← Dashboard

Backtesting

Replay a strategy over historical daily prices using the real trading engine.


Reading balances…
production
AI: gemini
Read this before trusting any number below. Backtest quality is decided by the price history you supply. This engine looks for data/history/<SYMBOL>.csv first, then Yahoo Finance (real NSE daily bars, no API key needed), then Alpha Vantage if a key is saved, and only falls back to a deterministic synthetic random walk. Synthetic results prove the machinery works and say nothing about whether a strategy makes money. Currently detected CSV symbols: none · looking in /app/data/history

Fetch fresh data

Download real daily price history for the symbols you want to test and write it to data/history/. Do this before a real backtest — otherwise the run falls back to synthetic prices, which cannot tell you anything about strategy quality.

Overwrite mode re-downloads the window from scratch, so a stale or split-affected file is cleaned out. Append mode keeps older bars the window no longer reaches.
No fetch run yet this session.
SymbolRowsFirst dayLast dayFile
Loading what is on disk…

Run a backtest

Symbols are comma-separated NSE tickers. Exits, charges and risk checks come from the same code that runs live paper trading.

Compare runs momentum, mean-reversion and a random control over the same window, plus buy & hold.
No backtest run yet.

Saved runs

#NameSignalSourceWindowReturnBuy & holdTradesWin %Max DD
No backtests saved yet.