Web app
React + TypeScript, Vite, Tailwind CSS and Recharts. Calculations and Excel, PDF and PowerPoint exports run in your browser.
Available nowAI Integration via Model Context Protocol (MCP) — connect AI assistants directly to the Lab.AI Integration via MCP — available nowLearn more →
Professional portfolio construction & analysis
Institutional-grade portfolio construction and analysis
for the European investor (CH · EU · UK).
A structured, rules-based environment to construct, stress-test, and compare ETF portfolios — with deep risk analysis, Monte Carlo simulation, and full transparency into every allocation decision.
For educational and illustration purposes only. Not investment advice.
Monte Carlo Simulation
Full Look-Through
0Analytical Modules
0Monte Carlo Paths
0Stress Scenarios
0Supported Base Currencies
0Satellite Asset Classes
0Thematic Tilts
Who it's for
Built for professionals and serious investors who demand transparency and rigour.
Portfolio Managers
Rapidly prototype allocation structures, validate inputs against coherence checks, and export institutional-grade PDF reports and PowerPoint presentations.
Independent Advisors
Build client-specific ETF portfolios with full rationale documentation. Compare scenarios side by side and stress-test against six historical events, from the Dot-com Crash to the 2022 Rates Shock.
Sophisticated Private Investors
Move beyond generic model portfolios. Define your own constraints, understand the mechanics of your allocation, and make decisions backed by risk metrics — not gut feel.
Welcome video
See the Lab in motion.
A short visual introduction to the Investment Decision Lab.
How it works
Three steps from blank canvas to institutional-grade portfolio.
01
Define
Set your constraints — base currency, investment horizon, risk appetite, equity target, number of ETFs, exchange preference, thematic tilt, and hedging.
02
Generate
Receive a fully structured allocation with ETF-level detail, asset class weights, geographic breakdown, and a plain-language rationale for every decision.
03
Analyse
Stress-test against historical crises, run Monte Carlo simulations, compare portfolios side by side, and export a professional PDF report or PowerPoint presentation.
Four analytical modules.
Covering the full decision cycle from construction to explanation.
Build Portfolio
Rules-based allocation, starting from the global market portfolio through to ETF implementation with look-through to the underlying holdings. Explore backtesting and Monte Carlo simulation, compare historical and forecast returns and volatilities, and analyse concentration, overlaps, tail risk, stress scenarios, correlation assumptions and home bias.
Compare Portfolios
Place two portfolios side by side with the same full set of analytics as the Build Portfolio module — from look-through to the underlying holdings to backtesting, Monte Carlo simulation, historical and forecast returns and volatilities, concentration and overlap analyses, tail risk, stress scenarios, correlation assumptions and home bias.
Explain My Portfolio
Enter any ETF or combination of ETFs — your own selection, not a generated portfolio — and explore it with the same full set of analytics as the Build Portfolio module: look-through to the underlying holdings, backtesting, Monte Carlo simulation, historical and forecast returns and volatilities, concentration and overlap analyses, tail risk, stress scenarios, correlation assumptions and home bias. A plain-language explanation makes the overall structure transparent.
Try itMethodology
Documented, deterministic portfolio calculations use return and volatility assumptions that can be forecast, custom or historical, with correlations from the forecast set. Current prices, NAVs and yields are not model inputs; protected delayed-price views remain separate.
Bring your own portfolio
Already hold ETFs? Run the analysis on what you own.
The Explain My Portfolio module lets you enter any ETF — by ticker or ISIN — and receive the same institutional-grade look-through analysis as a generated portfolio. Understand exactly what you hold, what risks you carry, and how coherent the structure is.
- Add any ETF by ticker or ISIN — not limited to generated portfolios
- Full look-through to underlying asset classes and geographic exposure
- Risk & performance metrics: expected return, volatility, Sharpe, max drawdown
- Scenario stress tests: 2000–02 Dot-com Crash, 2008 Global Financial Crisis, 2011 Euro Crisis, 2015 SNB Franc Shock, 2020 COVID Crash, 2022 Rates Shock
- Monte Carlo simulation across your investment horizon
- Plain-language rationale for every position in the portfolio
Live in the Lab
Structured output, not a spreadsheet.
Every portfolio generated includes a full breakdown across asset classes, regions, and instruments — with the rationale to back it up. Exportable as genuine PowerPoint or PDF for client presentations or investment committee review.
- Investor Profile Summary — Currency, risk, horizon, equity target, ETF count
- Target Asset Allocation — Cash, Bonds, Equities, Satellites with precise weights
- Geographic Breakdown — US, Swiss, European, EM, Japan, UK equity exposure
- Allocation by Bucket — Per-ETF holdings table with asset class and region
- Risk & Performance Metrics — Expected return, volatility, Sharpe ratio, max drawdown
- Scenario Stress Tests — 2000–02 Dot-com Crash · 2008 Global Financial Crisis · 2011 Euro Crisis · 2015 SNB Franc Shock · 2020 COVID Crash · 2022 Rates Shock
- Monte Carlo Simulation — Probabilistic range of outcomes over your investment horizon
- PowerPoint & PDF Export — Genuine PowerPoint files and standard or detailed PDF reports
Rules-based
Every portfolio follows a consistent, documented set of construction rules — not discretionary judgement.
European focus (CH · EU · UK)
Optimised for Swiss, EU, and UK investors — with home-bias multipliers and withholding-tax drag per base currency.
Scenario-tested
Portfolios are stress-tested against six historical crises and market shocks before you commit to a structure.
Coherence-validated
Inputs are checked for structural coherence before any portfolio is generated. No silent errors.
Monte Carlo Simulation
Project portfolio outcomes across thousands of simulated paths. Visualise return distributions, drawdown risk, and probability of meeting your investment goals.
Full Look-Through
Allocations are decomposed through ETF holdings to the underlying asset class and geographic exposure — applied across all risk calculations, stress tests, and portfolio breakdowns.
Your house view
Your assumptions.Your investment process.Your benchmark.
Tailor the Investment Decision Lab to your investment philosophy – from capital market assumptions, risk-free rates and inflation to asset allocation rules, regional tilts, ETF selection and concentration limits. Define your own global equity market benchmark and shape how portfolios are constructed, analysed and explained.
Your house view. Your rules. Your investment decisions.
Try it in MethodologyYour investment process, step by step
Inputs you can edit live in this view — values are stored locally and override the defaults everywhere in the app.
1. Asset Allocation: equity, cash, bonds & satellites
Rules editable2. Global Equity Benchmark (MSCI ACWI Proxy)
Weights editable3. Capital Market Assumptions: returns & volatility
μ / σ editable4. Risk-Free Rate & Inflation (per base currency)
Editable5. Regional Equity Weights: Home Bias & Tilts
Home-bias editable6. ETF Catalog & Overrides
Locally overridable7. Concentration Triggers
Triggers editableHistorical backtest
Test your portfolio against the market it actually lived through.
Daily adjusted-close prices and FX histories provide a clear record of how a selected ETF portfolio performed across its shared available history.
Historical portfolio performance
CAGR, annualised volatility and maximum drawdown
Currency effects in your selected base currency
Holding-level return contribution
Rolling 1-, 3- and 5-year outcomes
COVID sell-off and 2022 rate / inflation shock
The Monte Carlo range is overlaid on the same historical period, so model assumptions can be compared directly with what actually happened.
Available in the Lab — securely protected by passphrase
Price and FX histories are sourced from EODHD. The current licence does not permit publishing or redistributing data derived from those histories to third parties.
Return assumptions
Historical vs. Expected Returns — The Best of Three Worlds?
The past does not necessarily predict the future, and forecasts are not facts either. Expected returns and volatilities are fully adjustable: our market outlook need not be yours.
Three perspectives
01
Forward-looking assumptions
Long-term expected returns and volatilities.
02
Your own expectations
Fully customisable assumptions reflecting your market outlook.
03
Historical performance
Annualised returns and volatilities from up to 10 years of ETF prices.
Switch between them and compare the impact on portfolio construction, including Sharpe-ratio optimisation and thematic tilts, on Monte Carlo future outcomes, and on risk metrics.
Good decisions start with understanding the assumptions behind the numbers.
Screenshot: historical figures are measured returns and volatilities of reference funds, not forecasts. Correlations remain from the forecast set.
Transparent by design
Core portfolio calculations are rule-based, deterministic and run in your browser. Return and volatility assumptions may be forecast, custom or historical; correlations remain from the forecast set. Current prices, NAVs and yields are not model inputs. Optional market-data and backtest views, unlocked with an access code, show delayed prices on request for information only. The app also uses a catalog database and backend functions. Model assumptions and formulas are documented.
AI + Lab workflow
Two Portfolios. One Intelligent Comparison.
Deterministic finance meets AI intelligence.
The Lab generates a fully deterministic, rule-based portfolio proposal. A one-click Copy AI Prompt gives you a CFA-level briefing to paste into any AI — ChatGPT, Claude, Gemini, or any other. The AI responds with its own recommendation, including a plain-text import block you paste straight back into the Lab's Explain module. The Compare tab shows both side-by-side across every key metric.
01
Build Reference Portfolio
Create a rule-based portfolio tailored to your risk profile, horizon, currency, and preferences.
02
Generate AI Briefing
One click creates a CFA-level prompt for any AI model — ChatGPT, Claude, Gemini, or any other.
03
AI Creates Proposal
AI returns a portfolio proposal with a plain-text import block ready to paste.
04
Import into the Lab
Paste the block. The Lab automatically maps ETFs, allocations, and data.
05
Compare Side-by-Side
Both portfolios across all key metrics and scenarios — expected return, volatility, Sharpe, Monte Carlo, cost.
06
Decide with Confidence
Two independent perspectives. Zero manual data entry. Better decisions.
Available now
AI Integration via Model Context Protocol (MCP)
The Investment Decision Lab includes an MCP server for constructing portfolios from scratch, analysing portfolios from Build or Explain and making what-if changes in a natural-language conversation. Specify a base currency, risk profile and investment horizon, or build a portfolio around a target return.
Access by invitation code
An invitation code is required. Request access at info@bicon.li.
Works in Claude and ChatGPT. In ChatGPT, a paid subscription with Developer Mode is required.
Investor
Asks questions in natural language
Conversation
AI assistant · Claude or ChatGPT
MCP host / client. Orchestrates the tool calls across servers.
Separate MCP connections. The servers do not call each other.
Investment Decision Lab MCP server
- Portfolio construction, analysis and what-if changes
- Risk analysis
- Stress tests
- Monte Carlo simulations
- Portfolio comparisons
- Reports and presentations in the app's design
- And more
Market data MCP server
- Historical prices
- Market data
- Fundamentals
- And more
Brokerage MCP server
- Order preparation
Human approval and controls
Required before any hypothetical execution
Execution (hypothetical, not available in the Lab)
Professional investors can:
Connect Claude or ChatGPT to the Investment Decision Lab and work with your portfolio in plain language. Every figure comes from the Lab’s own engine, not from the AI.
What you can ask
Build a portfolio
Start from scratch in plain language: “Build a portfolio in CHF, High risk profile, 15 years.” You can also build around a target return.
Overview
Return, risk, Sharpe ratio and costs at a glance.
Capital market assumptions
Review the assumptions behind a proposal.
Monte Carlo
The range of outcomes over your horizon and the probability of a loss.
Savings goal
The monthly amount, start amount or years needed to reach a target.
Target return
The lowest equity share that reaches a return you set.
Stress test
How the portfolio would have done in past crises, from the 2008 financial crisis to the 2022 rate shock.
Scenario in words
Describe a scenario, such as tariffs plus an oil shock, and see its impact.
Market context
Today’s rates and inflation next to the model’s assumptions.
What if
Change the risk profile, the horizon or a fund, and compare the results.
Compare and rebalance
Compare two portfolios side by side and get a memo on moving from your depot to a target.
Depot statement
Turn a PDF, picture or pasted statement into a portfolio.
Reports
A client report, a PowerPoint deck in the Lab’s design, a one-pager, or an Excel order list.
Each result opens back in the app with one link.
The AI can display and critically review capital market assumptions, but cannot change them in chat. To change assumptions, update them in the app's Methodology and then pass the updated portfolio to the assistant again.
One integrated chat. No more copying and pasting between tools.
Analyse portfolios, review assumptions and interpret results in one natural-language conversation – supported by the Lab's quantitative models rather than generic AI-generated estimates.
The architecture also enables integration with external systems, including market data providers and brokerage platforms, through additional MCP servers. This creates the potential for seamless workflows from market data retrieval and portfolio analysis to optimisation, order preparation and approval.
MCP in practice
Two examples of analysis and preparation in a single AI conversation.
Compare a portfolio with a benchmark
A portfolio compared with an 80% ACWI / 20% CHF government-bond benchmark, with historical performance, volatility, drawdown and other risk metrics explained in the chat.
View full-size screenshot
Prepare an order list for review
An illustrative CHF 100,000 order list combines target weights from the Lab with delayed external prices, whole-share quantities, remaining cash and checks before placing orders. External data connections are separate from the Lab; this is preparation, not trade execution.
View full-size screenshotIllustrative outputs, not investment advice. Historical performance does not predict future returns.
From analysis to execution – with transparent methodologies, appropriate controls and human oversight.
Important notice: Connections to external market data providers and brokerage platforms are technically possible today through separate MCP servers but are not part of the Investment Decision Lab's functionality.
Full Look-ThroughCurrency HedgingMonte Carlo SimulationCVaR AnalysisPaginated PDF ReportsScenario Stress TestsVast Allocation BreakdownsTail RealismExcel ExportsWithholding Tax DragFull Look-ThroughCurrency HedgingMonte Carlo SimulationCVaR AnalysisPaginated PDF ReportsScenario Stress TestsVast Allocation BreakdownsTail RealismExcel ExportsWithholding Tax Drag
Independent assessment
Independent Review by Claude Cowork
Can a free, browser-based investment platform deliver capabilities normally associated with professional portfolio construction tools?
This independent review by Claude Cowork takes a deep dive into the BICon Investment Lab, assessing its portfolio analytics, Monte Carlo simulations, ETF portfolio construction framework, AI-assisted workflows, and overall user experience.
The verdict: a uniquely comprehensive platform that combines quantitative portfolio analysis and advanced AI-supported decision-making in a way that is rarely found in a single solution.
Explore the full analysis to discover what sets the BICon Investment Lab apart.
No other publicly available product combines rule-based UCITS ETF portfolio construction, deep quantitative simulation (with fat-tail distributions and crisis correlation regimes), look-through decomposition, currency hedging modeling, WHT optimization, and professional-grade export in a single zero-friction interface.
The prompt engineering in bicon.co is genuinely sophisticated for a free tool. The dynamic constraint injection, mandatory ETF validation guardrails, two-mode design, and structured import round-trip represent a well-thought-out human-AI workflow.
Behind the Lab
In short: the tech stack
Hosting
Vercel hosts the app and runs three Node.js/Express functions: admin interface, justETF fund look-ups by ISIN, and EODHD market-data prices.
Catalog database
Supabase PostgreSQL. The app reads via its read-only public API; a shared library validates all writes for the admin console and scheduled jobs.
Automatic runs
Scheduled GitHub Actions run Node.js scripts that read justETF’s public fund pages.
Code and tests
pnpm monorepo; Vitest unit tests and Playwright end-to-end tests in a real browser.
Three parties, three doors
Each route has its own access controls. All three lead to the same Postgres database.
End users
Browser
Read only
Supabase REST
anon key · RLS
Site admin
/admin
Token
/api/admin/*
Vercel function
Scrapers
GitHub Actions
Nightly · weekly · monthly
Node.js scripts
justETF public fund pages
Postgres
One source of truth
Release Process
How a change reaches the live app
Every change to Investment Decision Lab goes through the same quality path before it reaches you.
Design first. New features start as a short written design that is approved before any code is written.
Separate workspace. Each change is developed on its own branch, never directly on the live version.
Automated tests. Thousands of unit, interface and end-to-end tests run during development, and again automatically on every update. A change with a failing test cannot be released.
Independent reviews by two AI systems. Several Claude reviewers (Anthropic), each with its own focus such as correctness, security or language, check every change. OpenAI's Codex then reviews it again as a second, independent reviewer. Every finding is fixed or explicitly answered, and Codex reviews again until it finds nothing more.
Preview before release. Each change can be tried on a protected preview site before it goes live.
Controlled database changes. Changes to the data structure are tested in a dry run first and carried out by the owner personally.
Release by approval only. Nothing goes live without the owner's explicit approval. The live version is protected against direct changes.
Fresh data. The fund catalog is updated continuously, and the built-in reference copy is refreshed every week.
Check after release. After each release an automatic check confirms that the live site works.
The result: Often several releases per day – without compromising on quality controls.
Agentic Software Development does not mean less governance. On the contrary: today we can automate quality processes more rigorously than ever before.
And it is precisely this combination of speed, quality and human accountability that makes the difference.
Interested in a professional dialogue?
The Investment Decision Lab is a non-commercial technology showcase — not a product or service. If you want to discuss the methodology, explore the analytical capabilities, or have a professional conversation about portfolio construction, the BICon team is open to dialogue.