SIGMA v5.0 — the whitepaper.

ABSTRACT

SIGMA (Systemic Intelligence & Global Market Analysis) integrates eight classes of heterogeneous signals — energy, maritime, behavioural, textual, network, temporal and technical — into a composite score normalised on [0, 100]. Aggregation uses an exponential asymptotic function so that extreme scores require simultaneous deterioration across multiple layers, modelling the real dynamics of systemic crises. SIGMA is not a stock screener; it is a financial-intelligence engine that applies open-source analysis methodology to capital markets.

AGGREGATION FORMULA

SIGMA_RAW   = Σ ( layerScore_i × weight_i × sectorMultiplier_i )
SIGMA_FINAL = 100 × ( 1 − e^(−k · SIGMA_RAW) )

The asymptotic function guarantees SIGMA_FINAL ∈ [0, 100] and amplifies extreme stress.

The eight analytical layers

1. 01

   Metabolic Economy Model

   metabolicScore = f(β, age, immuneResponse, debtRatio)

   Models the economy as an organism with metabolic rate, immunological capacity, and collapse thresholds. β = response speed to external shocks. immuneResponse = system capacity to absorb stress without regime transition. Output: metabolicScore ∈ [0,1].

2. 02

   Structural Fragility — Physics-Based

   E_potential = ½ · k · x² | CSD ↑ as system → bifurcation

   Calculates potential energy accumulated in the system, analogous to a physical system near a critical point. Detects Critical Slowing Down through increasing variance and decreasing recovery rate from perturbations. Minsky Moment: boolean trigger when fragility exceeds the structural stability threshold.

3. 03

   Behavioral Psychology Scoring

   psychScore = NLP(hedging) + divergence(official_vs_market) + capitulation_signal

   Analyzes text from GDELT sources and reports via NLP. Detects: linguistic hedging anomalies, divergences between official statements and market behavior, signs of capitulation or narrative euphoria. Language precedes action.

4. 04

   Network Topology & Financial R₀

   R₀ = β_network · k_avg / γ | R₀ > 1 → systemic contagion

   Models the financial system as a graph of interconnected nodes. Financial R₀ = the average number of entities affected if a central node fails. Calculated via percolation models on scale-free networks. R₀ > 1 indicates conditions of potential systemic contagion.

5. 05

   NLP & Causal Chain Analysis

   chain: cause → mechanism → effect → P(impact)

   Identifies causal chains in textual corpus: cause → mechanism → effect → probability of impact. Detects absence-as-signal — when sources that normally speak go silent. Narrative/structural divergence is often the earliest regime signal.

6. 06

   Predictive Dynamics

   λ(t) = μ + ∫α·e^(−β(t−s))dN(s) | H = log(R/S)/log(n)

   Three complementary models: Hawkes Process for self-excitation of volatility events, Hidden Markov Model for detecting transitions between latent regimes, Hurst Exponent H for characterizing time-series memory (H<0.5 mean-reverting, H>0.5 trending). CSD Score integrates all indicators of proximity to bifurcation.

7. 07

   Bayesian Learning & Recalibration

   P(θ|data) ∝ P(data|θ) · P(θ) — continuous recalibration T+30/60/90

   SIGMA is not static. Layer 7 recalibrates the weights of the other layers based on historical performance verified at T+30, T+60 and T+90. If Layer 2 had lower accuracy in the last 90 days on the energy sector, its weight automatically decreases for that sector.

8. 08

   Technical Indicators

   RSI · MACD · Bollinger Bands · ATR · Volume Profile · EMA cross

   The classical layer acts as a validator and counterbalance for layers 1-7. It is not the primary element of the final score, but severe technical anomalies amplify signals from the upper layers. Integration is bidirectional: structural SIGMA informs the interpretation of technical signals.

DATA SOURCES