SYSTEM OPERATIONAL // V5.0

DECODE THE
NOISE
TRADE THE SIGNAL

VerasRecon analyzes WhatsApp/Telegram trading groups to detect market psychology, cognitive biases, and actionable signals — with special calibration for Latin American financial slang.

Launch Dashboard Learn More
verasrecon_core — python — 80x24
$ python pipeline.py --target=MERVAL_TRADERS_VIP
INFO: Parsing 14,203 messages...
INFO: Phone anonymization [CIS v5]... DONE
Analysis Phase 1: Emotion
-> Detected Sarcasm: "Sure it'll drop 🙄" (Inverted)
-> Dominant: ANXIETY (0.87)
Analysis Phase 2: Biases [ARG_CALIB]
-> "Dollar-Centrism": DETECTED
-> "Carry Trade Panic": LOW
Analysis Phase 3: Synthesis
-> Rule Check: ACTIONS > WORDS

VERDICT: BEARISH
REGIME: FEAR
CONTRARIAN SIGNAL: ⚠️ ACTIVATED
>> Verbal extremism high, but 0 sell confirmations detected.

THE PROBLEM

Why manual analysis is obsolete.

Time-Consuming

10k+
messages/day

Hours of scrolling through endless chat logs.

Signal-to-Noise

80%
irrelevant

Memes, greetings, stickers, and off-topic banter.

Misleading

High
Dissonance

What traders SAY often contradicts what they DO.

THE SOLUTION

VerasRecon 3-Phase Architecture

Automated Workflow

  • 1. Upload Export WhatsApp/Telegram history as .txt
  • 2. Parse CIS v5 parser removes PI & normalizes dates
  • 3. Analyze Multi-agent LLM pipeline executes
  • 4. Profit View synthesized verdict dashboard

# Quick Start

git clone https://github.com/veraslogic/verasrecon.git
cd verasrecon
pip install -r requirements.txt
streamlit run app.py
> Dashboard at https://verasrecon.streamlit.app
ARCHITECTURE

3-Phase LLM Pipeline

A sequential processing architecture designed to filter 60% of noise before applying expensive inference models.

PHASE 1: EMOTION

Llama 3.3 70B
  • • Dominant emotion detection
  • • Sarcasm identification (signal inversion)
  • • Toxicity scoring

PHASE 2: BIASES

Argentine Calibration
  • • Dollar-centrism detection
  • • Crisis memory trauma
  • • Carry trade anxiety

PHASE 3: SYNTHESIS

The Verdict
  • • Rule: ACTIONS > INTENTIONS > WORDS
  • • Contrarian signal calculation
  • • Composite risk score (0-100)

Argentine Calibration

Native lexicon containing 500+ financial slang terms.

> al horno, nos mataron, carry trade, MEP, CCL, el Toto

Sarcasm Inversion

Detects linguistic nuances that standard models miss.

> "Sure it'll drop 🙄" → BULLISH_SENTIMENT

Contrarian Signal

Buy signal triggers when verbal panic is high but sell actions are zero.

> IF panic > 80% AND sell_vol < 10% THEN BUY

Dissonance Detection

Measures the gap between group consensus and portfolio reality.

> Gap Analysis: Words vs Actions

CORE TECH STACK

Frontend
Streamlit + Plotly
Providers
Groq, OpenRouter, Google AI
Models
Llama 3.3 70B, Gemma 3, Gemini 2.5
Data
Custom Lexicon (500+ terms)