How Thuztra backtests work
The honesty promise is only as good as the machinery behind it. This page states exactly what our engine does — and what it doesn't.
One engine, everywhere
Every number Thuztra publishes — in the product, in research articles, on the landing page — comes from the same backtest engine. There is no separate marketing engine. A strategy is a rules tree (rank, weight, gate, rebalance); the engine evaluates it bar by bar against historical data and simulates the resulting portfolio, including the cash flows of every rebalance.
Data
Backtests run on Thuztra's archive of daily NSE price series — Nifty constituents and major indices. Daily bars only: no intraday extrapolation. When a ticker lacks sufficient history for the requested window, it is excluded and the exclusion is reported with the result, never silently dropped.
Costs
Every rebalance leg is charged a transaction cost (default 5 basis points round-trip, configurable). High-churn strategies visibly pay for their churn — that is intentional and realistic.
Taxes
Realized trades are matched FIFO into tax lots and Indian capital-gains tax is estimated: 20% short-term, 12.5% long-term above the ₹1.25 lakh annual exemption. Results report both pre-tax and after-tax final equity. This is an estimate for research purposes, not tax advice.
The fixed-deposit comparison
Every backtest is compared against the same deposit schedule growing at a 7% annual rate — the bank fixed deposit that is every Indian retail investor's real alternative. A strategy that can't beat the FD after costs and taxes is reported as exactly that.
Survivorship-bias diagnostic
Backtesting today's index members into the past lets yesterday's failures silently vanish from the universe. Every run carries a survivorship report — which tickers joined the index after the backtest start, which dropouts are missing — graded none / low / medium / high. Runs can also be gated point-in-time, restricting candidates to stocks that were actually in the Nifty 50 at each rebalance date.
The luck test (Monte Carlo permutation)
A strategy's edge is tested by re-running it on hundreds of shuffled versions of the same market: the bar order is permuted jointly across all tickers (preserving each stock's return distribution and the cross-sectional correlation, destroying temporal structure), prices are rebuilt, and the strategy is recomputed. The reported p-value is the fraction of shuffled markets on which the strategy did as well or better. A high p-value means the result is indistinguishable from luck.
Known limitations
Backtests are simulations, and honest simulation means stating the gaps: daily bars only (no intraday fills); slippage beyond the configured cost is not modelled; the archive covers Nifty constituents and indices, not the full listed universe; and any backtest window is a choice — where results are window-sensitive we report multiple windows. When these limitations narrow, this page changes.
Backtests are educational research, not investment advice. Past performance does not predict future results. Investments in securities are subject to market risk.