PortfoliosLab

What it is

PortfoliosLab is a browser tool for taking an allocation apart. Type holdings and weights, or load one of its stored public portfolios, and it returns growth against a benchmark, the Sharpe, Sortino, Omega, Calmar and Martin ratios, drawdowns, capture ratios, value at risk, five volatility estimators and a correlation matrix. Seven optimisers reshape the weights — mean-variance, risk parity, hierarchical risk parity, HERC, maximum diversification, nested clusters and Schur complementary allocation — with a training window and a walk-forward reoptimisation frequency rather than a single in-sample fit.

Around that sit three screeners (stocks, ETFs, mutual funds — 5,417 US ETFs on 14 September 2026), per-symbol pages carrying full monthly return tables, and portfolios that can be static, built from dated transactions, or driven by an imported NAV series.

It is one person's site. The about page is signed by Dmitry, the founder, who states plainly that he is not a financial advisor and does not represent a financial institution; no company name, team or investor appears anywhere.

Pricing

Four tiers billed through Stripe, with the annual toggle set by default and worth 17%. No paid tier has a trial — the free plan is the trial.

Free is ad-supported and narrow in a way the marketing copy is not. The plan matrix gives it one watchlist, 15 holdings, zero private portfolios, 10 calculations a month, and history "limited to the past 5 years", while the same page's banner offers "40+ years" on every plan. Plus lifts the history cap, raises calculations to 200 and removes ads. Pro is where exports, unlimited calculations and own-data import arrive. Max exists for two optimisers and a 3,000-call MCP quota.

Data & coverage

Daily, dividend-adjusted prices, and every history is a real one. A symbol runs back to its own inception — VFINX to 31 August 1976, BTC-USD to 5 October 2009 — and a portfolio runs back only as far as its youngest holding: the site's Harry Browne Permanent Portfolio starts on 18 November 2004, the day GLD listed, despite its other three funds being older. Nothing is stitched, synthesised or extended with index returns to fill the gap before a fund existed. That is the opposite trade from the neighbours in this category, and the thing to settle before choosing between them.

Coverage is wider than US-only: ETF screeners return 1,953 listings for the UK, 2,569 for Germany and 1,196 for Canada, and portfolios carry a reporting currency of their own.

Integrations

There is no REST API. What exists instead is an MCP server on an HTTP endpoint that takes a bearer token — an unauthenticated call on 14 September 2026 returned 401 "Missing Authorization header" — exposing symbol search, metrics, watchlist management, portfolio and diversification analysis and all seven optimisers. Only the calculating tools consume the monthly quota.

Screener results export as CSV or XLSX on Pro; data goes in as a CSV of holdings or a NAV series. A "Copy page" control renders any symbol, portfolio or calculator result as labelled text for pasting into a chatbot, which is the whole of the site's AI story — nothing here generates a view of its own.

Limitations

• The free plan cannot save a private portfolio at all. Anything you keep is public.
• Five years is the free history — the cap that matters most in a category bought for long-horizon questions — and 10 calculations a month is a low ceiling to hit while learning.
• No pre-inception history, ever. A 1970s or 2000-2003 allocation study is impossible unless funds you can name were trading then.
• No Monte Carlo, no withdrawal-rate simulation, no tax or contribution modelling. The tool list is entirely backward-looking analysis; the retirement half of this category is elsewhere.
• Two vendor pages disagree on the export cap — the plan matrix says up to 2,500 items, the screener guide says up to 10,000.
• One person, funded by ads and subscriptions. There is nobody to escalate to and no stated plan for the data if he stops. Liveness is not currently in doubt: product updates dated 13 and 24 August 2026, docs 31 August 2026, live prices on 14 September 2026.

Alternatives

Portfolio Visualizer backtests asset-class series to 1972 and adds factor regression and Monte Carlo, at $30 a month billed annually. Portfolio Charts goes back to 1970 in twelve home currencies, free and account-free, on annual data. Both answer the long-history question PortfoliosLab refuses to fake. Come here instead when the portfolio is made of specific funds you actually hold, when you want daily rather than monthly or annual numbers, or when you want an optimiser an AI assistant can call. The rest of the shelf is retirement and FIRE calculators.