Methodology

Rules before
prediction.

The public scanner uses unauthenticated Coinbase market data only. It evaluates all ten watched assets every run and publishes every finding that clears the rules when the market-wide protection is normal.

Qualification system

Disciplined filters.

  1. Complete public dataAt least 169 hourly candles and sufficient watchlist breadth are required. Missing data fails closed.
  2. Controlled pullbackThe asset must be below its seven-day high without entering a deep breakdown.
  3. Fee-aware upsidePossible upside must clear asset-tiered round-trip cost assumptions and the minimum net-return rule.
  4. Liquidity and trendSpread, 24-hour trend, seven-day trend, and relationship to the 72-hour average are checked.
  5. Reward versus invalidationReward/risk below 1.2× is rejected; 1.5× or better is preferred.
  6. Market protectionSevere BTC weakness or broad watchlist deterioration blocks actionable findings.
Core formulas

Inspectable math.

target = min(
  7d_high × 0.992,
  current_price × 1.08,
  72h_average × 1.06
)

net_upside = target / price − 1 − estimated_costs
reward_to_risk = gross_upside / downside_to_invalidation

The repository will expose the exact thresholds, implementation, tests, and algorithm version used for each snapshot.