools and a marketplace for ML models. What you do with the predictions is up to you, subject to your local laws.

Do I need to know machine learning to use this?

No. The free tier and paid tiers all give you fully functional picks, Consensus, and pickem tools without ever touching the model-building side. If you want to tinker, we have walkthrough content that gets you publishing your first model in a few hours.

What sports do you cover?

NFL, NBA, MLB, NHL are the primary coverage. We have growing support for college football, golf, soccer (MLS, Premier League), MMA, and tennis. The platform expands seasonally based on demand.

Where can I sign up?

sharksnip.com — free account in under a minute.

Pricing and billing

How much does it cost?

• Free / Sandbox: $0 forever — full Tinker access, 1 Sharp call per week (Mon reset), no live picks from your trained models
• Slate Pass: ~$5 one-off — live picks from one specific model, for one specific slate
• Grinder: $10/month or $99/year — live picks from all your models + 5 Sharp calls per month, rolling up to 10
• God Mode: $30/month or $299/year — everything in Grinder + 50 Sharp calls per month, rolling up to 100 + advanced Consensus + API
• Sharp call top-up: $1 per call — pay-as-you-go single Sharp call, available to every tier (Free included), no subscription

There's a $1 / 7-day Grinder trial. If you sign up for a sportsbook through our affiliate links, you get a free 7-day Grinder pass.

Why these tiers and not Pro/Elite?

Earlier versions of Sharksnip had Pro ($9.99/wk · $19.99/mo · $149/yr) and Elite ($79/mo · $599/yr). We collapsed and simplified to Free / Slate Pass / Grinder / God Mode in May 2026. Reasons: simpler funnel, lower price points convert better, and the new Sharp call feature replaces what the credit economy used to do (with much less cognitive load on users).

If you're an existing Pro or Elite subscriber, you're grandfathered at your current rate and tier mapping (Pro ~= Grinder, Elite ~= God Mode for feature access) until you self-initiate a change.

What's the difference between Grinder and God Mode?

• Grinder ($10/mo): Live picks from all your trained models. 5 Sharp calls per month, rolling forward up to 10. Full Consensus. Real-time picks. DFS optimizer. Discord access.
• God Mode ($30/mo): Everything in Grinder, plus 50 Sharp calls per month, rolling forward up to 100 (10× the Grinder quota — for active builders), Consensus dispersion analysis, market divergence alerts, custom slicing, API access (rate-limited), priority Sharp call queue, VIP Discord.

Both tiers can also top up extra Sharp calls at $1/call (no subscription required, doesn't expire, separate from monthly bucket).

If you're an active model builder, God Mode pays for itself in time saved (Sharp calls do meaningful ML work behind a single button click).

What is a Slate Pass?

A one-off purchase (~$5) that gives you live picks from ONE specific model for ONE specific slate. Example: you trained a spread model and you want to use it for this Sunday's NFL games. Buy a Slate Pass for that model + that slate. Run it as many times as you want during the slate window. No auto-renewal. Multiple Slate Passes can be active in parallel (different models, different slates).

Slate Pass is the "I'm not ready for a sub but I want this Sunday" option. It's a low-friction entry to paid features.

Can I cancel anytime?

Yes. Cancel from Account Settings. Cancellation stops future billing; you keep access through the end of your current billing period.

Do you offer refunds?

Subscription fees are generally non-refundable, though we issue exceptions for service failures or billing errors at our discretion. Slate Passes are non-refundable once the slate has started. Email support@sharksnip.com for billing issues.

What happened to credits?

We removed the credit economy in May 2026. The old system had credits at $0.10 each, credit packs from $4.99 to $499.99, earnable credits via daily logins, and credit-priced marketplace items (day pass / week pass / season pass / Builder Pass).

The new model is simpler: subscribe to a creator's model for live picks, or buy a one-shot Slate Pass. No fungible currency to translate, no FAQ on credit math, no daily-login mechanics.

If you have legacy credit balance, you can spend it on remaining marketplace items during the transition window. Residual balances can be converted to Sharp call quota at a rough 10:1 ratio (10 credits = 1 Sharp call).

What are Sharp calls?

Sharp calls are AI-powered button presses inside the model builder. Click a button (Optimize Weights, Run Monte Carlo, Improve Calibration, Suggest Features, Diagnose, Auto-tune), and Claude does the ML work behind the scenes. You see a before/after preview and decide whether to apply the change.

You don't write a prompt. You click a labeled button. The model sees the result with a structured diff, and you accept or reject.

How does the Sharp call quota work?

• Free: 1 Sharp call per week (resets every Monday, no rollover)
• Slate Pass alone: 0 Sharp calls (Slate Pass is just a live-picks unlock)
• Grinder: 5 per month, rolling forward up to 10 if unused
• God Mode: 50 per month, rolling forward up to 100 if unused
• Pack Breaker tournament prizes can grant bonus Sharp calls on top of your monthly quota
• $1/call top-up available to every tier as overflow

Monthly quota resets on the 1st of each month, in your local timezone. Weekly free calls reset every Monday. Bonus balance from Pack Breaker rolls forward (capped at 100).

Each click consumes one call from your quota — even if you reject the result. The cost was the API invocation; the choice to apply or not is yours, but you can't refund the call. Spend order is weekly → monthly → Pack Breaker bonus → $1 top-up balance, so paid top-ups don't get accidentally burned while subscription quota is still available.

How do Sharp calls work for free users?

Free users get 1 Sharp call per week, resetting every Monday. It's a real taste of the killer feature — click "Optimize Weights" or "Diagnose" on a model you trained, see what Claude does, accept or reject. The weekly call doesn't roll over (rollover is a paid-tier benefit), so use it or lose it each week.

If 1/week isn't enough, you have two options without subscribing: top up extra calls at $1 each (charged via Stripe one-time, no subscription), or hit a Pack Breaker tournament — top performers earn bonus Sharp calls. Subscribing to Grinder ($10/month) becomes mathematically smarter once you're using more than ~10 calls per month.

Do unused Sharp calls roll over?

Paid tiers, yes. Unused Grinder calls roll forward up to 2× your monthly allotment (so Grinder caps at 10 banked calls; God Mode caps at 100). The 2× cap prevents indefinite hoarding while still rewarding intermittent users — if you skip a month, you'll have a full extra month's worth banked when you come back.

Free tier's weekly call doesn't roll over. Each Monday is a fresh 1.

Pack Breaker bonus balance rolls forward separately (capped at 100 across tiers). Top-up balance from $1 charges rolls forward indefinitely with no cap and no expiry.

Can I buy a single Sharp call without subscribing?

Yes. The $1/call top-up is available on every tier, including Free. When you click "Optimize Weights" (or any Sharp call button) without quota balance, you'll see a confirmation modal — confirm and Stripe charges $1, the call runs immediately. No subscription, no hidden fees.

If you find yourself topping up more than ~10 times per month, Grinder ($10/month) is mathematically cheaper and includes rollover. We'll show that nudge in the confirm modal once you cross 8 top-ups in 30 days.

How do top-up Sharp calls interact with my monthly quota?

They're separate balances. The system spends in this order on each click: weekly quota → monthly subscription quota → Pack Breaker bonus → $1 top-up balance. So:

• Top-ups don't expire (unlike monthly quota, which renews each cycle)
• Top-ups don't count against the 2× rollover cap (the cap only applies to the monthly subscription bucket)
• Top-ups don't get accidentally burned while you have free monthly calls available
• You can stockpile top-ups across multiple months if you want

Top-up balance is visible in your account header alongside the other balances.

The marketplace

Who can sell on the marketplace?

Anyone. Free, Slate Pass, Grinder, or God Mode users can all publish models. There's no paywall on listing.

How much do creators earn?

50% of every transaction, paid in cash via Stripe Connect. Flat split, every creator, every transaction. No tier ladder. No promotion criteria. No demotion.

Why flat 50/50 instead of a tier ladder?

Earlier versions had Beginner / Established / Signature tiers (80% credits → 50% cash → 70% cash). We removed it for simplicity — see Creator Earnings & Status for the full rationale. The short version: flat 50/50 won the user research and is dramatically simpler to communicate, build, and run.

What can I sell?

Trained models, feature engineering pipelines, training scripts, backtesting harnesses, ensemble strategies, custom datasets (opt-in published), tutorials. See the full list in For Modelers.

What can't I sell?

Picks themselves (regulatory issue), real-time scraped sportsbook odds, copyrighted/unlicensed data, anything using insider info, "guaranteed winners" claims, malicious code. See Marketplace Rules.

How do I get paid?

50% of every transaction, in cash via Stripe Connect. Monthly cycle, $50 minimum payout, processed within 7 days of month-end. Stripe handles W-9 / W-8BEN paperwork and 1099 issuance.

What's forking?

Click "Fork" on any public model to get a starter that mirrors its architecture and features (NOT the trained weights). Modify, retrain on your own data, publish your version.

You're credited as a fork on the parent's profile, and you appear in the public fork tree. There's no royalty chain — the original creator doesn't earn from your fork's sales. The reward for being forked is visibility (forks lead users back to the parent), not direct earnings.

How do buyers verify a model is good?

Every published model has a public live track record: every inference is logged with the prediction and eventual outcome. Brier score, ROI, sample size, and recent results are visible on the listing page. Faked backtests don't survive contact with live data.

Can free users see backtests of marketplace models?

Yes. Free users can wire any creator's marketplace blocks into their own private model and see the backtest performance. This is genuine try-before-you-buy. The paywall is on live picks, not on backtests.

What if my model is no good?

Welcome to the club, everyone publishes some duds. Your model's performance shows publicly; buyers can decide. You can unpublish at any time. You can republish revised versions (which start a fresh track record). The platform is forgiving of failure and harsh on misrepresentation.

Shark Snips Consensus

What is Consensus?

A weighted aggregate of every opted-in marketplace model's predictions PLUS opted-in Pack Breaker tournament submissions. Top creators get 5× weight, new models get 0.25× weight, bad models are excluded. The result is a "wisdom of crowds" signal that no individual model on the platform can match.

Do I have to share my model's picks?

Only if you want to monetize them. Marketplace listing requires Consensus opt-in. If you keep your models private (i.e., don't publish), your picks stay yours alone — train models for your personal betting without leaking your edge.

Why don't you sell Consensus to sportsbooks?

Selling our aggregate signal to sportsbook operators would help them move lines against the same audience that pays us. It would also accelerate alpha decay across all our marketplace creators. It's not a market we'll ever serve.

We do sell aggregate research to media outlets, data aggregators, and academic researchers — different audience, doesn't kill alpha.

How is Consensus weighted?

Bayesian Brier-score weighting. Models with longer track records and better calibration earn more weight. Single-creator weight is capped at 5% of total to prevent any one creator from dominating. Forked models with high prediction correlation are deduplicated (treated as one combined vote). Pack Breaker tournament submissions enter at Probationary weight (0.25×) by default. See Consensus Methodology for the full math.

Why do I see a different consensus than my friend?

Tier-gated views:

• Free / Slate Pass: top consensus pick per slate (one game)
• Grinder: full consensus across all games + dispersion + 30d historical
• God Mode: market divergence alerts + custom slicing + full historical + API

Tinker (model building)

Do I need to install anything?

No. Tinker runs in your browser. We use TF.js (TensorFlow.js) which trains models entirely client-side.

Is Tinker free?

Yes. Tinker is free forever. Build, train, backtest as much as you want. The paywall is on live picks from your trained models, not on building.

What about my data — does Sharksnip see it?

For client-side training: no. Your data never leaves your browser. The trained model artifact is what you publish (~1-50MB JSON), not the training data.

What kind of models can I train?

• Linear and logistic regression
• Multi-layer perceptrons (MLPs / neural networks)
• Gradient boosted trees (XGBoost-equivalent)
• Simple ensembles (stacking, blending)
• Custom architectures via TF.js

For most sports betting use cases, the simpler models actually outperform complex ones — feature engineering matters more than architecture.

Can I import my own data?

Yes. Upload parquet, CSV, or JSON. Datasets stay in your browser unless you explicitly publish them.

How long does training take?

Depends on model + data size. Most NFL spread/total models train in 30 seconds to 5 minutes. Complex ensembles may take 15-30 minutes. God Mode subscribers get priority training queues.

Where's the data come from?

Sharksnip provides public-data feeds for major sports (nflverse-equivalent for NFL, MLB Statcast for baseball, etc.) for free. Players can also upload their own data.

For sportsbook odds (used in many models), we license from The Odds API and similar providers — paid users have access through the platform without separate API costs.

Pack Breaker (free tournaments)

What is Pack Breaker?

Free, skill-based prediction tournaments. Submit predictions for an upcoming slate (e.g., "pick the cover on every NFL Sunday game"), get scored against actual outcomes, win prizes if you're top of the leaderboard.

Is it really free?

Yes. No buy-in. No NFTs. No collectibles. No securities-law surface. Just submit predictions and see how you stack up.

What are the prizes?

Platform credit, never cash. Top performers win:

• Top 1: 1 month of God Mode (or equivalent Sharp call boost)
• Top 10: 5 bonus Sharp calls + a free Slate Pass
• Top 100: 1 free Slate Pass

Prizes go directly into your account; nothing tradeable, nothing for sale.

Why did you replace the old paid Pyramid League pack drops with this?

Reignmakers ($320 packs) and Sorare ($1,000+ legendary cards) crossed securities-law thresholds and got sued. Even at our $199 cap we were operating in a gray zone. Pack Breaker delivers the same engagement loop with zero legal risk.

What does Pack Breaker have to do with Consensus?

Tournament submissions can opt in to feed Consensus. They enter at Probationary weight (0.25×) by default — same as new marketplace models. This grows the Consensus dataset without requiring every user to publish a marketplace model.

Affiliate links and revshare

Why do you have affiliate links to sportsbooks?

It's a major revenue stream for us — when you sign up for a sportsbook through one of our links, the sportsbook pays us a commission. This funds the free tier and lets us offer Grinder at a low price. We disclose all affiliate relationships clearly.

Do affiliate links bias your picks or models?

No. Our picks are based on what our (and creators') models predict. The affiliate revenue is paid by the sportsbook for sending them users, not for steering specific bets. Where multiple books offer the same prop at different prices, the pickem unified line shopping always points you to the best price for you, regardless of which book has the higher CPA for us.

Do you get a cut of my actual bet losses?

Some affiliate deals are structured as revenue share (we get a % of net player losses). Others are flat CPA (we get a fixed amount per signup). We disclose the structure of each affiliate relationship publicly in our Affiliate Disclosures.

We're aware of the perverse-incentive concern. Our position: we don't recommend bets we don't believe in, and our published methodology is transparent enough that any user can verify it.

What's the free 7-day Grinder pass with sportsbook signup?

If you sign up for a sportsbook (DraftKings, FanDuel, etc.) through one of our affiliate links and make your first deposit, you get 7 days of Grinder free. The sportsbook funds it (not you), and we don't charge anything.

Account, security, support

How do I delete my account?

Account Settings → Delete Account. Or email support@sharksnip.com. We'll process within 7 days, retain for 90 days for legal/dispute purposes, then permanently delete.

Can I export my data?

Yes. Account Settings → Export Data. You'll get a JSON archive of your models, predictions, transactions, and account info.

What about two-factor authentication?

Available for all users; required for any creator receiving cash payouts.

How do I report a bug?

• In-app: gear icon → "Report a Bug"
• GitHub Issues (public): github.com/sharksnip/issues
• Email: support@sharksnip.com
• Discord: #bug-reports

What's your response time?

• Critical issues (account access, payment failures, data loss): within 4 hours
• Standard support: within 24 hours
• Feature requests: read but not always responded to individually; prioritized in monthly product reviews

Legal / regulatory

Are you available in my state / country?

We operate in the US, Canada, UK, Australia, and most EU countries. We restrict access from jurisdictions where sports betting analytics tools are prohibited. If you're unsure, our signup flow will tell you.

We do not facilitate gambling. Where sports analytics tools themselves are restricted (rare), we honor those restrictions.

Are you regulated?

We're not a sportsbook, so we're not regulated by gaming commissions in the way DraftKings or FanDuel is. We comply with general consumer-facing regulations: FTC affiliate disclosures, CCPA/GDPR for privacy, DMCA for IP, Stripe's KYC for payments, etc.

Do you sell my data?

No. We do not sell user data to third parties under any circumstances. See the Privacy Policy.

What about responsible gambling?

We have self-exclusion tools and visible-spend dashboards baked in. See Responsible Use Policy. Resources at 1-800-GAMBLER (US) and similar international helplines are available.

The company

Who runs Sharksnip?

Currently a solo-founder build with a community of early users. We're informal hires when great people show up — if you're a sports/quant nerd interested in helping, hit us up via Discord.

Are you funded?

Bootstrapped. We're aiming for profitable, content-led growth — the OddsJam playbook (which sold for $160M after running this way for 3 years) is the model.

What's your roadmap?

We share the rolling roadmap in our Discord and post quarterly updates on the blog. Highlights coming this quarter:

• Pricing page rewrite (Free / Slate Pass / Grinder / God Mode)
• Sharp call infrastructure (button-press AI helpers in Tinker)
• Marketplace MVP with flat 50/50 split
• Consensus aggregation pipeline
• Shark Snips of the Week content engine
• Free Pack Breaker tournaments

What if you get acquired?

We'd let users know with at least 30 days notice. Material changes to privacy or product would be communicated and, where applicable, opt-in/opt-out provided.

How can I support Sharksnip?

• Subscribe to Grinder or God Mode (the obvious one)
• Tell sharp betting friends
• Publish models on the marketplace (especially during launch — early creators get featured)
• Engage in Discord
• If you write or podcast about sports betting tools, we'd love a fair review (we don't pay for coverage, but we'll send you Grinder access for testing)

Have a question not covered here? Email hello@sharksnip.com or ask in Discord. We add to this FAQ regularly based on common questions.