These past-market results are research, not a promise of future returns. Historical Performance is the research record for TRINITY’s portfolio recipe: read the mix that won, see how it behaved, and copy the IDs that stay on every optimized portfolio after a person reviews the winner.
The winning recipe is a weighted mix of quality scores. TRINITY Trade Score leads, then relative strength and stage confidence. Those weights come from the current champion file — the same mix used when TRINITY builds new portfolios.
These figures are the current champion’s simulated history. Total return is —. The risk-adjusted score (Sharpe) is —. Use them with the mix above: first the recipe, then how it behaved, then whether it fits how you want to go steadily.
TRINITY ranks recipes on past results, including how the last stretch compared with the market after costs. Promotion to the live champion is a deliberate review step, so a change in mix stays visible and auditable.
Config hash — and run ID — are the current champion labels. On Certified you can copy them — like a batch number — and see which mix built the holdings.
How TRINITY thinks
TRINITY scores the market, records how the winning mix behaved, and leaves IDs on your book. Markets change, so treat these figures as the history of the recipe, then decide with a clear head.
The winning mix promoted from scheduled Historical Performance runs — used when TRINITY builds new portfolios.
TRINITY tests recipe mixes on past markets. A person reviews the winner. That champion is what TRINITY uses for new portfolio optimizations.
Quality signal weights and caps from the live fleet champion file in GCS.
Simulated champion performance from the promoted configuration — fleet-wide, not tied to a single portfolio run.
Simulated champion returns split by bull, bear, and sideways regimes in the research window.
Identifiers for the live promoted champion and latest TRINITY fleet run.
—
——TRINITY records how a portfolio recipe behaved in past markets before new books use that mix. You get the weights, the history, and the IDs to trace it.
After the research runs, one mix is reviewed and promoted. TRINITY uses that champion when building optimized portfolios.
TRINITY includes a model-training layer for later recipe versions. When more than one version is available, you will be able to compare them here.
FutureTRINITY runs on a schedule to refresh champions as markets evolve. Existing portfolios keep their original recipe IDs so you always know which mix built them.
These metrics come from the promoted champion configuration mirrored locally. They describe how this mix behaved in the research window. Use them with the weights below.
Copy these identifiers to match portfolios, support notes, and the promoted recipe file. The demo Certified tab shows the same fields after optimization.
—
—