Qovaryx 3.0 shipped — 4-tier authority, data moat, BYO-AI, 4 public risk tiers all positive every year 2021–2026
Qovaryx 3.0 is out. Full arc, no recipe leaks:
5-year full-pipeline backtest, $500 → all four public risk tiers positive every year 2021–2026 (real + synthetic option corpus, all wired safety gates engaged):
| Tier | 5-yr return |
|---|---|
| Conservative | +523% |
| Standard | +1,793% |
| Aggressive | +22,785% |
| Extreme | +47,326% |
Universal circuit breakers on every tier: daily loss halt, correlation cap (max 40% single sector), drawdown ack if account drops 10% in 7 days, slippage governor. Aggressive/Extreme require typed operator ack + mandatory milestone profit-pulling.
Two brand-new research posts, framings-only:
Head Architecture Advances — the four-tier execution-authority taxonomy (REAL_VALIDATED_SCOPED / SYNTH_VALIDATED_SHADOW / VERIFIED_FORECAST_CONTEXT / OFFLINE_FORECAST_ADVISORY), the batch-graded hypothesis funnel, the rehab queue, the retirement discipline. Nothing overrides an order without a signed head→consumer→mutation contract.
The Data Moat — the corpora that sit behind the specialist cluster: multi-decade daily equity coverage, multi-year real options-chain snapshots for the highly liquid names, app-faithful multi-symbol option-PnL labels, calibrated synthetic bridges with label-bridging validation, intraday minute-bar snapshots, and a privacy-preserving decision-outcome corpus.
On the trending "ChatGPT / Claude will trade for you" wrappers — they are not consistently profitable when benchmarked out-of-sample. General-purpose LLMs were trained to be helpful chat assistants, not to price options or size risk. Qovaryx uses BYO-AI (7 providers: OpenAI / Anthropic / Google / xAI / OpenRouter / Ollama / custom OpenAI-compatible) for chart analysis and position review — a human-facing decision-support surface — while order execution stays with our verified specialist heads under strict authority contracts. Different job, different tool.
The full trading cluster runs on any consumer CPU. No GPU. No cloud round-trip. Deep cleanup this milestone: 22 previously-dormant safety features wired end-to-end, ~1,750 lines of dead code removed, 80/80 focused test suite passes.
More: qovaryx.jehorizon.com · Public research devlog
Trading options involves substantial risk of loss. Backtest results are historical simulation, not a forecast. Not financial advice.