models, math, and the uncomfortable truth that most edges aren't real. four tools built to stress-test assumptions — mine and the market's.
[QA-01]Monte Carlo Strategy Simulator◈
ran 10,000 simulations of the same strategy. some end up 3x. some blow up. the math tells you which outcome is more likely — before you risk anything real. this is what backtesting can't show you alone.
Win Rate
55%
Average Win (R)
2.0R
Average Loss (R)
1.0R
Number of Trades
200
Starting Capital (₹)
Risk Per Trade
2%
Equity Curve Fan — 200 sampled paths from 10,000 simulations
Final Equity Distribution
each simulation is a different sequence of wins and losses at the same probabilities. the spread is the risk. the median is the plan.
[QA-02]Strategy Edge Quantification◉
win rate means nothing alone. a 40% win rate with the right R:R prints money. a 70% win rate with bad exits loses it. these are the numbers that actually matter — sharpe, sortino, kelly. computed from real inputs, not assumed.
most people look at returns. nobody looks at where the risk is actually coming from. this breaks down a portfolio mathematically — correlation, covariance, VaR — shows which position is quietly carrying the most danger.
Efficient Frontier — 500 random weight combinations · min-var and max-Sharpe highlighted
[QA-04]DCF Sensitivity + Monte Carlo Valuation⊕
a DCF gives you one number. one number is a lie. built this to show the full distribution of what a company could be worth — not just the base case, but the probability that the market is actually wrong.
Base Revenue (₹ Cr)
Revenue Growth Y1–3 (%)
Revenue Growth Y4–7 (%)
EBITDA Margin (%)
D&A as % of Revenue
Capex as % of Revenue
ΔWorking Capital as % Rev
Tax Rate (%)
WACC (%)
Terminal Growth Rate (%)
Shares Outstanding (Cr)
Net Debt (₹ Cr)
Current Market Price (₹)
Year-by-Year Projections (₹ Cr)
Sensitivity Matrix — Intrinsic Value / Share (₹) · green = above market price · red = below
Monte Carlo Intrinsic Value Distribution — 5,000 simulations (Box-Muller normal sampling)
the sensitivity matrix assumes inputs move independently. they don't. monte carlo lets them move together — which is closer to reality.