Dhanvi Experiments/001
EXPERIMENT 001Paper Trading / SimulationStatus: Completed

Multi-Agent Paper Trading

Testing whether specialized AI agents could coordinate across research, analysis and simulated trading during a complete market session.

5
Specialized agents
459
Opportunities evaluated
33
Trades selected
₹10,000
Simulated starting capital
The Filtering Story

Evaluation over execution.

Rather than executing on every signal emitted during the market session, the specialized agents rigorously evaluated 459 total opportunities and selected only 33 trade candidates that satisfied cross-modal parameters.

459
Trade candidates evaluated
Initial opportunity stream
~7.2% Selection rate
~7.2% progressed to selected trades
33
Trades selected
Final simulated paper execution
Agent Architecture

Five agents in coordination.

Each agent had a dedicated focus area during the session to avoid cognitive overload and ensure continuous validation.

01

News Agent

Context Ingestion

Collected and processed relevant market/news context.

02

Research Agent

Context Synthesis

Converted available information into broader research context.

03

Market Analysis Agent

Opportunity Identification

Evaluated market conditions and potential opportunities.

04

Company Analysis Agent

Fundamental Screening

Performed deeper analysis of candidate companies.

05

Trading Agent

Execution Selection

Used the combined intelligence to determine which opportunities progressed to simulated trades.

Execution Log

Session timeline.

The experiment ran across the full market session from 08:45 to 15:15.

08:45Pre-market research begins

System initialization and macro baseline calibration prior to open.

09:15News Agent stream ingestion

News Agent processes market open bulletins and filing disclosures.

10:00Research Agent synthesis

Converts unstructured information into broader market research context.

11:30Market Analysis Agent evaluation

Evaluates intraday conditions and identifies 459 candidate opportunities.

13:00Company Analysis Agent screening

Conducts deeper company-level screening on shortlisted candidates.

14:15Trading Agent simulated selection

Determines the 33 opportunities progressing to simulated paper trades.

15:15Experiment complete

Market session close. Ledger recorded and session metrics finalized.

Session Result
+1.89%
Paper-trading return for this single experimental session
₹10,000 simulated starting capital
About this result

This was a single paper-trading experiment conducted during development of Dhanvi. It does not represent live trading performance or expected future returns. Paper trading and simulations may differ materially from real trading due to factors including execution, fees, spreads, slippage, latency, liquidity and market impact. Past or simulated performance does not guarantee future results.

Experimental Hypotheses

What we were testing

Coordination
Could five specialized agents contribute to one continuous decision workflow?
Filtering
Could the system evaluate many potential opportunities while selecting only a subset for action?
Autonomy
Could the research → analysis → decision pipeline operate continuously across a market session?
Limitations & Scope

What Experiment 001 does not establish

A single paper-trading session does not establish:

  • expected future returns
  • long-term profitability
  • live trading performance
  • strategy robustness across market regimes
  • realistic execution performance
  • scalability to large amounts of capital
Experiment Summary Chain
5 specialized agents→459 opportunities evaluated→33 trades selected→₹10,000 simulated capital→+1.89% paper result
Single-session simulated result · Paper-trading prototype
Follow Dhanvi

Follow the experiments.

As Dhanvi develops, we'll continue documenting what we test, what works, what doesn't and what we learn. Join our early-access community to follow technical research updates.

All Experiments