How the Explorer works.

Purpose and simplified access

This is an educational, retrospective portfolio experiment. It assumes access to the selected overseas and domestic exposures, fractional holdings, and no eligibility barriers. It does not reproduce historical account-opening restrictions. No borrowing, short selling, derivatives, or personalised recommendation is involved.

An exposure is selectable only when every selected return year is supported. Assets are not given returns before they existed. Indian equity coverage starts with the 1991 return year. If a selected investment is unsupported, change the dates or remove it; the previous valid result stays visible until the configuration is valid.

Direct property, individual stocks, and Indian government bonds are not included. Government debt here is explicitly US Treasury-bill exposure. The deposit baselines are hypothetical, not actual rates offered in those years.

Investments and coverage

These are benchmark/model exposures, not executable products or fund recommendations. Standalone metrics use a ₹1 lakh lump sum, the selected dates, and modelled entry friction. Exact product availability, settlement, and charges can differ.

Random periods by default

First allocate your money, then select “Show my result.” The system draws a horizon uniformly from the supported lengths: 1, 3, 5, 10, or 15 years. It then draws a starting year uniformly among complete eligible windows for that horizon. Years and outcomes appear only after submission.

Coverage is checked for both initial holdings and any separate monthly contribution allocation. If an allocation includes an asset with a shorter history, only the horizons that fit are drawn. Eligible horizons are stated before drawing. Missing data is never replaced with an invented return.

Editing an allocation retains the revealed period; only submitting updates the result. “Try another period” draws again and avoids repeating the exact previous window where alternatives exist for the selected horizon. Other windows can overlap. A lucky result is not evidence that an allocation is generally better.

Manual dates are available only after explicitly selecting “Let me choose the years” inside More options. Those results are labelled as a user-selected period. “Use random periods instead” restores the random default. A new random draw can change the horizon as well as the starting year.

Results are organised into overall wealth, a chart, three numerical explanations, and a per-investment breakdown. Risk, access, comparisons, annual tables, and full historical-window analysis are optional expandable details.

Dates and monthly contributions

“First return year 2000” means purchasing at the opening boundary, end of 1999, then applying calendar-year 2000 returns. The table includes this opening boundary. 2000–2009 is ten return years. The standard data ends after the last complete supported year; there are no future projections.

Monthly savings arrive at the beginning of each model month. Because source data is annual, each year's growth is spread evenly using monthly return = (1 + annual return)^(1/12) − 1. Inflation is similarly spread within each year. This is a modelled contribution result, not an exact historical SIP.

Example: ₹1 lakh plus ₹1,000 each month over a zero-return year ends at ₹1,12,000 before costs. With a +12% annual return, the modelled monthly return is about 0.9489%; an initial holding ends at +12%, but contributions receive only the remaining months of growth. Actual monthly crashes and rallies can produce a different SIP result.

Contributions follow initial weights unless a different monthly split is enabled. Both weight sets must total 100%. No contribution earns returns before it arrives.

Allocation and rebalancing

Percentages and rupee amounts describe the same allocation. Weights use hundredths of a percentage point; rupee values are rounded for display. Editing a rupee amount converts it to the nearest supported weight. Other holdings are not silently changed. Remaining allocation is shown explicitly and can be assigned to savings.

Changing starting capital preserves weights. A 40% initial allocation is ₹40,000 of ₹1 lakh; it does not remain 40% after different investments grow at different rates. Ending weight is holding value divided by total wealth. Zero-value portfolios have no defined weights.

Default is buy and hold. Optional annual rebalancing restores initial target weights at each subsequent year's opening boundary, before monthly savings arrive. A simplified turnover model charges exit friction on overweight amounts and entry friction on underweight amounts, then allocates the remaining total to targets. Exact trades and lot sizes are not recreated.

CAGR means compound annual growth rate. Portfolio CAGR annualises the contribution-independent investment performance index, including modelled costs, rather than treating new savings as growth. With monthly contributions, XIRR remains a separate measure of return on timed cash flows. Investment rows are sorted highest-to-lowest by their standalone lump-sum CAGR over the selected period; that ranking is not an allocation recommendation.

The per-investment breakdown reports externally contributed money, ending holdings, and gain/loss. With annual rebalancing, gain subtracts net internal transfers as well as external contributions. Internal transfers sum to zero, so holding gains reconcile to the overall portfolio gain; modelled fees remain deducted.

For an unrebalanced lump sum, each row's return is its gain divided by its initial allocation. With monthly savings or rebalancing, “Period return” instead shows a standalone lump-sum holding over the same years, including entry friction. It is not the return on timed cash flows. The overall monthly-contribution result uses XIRR; the table explains the distinction.

Returns and income

Each holding compounds its annual series. closing wealth = opening wealth + external contributions + investment gains − costs − withdrawals. There are no external withdrawals in this version. Reinvestment and transfers are not new contributions.

Example: ₹1 lakh grows 10%, then falls 10%, ending at ₹99,000. Percentages compound; they do not cancel.

US equity total returns include reinvested dividends. Sensex returns here are price-only and omit dividends, understating returns relative to a dividend-reinvesting equity benchmark. No extra dividend is added to any total-return series. Gold and USD cash have no recurring income in this model. Treasury-bill income is reinvested within its return series.

The deposit-interest figure sums modelled savings and FD interest retained in the portfolio. It is not an additional payment on top of terminal wealth. No spend-income toggle or monthly dividend/rent promise is provided.

CAGR applies to the no-flow lump sum: (ending wealth / initial capital)^(1 / years) − 1. With monthly savings, XIRR uses model-month contribution timing and the terminal holding value. It is unavailable if the solver cannot find a valid solution. There is no exact day-level transaction simulation.

Currency conversion

All results are in INR. USD holdings use (1 + USD return) × (current annual-average INR per USD / previous annual-average INR per USD) − 1. Annual-average FX is a material approximation: it is not the exchange rate actually available at either year boundary.

Example: a 10% USD asset return and 5% dollar appreciation combine to 15.5% before costs, not 15%. USD cash earns only the FX movement and no interest. US equities, gold, and US bills combine their USD asset movement with this FX proxy.

Local gold premiums, foreign-account restrictions, exact conversion dates, and currency hedging are omitted.

Purchasing power

India's annual consumer inflation is compounded into a relative price index. real wealth = nominal wealth / cumulative inflation factor. Each selected period uses the opening boundary's purchasing power. The underlying World Bank annual CPI-change series uses annual-average price information, so this is a general purchasing-power measure, not an exact household cost index at the endpoint.

Example: 6% nominal growth with 5% inflation is 1.06 / 1.05 − 1, about 0.95% real growth. ₹2.4 lakh after ten years of assumed 5% inflation is worth about ₹1,47,339 in starting-year rupees. The 5% is an illustration, not our historical inflation assumption.

Monthly contributions are deflated when they enter. Real terminal wealth is compared with contributions expressed in the same opening-year rupees; a raw nominal sum is not used to claim real profit. Purchasing power across different rolling windows has a different reference year for each window.

Risk and recovery

The performance index is unitised around contributions so new savings do not mechanically erase a market loss. Purchase and rebalance friction reduce performance. Worst year is the lowest annual performance return; negative-year count counts returns below zero. Annual variability is the sample standard deviation and requires at least two years.

Year-end drawdown is observed performance index / previous observed peak − 1. We include the pre-purchase reference to reflect initial friction. Nominal peak recovery measures years from that peak until it is regained at an annual observation. If not regained, the result says “Not recovered.” Being below the latest peak is reported separately.

Example: a −25% decline is ₹25,000 on a ₹1 lakh reference holding. It is not automatically the visitor's rupee loss because their balance and contribution timing differ.

Annual data misses losses and recoveries between endpoints. These are not actual intra-year maximum drawdowns. Credit/default, custody, regulatory, and operational risk are not assigned fabricated probabilities. Smooth hypothetical deposits do not mean zero risk. There is no official or invented universal risk score.

Liquidity and access

Access buckets show estimated net INR obtainable within 1, 7, or 30 calendar days at the selected terminal boundary, assuming unchanged market prices. The metadata's conservative access-time assumption determines inclusion; estimated sale friction reduces the amount. Buckets are cumulative and must not be added together.

FD is a hypothetical annual-rate balance model. Monthly additions accrue fractional-year interest; individual deposit maturities are not tracked. All FD balance is assumed accessible within 7 days at the terminal boundary. Actual lock-ins, mid-term withdrawal and early-exit interest reductions are not modelled.

Example: ₹50,000 of a holding with assumed 7-day access and 0.2% exit friction contributes ₹49,900 to the 7-day bucket. That says nothing about what happens if its price drops before sale. Liquidity means access, not protection from loss.

These are model assumptions, not verified broker/exchange/bank service standards. INR conversion and platform interruptions may take longer.

Costs and taxes

Each investment's exact modelled purchase and sale percentage is listed above. Initial purchases and monthly contributions pay entry friction; annual rebalance turnover pays both sides. These percentages are illustrative constant assumptions, not historical fee records.

Ending wealth is holdings value plus cash, not full liquidation proceeds. Exit friction is applied when modelled sales occur or in the liquidity scenario. No separate annual fund/storage fee is deducted; the omission may overstate outcomes relative to a real product. Embedded fees, where applicable, must not be charged again.

All results are pre-tax. Capital-gains tax, income tax, crypto taxes, remittance charges beyond stated friction, exemptions, and user-specific tax positions are not calculated. These are not exact take-home outcomes.

Historical windows and A/B comparisons

For each horizon, the Explorer enumerates every annual-start window with complete common coverage for the selected exposures. The same initial capital, contribution amount and timing, cost assumptions, and rebalance rule apply. Lowest, median, and highest terminal wealth are observed outcomes, not forecasts.

Windows overlap and are not independent experiments. A 30-year window needs thirty complete return years; Indian equities have no 30-year result beyond 1991 here. A single supported window cannot establish an outcome distribution. Loss counts compare ending wealth with nominal or consistently deflated contributions, not a predicted probability of loss.

Saved A is a fixed snapshot. When capital, contributions, dates, or rebalance settings differ from current B, the interface flags the mismatch and hides the chart overlay. Contribution-weight differences remain visible in the saved configuration. Both are experiments, not rankings of recommended portfolios.

Active estimates and hypotheticals

No future-return assumptions, random emergencies, rent, or property appreciation are active. Future ideas in planning documents are not calculation rules.

Sources and updates

Data version 2026-10-14-v1; retrieved 2026-10-14; complete coverage through 2025. Sources are normalised by the repository's preparation script; their retrieval hashes are recorded in the dataset manifest. This is a sourced prototype with identified approximations, not exchange-certified investment performance.

• NYU Stern / Aswath Damodaran annual returns
• World Bank: official exchange rate, annual average
• World Bank: India consumer inflation
• World Bank: India deposit interest rate — no usable observations returned; not used for the current FD model.
• Existing Vimista Sensex compilation (exchange verification pending)

The Sensex compilation was inherited from the existing timeline dataset; its latest available December observation is used per year. US returns are reported to the precision of the source table. Source updates may revise earlier data. Re-run preparation, validate coverage, review changed values, and increment the data version before release.

The site does not fetch market data while you interact. Numbers come from the versioned dataset shipped with the page. Personal portfolios are not sent to a backend or retained after reload.