EDUCATIONAL TOOL Historical study on 30-day-lagged data. These are educational findings about how documented methodologies behaved in the past. Past behaviour is not indicative of future results. Nothing here is a signal, recommendation, or forward projection. Not investment advice. Read the full compliance posture →
Bharath Shiksha

Practice Suite · Module 07 · Evidence Engine

The evidence, not the story.

Every codified methodology, run across a decade of real, 30-day-lagged NSE data, then judged with the same statistics an institution would use: out-of-sample validation, Monte Carlo drawdown, confidence intervals, and a multiple-testing correction. We publish what holds up and, just as important, what does not.

Loading the evidence report…

Instruments-
Real bars-
History-

Click any methodology for its equity curve, regime-by-regime breakdown, per-instrument results, and R-multiple distribution.

Methodology Verdict Expectancy (R) Win % t-stat Trades Out-of-sample Profit factor Median maxDD

How each verdict is decided

Each methodology is run on every instrument. Every trade uses next-bar-open entry, a fixed 1% risk per trade, a 2R target, a 30-bar time stop, and the FY26 NSE retail cost model (brokerage, STT, GST, slippage). We then pool the per-trade R-multiples and compute:

  • Expectancy: average R per trade after costs. Positive means the method made money.
  • t-stat: how far the expectancy is from zero relative to its noise. We only call something an edge when it clears a Bonferroni-corrected threshold across all methodologies tested (so we are not fooled by luck across many tries).
  • Out-of-sample: the last 30% of history, held back. An edge must survive here too.
  • Median maxDD: the typical worst drawdown across 1,000 Monte Carlo reshuffles of the trade order.
  • Regime-dependent: no edge overall, but positive in a specific market regime (trend / range / volatile).
  • Negative edge: statistically loses money after costs.