This is a two-signal system built on Giovanni Santostasi's Bitcoin Power Law Theory.

Signal 1 — Power Law Deviation

Bitcoin's price follows P ~ t^5.82 (days since the genesis block). When price is below this line, the strategy buys more. When above, it buys less. The response is exponential: deeper discounts trigger progressively more aggressive buying, with 3.0x maximum reserved for discounts of -1.37σ or deeper.

Signal 2 — Residual Momentum

The 120-day rate of change of the power law residual. If price is falling toward fair value, this amplifies buying. If price is running away above fair value, it dampens buying. This filters out the "catching a falling knife" problem.

Validation

Tested out of sample on 2022-2025 data the strategy never saw during optimization:

+24.5% more BTC per dollar vs flat DCA. 100% of 10,000 bootstrap samples positive. All 36 parameter combinations positive. 4 of 5 rolling time windows positive.

Sell Signal — When to Take Profits

The power law works symmetrically. If being below the line means "on sale," being above it means "overpriced." The sell signal uses the same σ deviation:

Below +0.5σ: HOLD — accumulation zone, no reason to sell.
+0.5σ to +1.0σ: WATCH — getting warm, be ready.
+1.0σ to +1.5σ: TRIM — consider selling 10-20%.
+1.5σ to +2.0σ: TAKE PROFITS — sell 30-50%.
Above +2.0σ: AGGRESSIVE SELL — sell 50-75%. Bubble territory.

Every historical bubble has pushed above +2σ and then returned to fair value. The sell signal doesn't predict the top — it tells you when odds favor taking chips off the table.

What Did NOT Work

LPPL bubble timing, macro PCA gates, and noise filters all failed out-of-sample testing. The power law + momentum is the entire edge.