How QTP Operates

Every trade passes a committee of independent checks before a single share is bought — and every check is measured against real, cost-survived outcomes. This is a structural view of the pipeline.

ILLUSTRATIVE — NO LIVE VALUES SHOWN

The committee — every signal runs this gauntlet

Sourcing

Broad Scanner

Surfaces candidate setups from the market in real time.

Analyst

Research (LLM)

Reads each candidate's context and evidence.

Scorer

VC Gatekeeper

Scores conviction; weak setups rejected here.

Macro

Regime Sentinel

Blocks trades fighting the prevailing trend.

Risk Mgr

Gate-K

Sizes by measured edge; enforces protection & no-revenge rules.

Execution

Broker

Places the order with a protective bracket.

Position Mgr

Trailing Stop

Manages the exit, locks profit, guards the stop.

What the gates block — by reason (historical base rate)

Counter-trend (wrong side of regime)

~44%

No-revenge cooldown

~20%

Unprotected / bad stop

~12%

Stop too wide for the edge

~8%

Measured negative edge

~5%

These are how often each reason fires — a picture of the discipline, not a claim of dollars saved. We never advertise avoided-loss figures: whether a rejected trade "would have lost" is unknowable, and honesty about that is the point.

Why this is different

The numbers are real. Sizing comes from a broker-reconciled ledger of actual fills — net of real slippage and fees — not from a backtest.

Risk decides before conviction. The Risk Manager sizes and can veto after the analysts have their say — position risk is never an afterthought.

Unproven signals stay on the bench. A strategy sizes at a tiny probation stake until it logs enough real trades to prove a statistically real edge. No promotion by hope.

Every decision is auditable. Each signal's full journey — every gate's verdict, the final order, the exit — is reconstructable from the record.

The one thing most systems won't show you

A signal-scoring model inside QTP was found — by QTP's own backtest — to be anti-predictive: its confident scores ranked setups backwards. Rather than ship it and hope, it was pulled back to recalibration. A system honest enough to bench its own model is a system you can trust with the ones it keeps.

QUANTLYS · QTP — HOW IT WORKS · ILLUSTRATIVE DATA ONLY