Zero-Budget Policy (2026-04-21)
Directive: Build it ourselves. No paid third-party services during alpha. Investment in paid tools only starts after NuroPicks + MGT revenue covers them.
Approved zero-cost services
| Need | Service | Cost | Why |
|---|---|---|---|
| Database | Neon (free tier) | $0 | Already wired, generous free tier |
| Hosting (web) | Vercel hobby | $0 | Existing. Fluid Compute, 100GB bandwidth |
| Hosting (bot) | Railway free tier | $0 | Existing. Switch to free hobby plan if current paid |
| Odds data | The Odds API free tier | $0 | 500 req/mo. 8 sports x 1 daily slate pull = 240 req/mo, fits |
| AI narrative | Claude Max subscription | $0 new spend | David already pays $200/mo Max, use that, not API |
| Payments (card) | Whop | 3-5% of revenue | No upfront, only on subscription revenue |
| Payments (crypto) | NOWPayments | 0.5-1% of revenue | No upfront |
| Resend free tier | $0 | 3k emails/mo free | |
| Analytics | Self-hosted on Neon | $0 | Funnel lives in our own funnel_events table |
| Track-record verifier | Self-hosted on nuropicks.com/record | $0 | Immutable ledger + public API + export is the verifier |
| CLV benchmark | Self-compute from The Odds API consensus | $0 | Compute CLV vs consensus-of-N-books, not Pinnacle specifically |
| Content hosting | Self on Vercel | $0 | Blog + changelog already on the site |
Killed (was in earlier plan, now removed)
| Was | Replaced with | Saves |
|---|---|---|
| OddsJam $199/mo for Pinnacle closing lines | Self-computed CLV vs consensus of free books via The Odds API | $199/mo |
| CAPTRACKER.markets enrollment (unknown pricing) | nuropicks.com/record is the ledger; post X receipts publicly; earn trust through repetition | likely $50-200/mo |
| SharpRank enterprise application | Same as above; revisit post-revenue | likely $100+/mo |
| Pinnacle API partnership | Consensus-of-books CLV | $0 vs unknown partner fee |
| Claude API direct for narrative | Claude Max subscription + ChatGPT/Copilot fallback | $0 new vs ~$1,080/mo |
| Paid Google Workspace | Use existing personal Gmail until LLC needs it | ~$6/mo |
| Mercury bank monthly fee | Mercury free tier (no monthly fee if above $0) | $0 |
| E&O insurance | Skip during alpha; revisit at 100 paid members | ~$50-100/mo |
Approved paid items (already committed)
- WY LLC filing: $102 one-time (already spent 2026-04-20)
- Northwest Registered Agent: $125/yr (already spent)
- Domain registrar (nuropicks.com only during alpha; .ai + .gg deferred until cash flow covers renewals): David's existing registrar spend
- Claude Max: $200/mo (David already pays for general dev use)
Triggers to start paying for things
- Month 1 cash flow positive (subs revenue > $500 MRR): start OddsJam trial OR build out our own closing-line scraper more aggressively (choose whichever costs less for the return).
- 10 paying members: revisit E&O insurance quote.
- 50 paying members: consider Google Workspace for LLC + mail forwarding.
- 100 paying members OR $3k MRR: consider CAPTRACKER verifier if still around.
Self-build roadmap (what we build instead of buy)
Self-verifier (immediate)
- Public immutable ledger already live at /record
- Weekly automated post to X with the ledger snapshot
- Monthly PDF "audit report" auto-generated and pinned in Discord
CLV computation (week 1-2)
- Modify picks publisher to snapshot odds at post time AND at
commence_time - 10 minutes - Compute CLV as posted_decimal / closing_decimal - 1, averaged across N books
- Default N = median of whatever books The Odds API returns per market
- Stored on picks.closing_odds + picks.closing_at columns (already exist)
- Modify picks publisher to snapshot odds at post time AND at
Odds ingestion (week 1)
- Single daily pull of slate from The Odds API free tier
- One pull per sport per day = ~8 req/day
- If we need more frequency, add a second pull at slate-close time
- Cache aggressively in Neon (we already have the picks table; add an
odds_snapshotstable for raw data)
Model narrative (week 2-4)
- Claude Max subscription used via Claude Code session scripts for batch narrative generation
- Alternative: local GPT4All / Ollama with Llama 3.1 for zero-cost narrative
- Keep narrative short (under 400 chars) to minimize cost if we do migrate to API
Email notifications (now)
- Resend free tier 3k/mo
- For anything above, use mailto: links
Rules for future "can we buy this?" requests
Question flow:
- Can we build it in under 8 hours of work? If yes, build it.
- Is there a free tier that works for first 6 months at expected scale? If yes, use free tier.
- Is the paid tier < 1% of projected month-6 MRR? If yes, consider post-revenue.
- Otherwise, skip.
This doc overrides earlier plans that assumed paid tooling (VERIFIER_INTEGRATION_PLAN.md, AFFILIATE_APPLICATION_CHECKLIST.md paid-tier assumptions).