// Reference

Glossary

Terms used across Knox's writing. InDecision Framework vocabulary, trading concepts, military doctrine, AI terminology.

ABCDFHINOPRSTZ
A

Autonomous Task Completion

The length of a real-world task that an AI model can complete without human intervention. As of 2026, doubling every 4–7 months. The metric most relevant to understanding the practical capability trajectory of AI systems.

B

BEARISH

InDecision bias label indicating the bear case score significantly exceeds the bull case. Spread typically ≤-15. A High Conviction BEARISH signal is the mirror of BULLISH — same threshold, opposite direction.

BULLISH

InDecision bias label indicating the bull case score significantly exceeds the bear case. Spread typically ≥15. One of five bias labels output by the InDecision Framework scoring engine.

C

Center of Gravity (CoG)

Clausewitz's concept from On War: the hub of all power and movement in any system. Neutralize it and the adversary collapses. Misidentify it and you waste resources attacking peripheral targets. Applied to competitive analysis, markets, and geopolitics.

Conviction Band

The range classification of an InDecision spread. High Conviction (HC): |spread| ≥15. Moderate: |spread| 3–15. Neutral: |spread| <3. The band determines whether a signal justifies action.

Coup d'œil

Clausewitz and Boyd's term for the ability to see the whole situation instantly and grasp what matters. The skill that separates elite decision-makers from competent ones. Not intuition — trained pattern recognition operating at speed.

D

DualCaseAggregator

The swing-timeframe engine in the InDecision Framework. Builds independent bull and bear cases using 5 weighted scorers. Used for 4H+ and daily analysis. The DualCase model is what produces the bull/bear spread that determines conviction.

F

Friction

Clausewitz's term for the gap between theory and reality in any complex operation. Software systems, markets, and military campaigns all operate under friction — the accumulation of small failures, delays, and miscommunications that degrade the plan. Plan for it explicitly.

H

High Conviction (HC)

InDecision classification when the bull-bear spread is ≥15 (BULLISH) or ≤-15 (BEARISH). Historically the most accurate signal band. The threshold exists because close calls — spreads of 3 to 14 — produce lower-confidence signals that do not justify the same position sizing.

I

InDecision Framework

A 6-factor weighted scoring model for crypto market analysis. Generates independent bull and bear cases (0–100 each); the spread determines conviction level and bias label. 82.5% accuracy over 7+ years live trading.

→ Read the framework breakdown

IntraCaseAggregator

The intraday engine in the InDecision Framework. Runs at 5m/15m/1h timeframes. Three scorers: Technical RSI/MACD/BB (70%), Volume (20%), Cascade (10%). Designed for short-duration signals and PolyEdge's binary options positioning.

N

NEUTRAL

InDecision bias label when the bull-bear spread is between -3 and +3. Signal is too close to call. No position is taken. NEUTRAL is a valid output — it means the framework sees no exploitable asymmetry.

NEUTRAL_BEARISH

InDecision bias label for a moderate bear lean. Spread between -3 and -15. Moderate conviction band. May justify a cautious position but does not meet High Conviction thresholds.

NEUTRAL_BULLISH

InDecision bias label for a moderate bull lean. Spread between 3 and 15. Moderate conviction band. Signals are directionally aligned but below the High Conviction threshold.

O

OODA Loop

Colonel John Boyd's decision cycle: Observe, Orient, Decide, Act. The Orient step — mental model construction — is the most critical and most neglected. Speed through the loop beats mass. Originally developed for fighter pilot combat; applies universally to any competitive environment.

→ Read: Boyd's OODA Loop
P

PolyEdge

Knox's 8-factor AI trading bot running live on Polymarket binary options markets. InDecision Framework powers the core signal layer. Fee-aware execution: skips bets when market odds ≥85%. 1,063 automated tests, 93% code coverage.

Principal-Agent Problem

The misalignment of incentives between a principal (who delegates) and an agent (who acts). Root cause of most engineering team dysfunction. The fix is incentive redesign, not management pressure.

→ Read: The Principal-Agent Problem in Engineering
R

Recursive Self-Improvement

When an AI system participates in building its successor — debugging its own training, managing its own deployment. GPT-5.3 Codex is the first documented instance. A Gödelian moment with alignment implications.

→ Read: The Bottleneck Was a Feature
S

Spread

In the InDecision Framework, the difference between the bull case score and the bear case score. The spread determines conviction level: ≥15 is High Conviction, 3–15 is Moderate, <3 is Neutral. The spread is the primary output of the DualCaseAggregator.

T

Tesseract Intelligence

Knox's primary mission: a competitive intelligence platform combining engineering depth, market analysis, and strategic doctrine. All roads converge here — 16 years of engineering, 7+ years of market analysis, 25+ years of strategic thinking.

Thesis-First Analysis

The practice of forming an independent view from primary sources before consulting AI, consensus, or secondary analysis. The sequence that preserves analytical edge. Outsourcing orientation — the most critical OODA step — surrenders the advantage.

Z

Zero-Lag Execution

The operational ideal: a decision pipeline from signal to action with no human execution steps — only human judgment. The goal of the OpenClaw AI OS architecture. Every automated pipeline is a step toward this ideal.