BraiinsRatchet/PROGRAM.md

Program Charter

Goal

Maximize expected BTC profit, or minimize BTC loss, for manually buying BTC hashpower on Braiins and pointing it at OCEAN/DATUM.

Operating Premises

• OCEAN uses TIDES: shares are paid only if they are inside the current share-log when OCEAN finds a block.
• OCEAN block discovery is stochastic and memoryless. A recent drought is not evidence that OCEAN is "due".
• The useful edge is expected value versus Braiins market price, plus operational quality and timing around observable fee/reward conditions.
• The system should improve through repeated small experiments, not through one large theoretical bet.

Ratchet Loop

1. Collect read-only snapshots from OCEAN, Braiins, local DATUM/Knots where available, and manual experiment results.
2. Score candidate policies against current guardrails and historical/paper-trade results.
3. Emit a manual recommendation.
4. If the recommendation is executed manually, record the exact order parameters and later realized rewards.
5. Keep changes to strategy.py only when they improve the measured score under comparable risk.

Hard Guardrails

• No code path places, modifies, or cancels Braiins orders.
• No owner token may be stored, loaded, or requested by the code.
• Watcher-only Braiins token may be provided via environment variable only.
• No secrets in Git.
• No containers or VMs.
• Runtime files stay inside this repository.
• Default branch name is master.
• Production-like behavior starts with monitor-only and paper trading.
• First live canary, if manually executed, should use minimum viable spend only.

Initial Scoring Metric

Use BTC-denominated expected value:

expected_net_btc = expected_ocean_rewards_after_fee - braiins_cost_btc
score = expected_net_btc - risk_penalty_btc - execution_penalty_btc

The strategy must show break-even price, discount to break-even, spend, duration, and maturity assumptions.