Adaptive Semi-Annual · ADB · Trim · Five-model risk-managed allocation with a daily kill switch that caps equity at 20% when the treasury-flow trend signal AND the US-equity supertrend confirm bearish at the same time. The flagship. Rebalance on January 1 and July 1, plus the same 5pp off-cycle threshold. Even lower tax drag; most realized gains end up long-term. + ADB v2 canonical kill stack: a daily kill switch caps equity at 20% when DTS-TF and US-equity supertrend both confirm bearish; the equity carve routes into a momentum-weighted defensive basket (TLT / GLD / DBC / BTAL / DBMF / KMLM, BIL fallback). Plus the Trim overlay: on Goldilocks days (inflation easing AND trend up) the held SPY routes to the SPMO momentum factor, and relatively-weak sleeves (EEM / DBA / SCHP / XLE — capped when their 6-month return trails SPY) trim into managed futures. See methodology.

Pick up to five lines to compare. Mix-and-match across the three Limnal models (Adaptive, the kill-switched flagship; Adaptive+, the same plus an opportunistic SSO leverage swap; Limnal Simple, the coarse 4-sleeve portfolio) and five rebalance cadences (Weekly through Annual), against the three passive benchmarks. The first variant you pick is treated as the headline — it gets the bold ink so the eye reads it as the protagonist.

Tail-risk view of the Limnal portfolio over the full backtest window. Historical reads percentiles directly off realized daily returns — fat tails baked in. Monte Carlo fits a Gaussian to the same series and draws 10,000 samples; the gap between the two is the size of the fat tail Gaussian VaR understates.

Each rebalance, rolled up into six top-level buckets. Stack order from the ground up is risk-off → risk-on (cash, alts, commodities, credit, fixed income, equity), so the chart shape itself encodes the model's risk posture as a function of time. Trace the cumulative-growth chart above against this one to see what the model was holding when it earned (or lost) ground.

Two envelope-level scores feed the allocation: the macro regime score (growth · inflation · liquidity composite) and the DTS score (Daily Treasury Statement liquidity proxy). Both run on a 1.0 → 3.0 scale where 1.0 is risk-off and 3.0 is risk-on. Trace this against §2 above: when the macro line dips toward 1, you should see the equity band in the allocation chart give ground to cash and fixed income; when it pushes toward 3, the equity band swells.

A standard fund-factsheet view: each cell is one portfolio's return for one calendar month, with cells color-graded by sign and magnitude. Limnal's row should read as fewer extreme months; the model's risk-management goal is to dampen the tails on both sides without giving up the median return.

A single backtest window can be lucky. The honest test is does the edge hold across many overlapping start dates? For each window length below, every possible starting date in the year-long base is tested independently. Limnal's edge is most decisive at the 180-day horizon: the noise washes out and the signal stabilizes.

The full description of how the regime is scored, how the score becomes an allocation, what data feeds the models, and the honesty caveats around the published numbers lives on the methodology page. Short version: six independent quantitative strategies score the regime daily on a 1.0 → 3.0 scale, the consensus drives a sleeve-by-sleeve allocation across a fixed 17-ETF universe, the backtest uses point-in-time inputs (no look-ahead) over the roughly 10-year window (January 2016 to today) for which the full model suite has data, and the published numbers are frictionless — real-world transaction costs, taxes, and slippage will subtract from realized returns.