Market overview
Trading Lab is a quant research and backtesting platform for Indian markets (NSE/BSE equities, INR). This public demo runs entirely in your browser on fictional instruments with synthetic prices: they are not live quotes and not real securities. See Research for what is built and what is not.
WATCHLIST · FICTIONAL INSTRUMENTS
| Instrument (fictional) | Last | Series chg | Trend | Strategy |
|---|
LAST SYNTHETIC BAR
CHART ·
RSI 14
Triangles mark the strategy's fills (up = buy, down = sell). The strategy and its inputs are set in the Backtest workspace.
Backtest workspace
A moving-average crossover on a seeded synthetic series. A signal at one close becomes an order at the next open, and every fill pays a cost. Change an input and every chart, metric and table below recomputes from one simulation.
EQUITY · STRATEGY vs BUY-AND-HOLD
DRAWDOWN
DAILY RETURNS DISTRIBUTION
MONTHLY RETURNS · % · ROW = SYNTHETIC YEAR, COLUMN = MONTH
BENCHMARK COMPARISON · SAME SYNTHETIC SERIES
| Metric | Strategy | Buy & hold | Difference |
|---|
TRADE HISTORY
| # | Entry bar | Exit bar | Entry | Exit | Net return | P&L | Bars held | Status |
|---|
- Generate a price series from the seed (drift regimes plus noise).
- Signal: in the market while the fast average is above the slow average, using only data up to that close.
- Fill at the next open, charging the cost on each side.
- Score against buying on day one and holding.
Data workspace
Where the numbers come from and whether they pass the validation rules. In this public demo every price is generated; the private research lab applies the same kind of rule-based flags to real historical data and never silently edits a bar.
PROVENANCE
VALIDATION RULES · LIVE RESULT
| Rule | Rows | Failing | Result |
|---|
MOST RECENT 20 BARS (SYNTHETIC)
| Bar | Open | High | Low | Close | Volume |
|---|
Research overview and architecture
A research platform for testing trading ideas on Indian equity data without fooling yourself: Python backtests, statistical evaluation against honest benchmarks, and paper-trading simulation before anything is trusted.
- 01Quantitative research on historical Indian equity data (NSE/BSE, INR).
- 02Backtesting and evaluation in Python, against buy-and-hold, with statistical tests.
- 03Paper-trading simulation to examine a strategy before any deployment.
Everything interactive on this site runs in your browser on synthetic price series generated from a seed. The companies are fictional. This site publishes no market data, no private strategies and no research results.
Pipeline (live values from the current simulation)
- Data
- Signals
- Backtest engine
- Evaluation
- Paper trading
TECHNOLOGY
- Research core
- Python, pandas, NumPy
- Tests
- pytest; calculations cross-checked by hand
- Private dashboard
- Local web dashboard (not deployed)
- Storage today
- CSV and append-only JSONL files
- This public site
- Static HTML, CSS, ES modules, SVG; no libraries, no backend
- Hosting
- Cloudflare Pages (static)
STATUS · WHAT IS REAL
| Capability | Status |
|---|---|
| Python backtest engine with costs and next-open fills | Implemented |
| Performance metrics and buy-and-hold comparison | Implemented |
| Daily data validation rules | Implemented |
| Exact-arithmetic FIFO portfolio ledger | Implemented |
| Statistical evaluation (bootstrap, null tests) | Partial |
| Paper trading | Partial: manual local runs |
| Indian equity data coverage | Partial: limited history, selected symbols |
| PostgreSQL data store | Planned |
| Scheduled, production paper trading | Planned |
| Public hosted backend | Planned |
METHOD
- No look-ahead. A signal seen at a close is acted on at the next open. A test checks that changing future bars never changes past results.
- Costs on every side, applied to each buy and each sell.
- Benchmark first. Every result is shown beside buy-and-hold.
- One state. Every chart, table and metric here comes from a single simulation, and integrity checks compare independent calculations.
- Flags, not deletions. Data validation marks suspect rows; it never removes them.
LIMITATIONS
- Synthetic series are random draws. They show how the machinery behaves, not how any market behaves.
- The bootstrap treats days as independent; real returns cluster in volatility.
- One crossover rule, one asset at a time, no slippage model, no taxes, no corporate actions in the strategy engine.
- Real-data coverage in the private lab is limited and is not claimed to be broad or decades long.
- Paper trading is partial, and simulated fills are not real fills.
- Past or simulated results do not predict anything.
It is an educational demonstration, not a broker or investment adviser. It has no accounts or saved data, no brokerage connection, and sends nothing anywhere.
Benchmark and statistical evaluation
A backtest number is one draw from a noisy process. These tools ask how much of it could be luck. They use the instrument and parameters set in the Backtest workspace.
BOOTSTRAP · ANNUALISED SHARPE
NULL TEST · SLID POSITION PATTERN
SELECTION BIAS · BEST OF 8 PARAMETER PAIRS
200 INDEPENDENT SERIES · STRATEGY MINUS BUY-AND-HOLD
Paper-trading simulation
The same engine, advanced one bar at a time. Each evening the rule decides; the order fills at the next open. Step through the series and watch the books reconcile against the backtest. This is a simulation on synthetic data, not a connection to any broker.
SIMULATED EQUITY · ORDERS
BOOKS
RECONCILIATION AGAINST THE BACKTEST
| Check | Result |
|---|
ORDER BLOTTER (latest first)
| Bar | Side | Fill price | Units |
|---|
Exact-arithmetic portfolio ledger
The portfolio accounting uses integer paise and FIFO lots, so cash, realized and unrealized profit and fees reconcile to the paisa. Two fictional companies, SYNA and SYNB; every quantity and price is invented.
| Date | Type | Instrument | Qty | Price | Amount | Fee |
|---|
| Instrument | Qty | Cost (FIFO) | Price | Value | Unrealized |
|---|
Reports
A written summary of the current simulation, generated from the same state as every chart. Use the button to print or save it as a PDF.