One for every risk appetite.

How MindForge backtests

Every backtested percentage on this site comes out of the same simulation, run the same way for all three strategies. We publish how it is built — not what the models score on, which stays proprietary — so you can judge the numbers by the rigour of the process rather than by how good they look.

1. The universe is fixed before the test Each strategy tests against a defined list of NSE-listed securities set in advance. Nothing is added to it because it later turned out to do well.
2. Nothing is scored on data it could not have had At every historical rebalance the model sees only prices and volumes up to that date. Over a year of history is burned as warm-up before the first month is ever scored.
3. Extremes are tamed, not deleted Every input is trimmed at its statistical extremes and put on a common scale before ranking, so one freak print cannot carry a stock into the book. Names with too little trading history to score are left out rather than guessed at.
4. Every month counts The model rebalances on the last trading day of each month and holds until the next. Every month in the window is scored — none is skipped, re-run or chosen after the fact.
5. Turnover is charged On both equity strategies each rebalance pays a one-way trading cost on what actually changed hands, at a wider rate for the small- and micro-cap book. Returns shown are after that charge.
6. The assumptions live in one file Cost rates, the risk-free rate and each strategy's benchmark are declared once, centrally, and read by the backtest, the charts and this website alike. No run carries its own private numbers.
7. The benchmark is the real index Each strategy is measured against the actual index path over the same dates — not a smooth long-run average. So the benchmark's own return, risk and drawdown all describe the same window.
8. One run, one set of numbers Return, volatility, Sharpe, drawdown, win rate and best month are all read off the same equity curve from the same run. No metric comes from a different, friendlier window.
9. Published straight from the run Every figure on these pages is written in from the simulation's own output files each month by the publishing script, then re-derived from the equity curve by a separate check before anything goes live.

The full methodology, the exact cost assumptions, benchmark construction and every known limitation are set out in Backtest methodology & known limitations. Backtested performance is hypothetical, has inherent limitations and is not indicative of future results.