delta · controlled evidence
agδ

delta

Drawdown-first market decomposition.

Delta is risk management, not alpha. It puts drawdown control first: an integer-deterministic decomposition with a pre-entry kill-switch that rejects trades before execution. On high-volatility assets it holds portfolio drawdown to a few percent where buy-and-hold takes tens — and it openly underperforms broad-market returns.

evidence ledger

measured facts, not exposed machinery.

Out-of-sample backtests across bull / bear / covid regimes (mass-backtest synthesis, 2026-05-08). The drawdown edge is the claim; the return shortfall is stated, not hidden.

All Delta figures are from backtested, out-of-sample simulation — not a live trading track record. Simulated results have inherent limitations and do not reflect actual trading, liquidity, or execution. Past or hypothetical performance is not indicative of future results. Nothing here is investment advice or an offer or solicitation to invest.

max drawdown
1.4–3.2% · vs 16–34% B&H
OOS return
underperforms B&H · every regime
OOS Sharpe
≈0.84 BTC · ≈1.14 ETH
pre-entry gate
pre-entry kill-switch · rejects before execution
regime
regime detection
determinism
integer-deterministic · x86-64 ↔ ARM identical
public layer

Drawdown-first.

On high-volatility assets — crypto, volatile equities — Delta holds portfolio drawdown to 1.4–3.2% where buy-and-hold takes 16–34%. The pipeline is integer-deterministic: identical input gives a bit-identical regime fingerprint on x86-64 and ARM.

stated openly

Not a broad-market alpha engine.

Delta underperforms buy-and-hold on returns and Sharpe in every out-of-sample regime. It is built for drawdown protection on volatile assets — avoid mature large-cap defensives and anything under ~15% annualised volatility. The decomposition mechanics stay sealed.