AHF Research runs a long-only, systematic strategy that rotates between an aggressive equity allocation and a defensive cash-and-managed-futures allocation based on a quantitative regime signal. The dashboard you're looking at shows the live signal, current positioning, and the full backtested track record.

Everything is rules-based. No discretionary overrides, no manual position changes, no intraday calls. The same signal pipeline that drives the live exposure also produced every bar of the historical performance shown on the Overview and Performance tabs.

Three exchange-traded funds, weighted differently depending on the regime state:

The signal classifies the market into one of two states. Each state maps to a fixed weight table:

The portfolio rotates only when the signal flips between states. There is no continuous re-weighting inside a state and no intraday trading. Historically that's averaged about seven flips per year.

The risk-on / risk-off signal is produced by a five-component multi-factor regime model. Each component measures a different aspect of market conditions; the components are combined into a single composite that drives the state assignment. The composition and weighting are part of the underlying research and are not published.

Most retail signal services don't publish anything you can verify. Two things distinguish what's running here:

• Out-of-sample holdout. Parameter selection is locked to a historical window that ends in 2019. Every bar of market data after that — the COVID crash, the 2022 bear, the 2023–2024 recovery, and everything since — is unseen by the optimizer. The performance you see across those periods is genuinely out-of-sample, not a curve fit.
• Score-gated parameter selection. New parameter sets only replace the live ones if they strictly beat the current best on a fixed scoring rule applied across the entire historical window. Re-runs aren't allowed to silently downgrade the live model under a different scoring regime.