Methodology
How the engine works, where the numbers come from, and why every assumption is conservatively framed.
Objective: max Sharpe, not max return
The optimizer maximises the Sharpe ratio — expected return minus the risk-free rate, divided by portfolio volatility. Maximising raw return would push 100% into the highest-forward-return asset; the Sharpe objective rewards diversification because adding a low-correlation asset reduces portfolio volatility even if its own return is moderate.
The risk-free rate used is the India 10-year G-Sec yield: 6.8% (as of 2026-08-03).
Implementation: 4,000 portfolios are sampled via stick-breaking (Dirichlet-ish), each scored on its Sharpe ratio. The best candidate is returned. This is a Monte-Carlo approximation — not an analytical closed-form solution — but 4,000 samples over ≤9 assets gives a result within ~2% of the analytic optimum.
The 12% floor
Assets with a forward expected return below 12.0% are excluded from the optimizer by default. The studio has a toggle to include them for comparison, but the default removes them so the math is never poisoned by below-floor drag. The UI always flags which assets were excluded.
The floor is not a guarantee: the 12% Indian-equity forward is itself an estimate sitting on the borderline. If earnings growth disappoints, the flexi-cap forward could fall below 12% too.
Projection and Monte Carlo
Scenario bands use three deterministic paths:
- Bear:
expectedReturn − volatility(floored at −50%) - Base:
expectedReturn - Bull:
expectedReturn + volatility
Each path compounds a lump sum plus monthly SIP using annuity-due formula with monthly sub-compounding.
Monte Carlo simulates 1,000 wealth paths with annual returns drawn from
Normal(expectedReturn, volatility) using a seeded deterministic PRNG
(mulberry32 + Box-Muller). The p10 / p50 / p90 percentile bands are shown as a cone.
This is NOT a fat-tailed model — equity returns in practice have negative skew and
excess kurtosis. The cone is conservative directionally but understates tail risk.
All projections are estimates. Markets can fall 40-60%. Past returns are not a guarantee of future performance.
Sources table
| Asset | Fwd return | Vol | Below 12%? | Risk notes | Source |
|---|---|---|---|---|---|
| Indian Equity — Flexi/Multi-cap | 12.0% | 16.0% | No | Core diversified equity. Forward net of ~1.5% ER + ~12.5% LTCG sits ON the 12% line. 10Y track record (12.72%) inflated by 2020-2024 re-rating. Recent 1Y weak (+3.56%). Vol 14-20%; expect 40-60% peak-to-trough drawdown over a full cycle. | source |
| Indian Equity — Mid-cap | 13.5% | 20.0% | No | Forward 13-16% pre-tax lifts blended equity above 12%. Carries 40-60% drawdown risk; sized to survive without forced sale. 2 recent flat years. | source |
| Indian Equity — Small-cap | 14.0% | 24.0% | No | Highest forward CAGR after 2 flat years (FY27 earnings-bounce tailwind). Most volatile sleeve — cap exposure for drawdown tolerance. SEBI high risk. | source |
| International Equity — US/Global FoF | 5.0% | 17.0% | Yes | Forward only 4-6% INR net (Vanguard VCMM Jun-2026 base 4.2-6.2% USD gross, minus 1.1-2.0% double-layer FoF fees + 12.5% LTCG, plus rupee slip). Held for geographic + currency diversification, NOT return. S&P500 drawdowns -56.8% (GFC), -33.9% (2020). 30% probability of near-zero decade. | source |
| Gold — ETF / SGB | 11.0% | 16.0% | Yes | Trailing 43%/32%/23% are RALLY ARTIFACTS dominated by one-off +74.5% in 2025 — do NOT extrapolate. Already corrected 20%+ off Jan-2026 peak. Honest forward ~10-13% INR (9% USD base + rupee slip), wide band. Non-yielding diversifier held for negative correlation in equity drawdowns, not the 12% claim. | source |
| REITs / InvITs | 9.0% | 11.0% | Yes | Yield-driven: REITs 6-8%, InvITs 8-11% DISTRIBUTION (partly return-of-capital, not pure income). Forward WEAKENED in 2026 — yield spread over G-Sec compressed to <150bps (from >300bps in 2021). Slab-taxed on interest. ~7-11% pre-tax, structurally short of 12%. | source |
| P2P Lending (RBI NBFC-P2P) | 8.0% | 6.0% | Yes | CRITICAL: advertised 10-18% is GROSS/PRE-DEFAULT MARKETING, not net track record. RBI (Aug-2024) BANNED assured-return marketing, credit guarantees, secondary-market liquidity. Lender bears 100% of default loss; unsecured; NO liquidity (locked till repay); NO deposit guarantee; slab-taxed (no LTCG). Realistic net ~6-10%. RBI publishes NO official NPA time-series (4-12% inferred). Caps: ₹50L aggregate, ₹50k per borrower. Enforcement real (LenDenClub ₹1.99cr, NDX ₹1.92cr fined). | source |
| Arbitrage Funds | 6.5% | 2.0% | Yes | Cash-vs-futures carry, equity-taxed, near-zero drawdown. ~6-7% forward, compresses when volatility is low. Ballast/liquidity, not growth. Confirmed across Value Research/ICICIdirect/MySIPonline. | source |
| Corporate Bond (AAA / high-yield) | 8.0% | 4.0% | Yes | AAA 7.5-8.5%, sub-AAA high-yield 9-11% with real credit/default risk. Slab-taxed. Rate-sensitive. Caps below 12% even at the high-yield end. Ballast, not growth. | source |
The honesty layer
- International equity (~5% INR forward) — held for geographic + currency diversification, not return. Trailing figures are a US-bull + rupee-fall artifact. Vanguard VCMM Jun-2026 base forecast: 4.2–6.2% USD gross before double-layer FoF fees and LTCG.
- Gold (~11% INR forward) — the trailing 43% 1Y / 32% 3Y CAGRs are dominated by a single +74.5% event in 2025. Gold has already corrected 20%+ off its Jan-2026 peak. Honest forward is 10–13% INR with a wide confidence band.
- P2P lending (~8% expected, net ~6-10%) — advertised 10–18% is gross/pre-default marketing. RBI (Aug-2024) banned assured-return marketing, credit guarantees, and secondary-market liquidity. Lender bears 100% of default loss. No deposit insurance. Slab-taxed. Illiquid until repayment. Realistic NPA 4–12% (inferred; RBI publishes no official series). Caps: ₹50L aggregate, ₹50k per borrower.
- REITs / InvITs (~9%) — yield-driven; yield spread over G-Sec compressed to <150bps in 2026 (from >300bps in 2021). Distribution partly return-of-capital. Slab-taxed. Structurally short of 12%.