Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/deskiziarecords/QUIMERIA-HYPERION/llms.txt

Use this file to discover all available pages before exploring further.

Most QUIMERIA-HYPERION issues fall into two categories: veto conditions that block trade execution and import or path errors that degrade the pipeline to numpy fallback mode. This page walks through both, covers all error codes from the Ring 0 decision layer, and documents the known sensor weaknesses you need to account for in live deployments.

​
Veto decision tree

When r['veto']['trade_allowed'] is False, read r['veto']['reasons'] to identify which Ring 0 gate triggered the halt, then follow the resolution steps below.
Trigger: The λ7 Macro Causality sensor detected an adverse DXY or lead-asset correlation and issued a hard veto. This fires when the macro environment is in direct opposition to the structural bias.Check first:
  1. Verify the DXY/SPX correlation data is current — stale or missing macro data causes false vetoes.
  2. Confirm the r['sensors'] entry for lambda7 and check its raw score. A score below 0.3 alongside a hard veto indicates the correlation threshold is calibrated correctly.
Resolution:
  • If correlation is legitimate: wait for macro alignment before resuming execution.
  • If sensor appears miscalibrated: verify input feed for the lead asset is not missing bars. Do not override a λ7 veto during Privileged Mode (central bank intervention windows).
Never override a λ7 Macro Gate veto during scheduled central bank events (FOMC, ECB, BoE announcements). The veto exists precisely for these periods.
Trigger: The Mandra Gate computed a ΔE (information energy gain) below the minimum threshold of 0.02. The market is not delivering enough information content to justify a position.Check first:
  • Is the market in a Reverse Period (session transition, low-liquidity window)? If yes, this is expected behavior — wait.
  • Check r['mandra']['delta_e'] directly. Values near zero but positive indicate borderline entropy; values negative indicate chaotic price action.
Resolution:
  • For delta_e between 0 and 0.02: wait for the next AMD phase transition. The entropy gate typically clears during Manipulation and Distribution phases.
  • For negative delta_e (ERR_MANDRA_ENTROPY): stop execution immediately. Do not attempt to force trades — the market is in a chaotic state that the model cannot reliably predict.
Threshold: ΔE_min = 0.02 (adaptive range: 0.005–0.08 depending on volatility regime).
Trigger: The λ8 Light-Cone Violation detector identified statistical evidence of information leakage or front-running. This fires when price movements precede their causal triggers by more than the expected propagation window.Check first:
  1. Review recent events.log entries for repeated early price moves before macro releases.
  2. Check r['sensors'] for lambda8 score — a score above 0.8 alongside a veto indicates high-confidence detection.
Resolution:
  • Audit the data feed for latency anomalies or timestamp drift.
  • If the feed is clean, the issue may be the Adelic Tube radius being too narrow (see Known Weaknesses below). Widen the p-adic radius in core/adelic/ configuration.
  • Do not resume trading until the λ8 score drops below the veto threshold for at least 3 consecutive bars.
Trigger: The 7ZERO HMM tracker classified the current regime as 0 (sideways/ranging). The SMK requires a stable trending (1) or reversal (2) regime before approving signals.Check first:
  • Read r['gmos']['regime']. A value of 0 means the HMM has not found a dominant state.
  • Check r['gmos']['confidence']. Low confidence (< 0.6) combined with regime 0 means the market genuinely has no directional structure.
Resolution:
  • Force Accumulation logic: watch for λ1 volatility exhaustion (V_t / ATR₂₀ < 0.7) as the signal that the ranging phase is ending.
  • Monitor r['amd']['state'] — a transition from ACCUMULATION to MANIPULATION is the earliest structural signal of regime change.
  • Do not increase Kelly position sizing while the HMM regime is 0 or the confidence is below 0.6.
Trigger: The topological fracture detector found persistent H₁ homology loops exceeding the threshold. This indicates a fractal structural break in the price manifold.Resolution:
  • Wait for manifold compaction (H1 reset). This typically resolves within 3–10 bars after the fracture event.
  • Check logs/events.log for the bar index where TOPO:H1_FRACTURE was first logged and count bars since then.
  • This condition requires ripser to be installed. If ripser is missing, the topological detector falls back to a numpy stub and will never trigger HALT_H1_LOOP.
Trigger: KL divergence between the current and reference regime distribution exceeded 1.3× the threshold, indicating a statistically significant regime shift.Resolution:
  • This condition requires manual manifold recalibration. The reference distribution used for KL comparison may be stale if the instrument has undergone a structural change (e.g., ETF inclusion, macro regime shift).
  • After recalibration, the ERR_REGIME_FRACTURE error clears automatically on the next bar.

​
Error codes reference

CodeRoot causeResolution
ERR_REGIME_FRACTUREKL divergence > 2× thresholdManual manifold recalibration
VETO_LIAR_STATEλ3 phase inversion: φ_diff > π/2Wait for φ_diff < π/2 (typically 2–5 bars)
HALT_H1_LOOPTopological fracture — H₁ loop sum exceeds thresholdWait for manifold compaction (H1 reset)
ERR_MANDRA_ENTROPYΔE is negative — market is chaoticStop execution immediately
FUSION:LAMBDA_VETOAny λ sensor issued a hard vetoInspect r['veto']['reasons'] for the specific sensor
CONF:INSUFFICIENTFusion confidence < 0.2Insufficient signal agreement — wait for consensus
MANDRA:ΔE<0Negative information energy gainDo not trade — chaotic state
L3:LIAR_STATEFFT phase inversion > π/2 in λ3 HarmonicWait for phase realignment

​
Known weaknesses

During ultra-high-frequency volatility bursts (e.g., news events with bid-ask spread widening > 5 pips on EUR/USD), λ3 may over-veto via VETO_LIAR_STATE even when the structural DNA — DealingRange, FVGs, and AMD state — is clearly bullish.Symptom: VETO_LIAR_STATE fires on 3+ consecutive bars during a strong displacement move. The chart shows a clear Order Block and FVG entry opportunity, but the veto panel shows repeated λ3 blocks.Mitigation: Monitor for Opportunity Decay — the window where a valid signal was blocked by λ3 and the move is now underway without a position. Log these instances to a separate file for post-session review and use them to calibrate the λ3 phase threshold (default π/2, adaptive range π/4–3π/4).
The p-adic radius in the Adelic Tube (core/adelic/) determines the tolerance for causal force vectors. When the radius is too narrow, valid institutional price delivery is flagged as a λ8 Light-Cone Violation.Symptom: λ8 Kill-Switch fires repeatedly during clean trending sessions with no evidence of front-running. r['sensors'] shows lambda8 scores above 0.7 on trend-continuation bars.Mitigation: Widen the p-adic radius incrementally. There is no universal correct value — it depends on the instrument’s typical volatility and delivery speed. Backtest with at least 3 months of data before tightening the radius in production.

​
Numpy fallback mode

If SMK_DIR is not set or module imports fail, smk_pipeline._find_smk_root() cannot locate the project root. The server starts but runs with all 18 detector modules replaced by numpy stubs that return neutral scores. How to detect fallback mode: 
curl http://localhost:8000/api/status

{
  "modules_ok":    0,
  "import_errors": 18,
  "pipeline":      "numpy_fallback"
}
All r['sensors'] values will be 0.5 and r['veto']['trade_allowed'] will always be False in fallback mode. Resolution:
1

Verify SMK_DIR is set

echo $SMK_DIR
# Should print the project root path, e.g. /home/you/QUIMERIA-HYPERION
2

Verify PYTHONPATH includes both the root and backend

echo $PYTHONPATH
# Should include: /path/to/QUIMERIA-HYPERION:/path/to/QUIMERIA-HYPERION/backend
3

Restart the server with the correct environment

export SMK_DIR=/path/to/QUIMERIA-HYPERION
export PYTHONPATH=$SMK_DIR:$SMK_DIR/backend:$PYTHONPATH
cd backend
python -m uvicorn main:app --host 0.0.0.0 --port 8000 --reload
4

Check session.log for specific import errors

curl "http://localhost:8000/api/logs/session.log?lines=50"
Each failed module import is logged with the specific exception so you can identify missing dependencies.

​
SMK_DIR resolution order

smk_pipeline._find_smk_root() searches for the project root in this order:
  1. SMK_DIR environment variable (highest priority)
  2. Walk up from backend/ looking for requirements.txt
  3. Fall back to numpy stubs if neither succeeds
Always use SMK_DIR in production — the walk-up heuristic can resolve to the wrong directory if the backend is run from an unexpected working directory.

​
Breakage-prevention rules

Follow these rules before and during live operation to avoid the most common failure modes:
  • Never execute without Mandra Gate ΔE > 0.02 confirmation.
  • Always verify ATR₂₀ benchmarks before triggering the λ1 Phase Entrapment threshold.
  • Never override a λ7 Macro Gate veto during Privileged Mode (central bank intervention).
  • Always ensure HMM Regime is STABLE (1 or 2) before increasing Kelly position sizing.
  • Never increase Kelly fraction beyond 0.5 — the UQPCE uncertainty engine clips it at this value.

​
Safe upgrade to v1.1

1

Verify all 25 modules are status OK

curl http://localhost:8000/api/status
# Confirm: "modules_ok": 25, "import_errors": 0
2

Register new forensic modules

python backend/plugin_manager.py
3

Wipe the ATR₂₀ cache

Stale ATR₂₀ values in the cache cause incorrect λ1 Phase Entrapment triggers. Restart the server with a fresh data load after any kernel change.
4

Back up core kernel and risk modules

cp -r core/kernel/ core/kernel.bak/
cp -r risk/ risk.bak/
Keep this backup until you have run at least one full trading session on v1.1 without errors.

Build docs developers (and LLMs) love

Get started for freeTalk to us