Strategies
Three cocktails,
one bitter base
Each strategy pairs one trading policy with one model version, under a bitter-cocktail name and built on Centaurium 1.1. Same forecaster, three different horizons and risk settings — here is the backtest for each, and the records of the ones we have closed.
The current line
Same Centaurium forecaster, three different shapes of risk.
Old Pal 1.1
Higher risk · concentrated
Centaurium 1.1
The sharpest sibling — concentrated, high-conviction, with a volatility brake.
+21.4% 7yr · −31.0% max DD
Americano 1.1
Lower risk · vol-managed
Centaurium 1.1
The lightest pour — volatility-managed exposure, built for shallower drawdowns.
+10.7% 7yr · −11.5% max DD
Black Manhattan 1.1
Overnight · event-driven
Centaurium 1.1
The night shift — holds a handful of names from the close to the next open, and nothing in between.
+17.5% 7yr · −13.5% max DD
Backtest, side by side
In-sample 2022–2024, against the S&P 500 under the identical 7-year cycle. Two of the three sit below the index on raw return and well above it on drawdown — that is the trade. Old Pal 1.1 is the exception in both directions: it compounds faster than the index, and it draws down deeper.
| Strategy | 2022 bear | 7yr / year | Max DD | Calmar |
|---|---|---|---|---|
| Old Pal 1.1 | −12.0% | +21.4% | −31.0% | 0.51 |
| Americano 1.1 | −7.7% | +10.7% | −11.5% | 0.60 |
| Black Manhattan 1.1 | −4.9% | +17.5% | −13.5% | 0.95 |
| S&P 500 (SPY) | −18.1% | +18.2% | −25.0% | 0.44 |
Higher Calmar = more return per unit of drawdown. That ratio — not the raw return — is where a vol-managed strategy earns its keep. Every figure here is the raw in-sample backtest. See a full strategy backtest →
Closed records
These ran on Centaurium 1.0 and stopped on 8 September 2026. We keep their pages up rather than deleting them: a strategy that no longer runs is part of the track record, not an embarrassment to hide.
Americano 1.0
Lower risk · vol-managed
Centaurium 1.0
The lightest pour — volatility-managed US mid-cap exposure, built for shallower drawdowns.
+11.0% 7yr · −9.9% max DD
Boulevardier 1.0
Balanced · index-relative
Centaurium 1.0
The balanced classic — index-like exposure with a regime-adaptive tilt, read against an index fund.
+15.6% 7yr · −24.7% max DD
Old Pal 1.0
Higher risk · concentrated
Centaurium 1.0
The sharpest sibling — concentrated, high-conviction, aimed at long-run compounding.
+19.4% 7yr · −37.5% max DD
Why a “7-year” number?
One good year proves nothing. So we weight the three real backtest years the way a realistic cycle does, then compound them. Here it is for Old Pal 1.1:
Markets do not repeat 2024 forever. We assume a realistic cycle — of every 7 years, 1 looks like 2022 (bear), 3 like 2023 (neutral) and 3 like 2024 (bull) — and compound the three actual backtest years accordingly.
Compounds above the index in the backtest, with a deeper maximum drawdown than the index. Most of the return came from a single year.
Each strategy card runs this same calculation on its own years. The weighting (1 bear, 3 neutral, 3 bull) is fixed and applied to the index too, so the comparison can’t be gamed by picking a flattering window.
How we name and version strategies
The model is the bitter botanical; a strategy is the drink mixed from it. Each carries a cocktail name — a fixed identity that never shifts mid-version.
One exact policy on one exact model. A policy change bumps the major; a model re-point bumps the minor. Every version freezes its own track record.
The current line is Americano 1.1 (defensive), Black Manhattan 1.1 (overnight), Old Pal 1.1 (concentrated). The name tells you where on the risk frontier it sits, and it never moves once published.