Methodology · powering the hub & academy

The Convexity
System.

A filter system that only allows trades where option pricing, market structure, and volatility state create asymmetric payoff potential. Six layers. Parallel engines. One quorum.

Optimize E[log(R)] — not maximum possible return. That naturally balances capital preservation, compounding, position sizing, and the occasional home run.

Driver 01
Relative Volume
Driver 02
Short Interest
Driver 03
Borrow Rate
Driver 04
Days To Cover
Driver 05
Gamma Exposure
Driver 06
Call Acceleration
Driver 07
Catalyst Quality
Driver 08
Liquidity Gap

// the 20% of signals driving 80% of squeezes. ignore most other data.

Architecture

Six layers, in order.

Each layer can veto. Nothing reaches execution unless every layer agrees.

01
Trade less

Layer 01 · Market Filter

Liquidity, spreads, volume expansion, trend alignment, IV regime. Rejects 90–95% of setups before they reach the engine.

02
Score asymmetry

Layer 02 · Convexity Engine

Delta acceleration, gamma exposure, vega sensitivity, theta decay, expected move. Only fires when reward > 3× expected loss.

03
Hard rules

Layer 03 · Risk Engine

Max 1–2% per trade, 3–5% daily loss cap, weekly drawdown stop, auto-disable after loss streaks.

04
Architecture

Layer 04 · Portfolio Logic

60% reserve · 30% active · 10% experimental. Inside experimental: 80% proven setups, 20% exploration.

05
Compounding edge

Layer 05 · Learning Layer

Stores Setup IDs — not tickers. Tracks PF, max DD, decay rate. Updates monthly, not trade-by-trade.

06
Guardrails

Layer 06 · Human Override

Robot cannot revenge trade, double down, remove a stop, or size up after losses. Period.

Quorum

Parallel engines vote. Meta controller decides.

Old way: News → Flow → Greeks → EV → Execution. If news fails, everything is wrong. New way: independent engines vote — one failure doesn't crash the machine.

Market State
Setup Detector
Options Engine
Business Engine
Catalyst Engine
Risk Engine
↓ each votes ↓
Meta Controller
Quorum reached → Trade

Reliability-Adjusted Edge

Trade size shrinks automatically when confidence or data quality drops.

RAE = EV × Confidence × DataQuality
ex: 2.0R × 0.5 × 0.7 = 0.7R

Weight Update Rule

Good setups get stronger weight. Bad setups decay. Weak signals disappear.

W_new = W_old × (1 + α × PerformanceSignal)
This is how the machine learns from mistakes — mathematically.
Home-run detection

Where squeezes actually happen.

Sector-level squeeze potential scoring. Higher = thinner float, more borrow stress, fatter tails.

Biotech
10/10
Small-Cap Healthcare
9/10
EV
9/10
Solar
9/10
AI Speculative Tech
8.5/10
Consumer Discretionary
8/10
Critical Minerals
8/10
Transportation
7/10
Utilities
6/10
Discipline

Every loss fits one bucket.

No vague "I got unlucky." That's amateur language.

Mistake
Timing error
Mistake
Direction error
Mistake
Volatility misunderstanding
Mistake
Over-sizing

Run the system. Don't read about it.

Every Nexus subscription includes the Convexity dashboard in the hub. The academy bundle teaches the math, the engines, and the override discipline that makes it work.