Which stocks you pick is only half the decision — how much of your money goes into each one is the other half.
The simplest approach — put the same dollar amount into every holding — quietly makes a series of bets nobody actually chose. It treats a stock that swings 15% a year the same as one that swings 60%. It ignores that two "different" holdings might move in near lockstep with each other, or sit in the same sector, or carry the same market beta. And it gives a name with a terrible track record the exact same weight as one with a great one. This module runs three real diagnostics against your actual basket to check for exactly that, then offers five allocation methods, each built to correct one or more of the mistakes equal-weighting makes silently.
Volatility dispersion, correlation/sector/beta clustering, and Sharpe-ratio dispersion — checked automatically against your specific holdings before you ever pick a method, not generic advice.
Equal-Weighted, Inverse-Volatility, Critical Line Algorithm, Regime-Adaptive Allocation, and Hierarchical Risk Parity all run over the exact same basket and window, so their outputs are directly comparable.
Markowitz, H. (1952). "Portfolio Selection." The Journal of Finance, 7(1), 77–91.
The original mean-variance framework — showed that a portfolio's risk isn't the average of its holdings' individual risk, but depends on how they move together. Still the basis of this module's Critical Line Algorithm.
López de Prado, M. (2016). "Building Diversified Portfolios that Outperform Out-of-Sample." The Journal of Portfolio Management, 42(4), 59–69.
Introduced Hierarchical Risk Parity — allocates by clustering holdings on return correlation first, sidestepping mean-variance's own instability when the covariance matrix is estimated from noisy data.
Every method below runs on this platform's own real engine, over your actual basket — the three diagnostics run automatically once you've added holdings, before you ever have to pick a method yourself.
The process
One staged flow, four steps — the same window and the same basket feed every method, so nothing you compare later is measured on a different basis.
Build your basket
Search and add tickers, or start from a preset. The three risk diagnostics run automatically the moment you have two or more holdings.
Set a risk mentality
Low, Balanced, or High risk appetite — a starting-point suggestion only, shown as a badge on one method tab. It never runs anything or locks you into a choice.
Set your rebalance window
Pick a rebalance frequency and estimation lookback. Every method uses the same window, so what you see next is apples-to-apples.
Analyze, or compare all five
Open any method's own tab for a full breakdown, or run Equal-Weighted, Inverse-Volatility, CLA, RAA and HRP together in one batch, side by side.
What we offer
Five allocation methods on the same basket and window — from the no-data baseline through to the two research-grounded approaches above.
Equal-Weighted
1 / nEvery holding gets the same weight, reset at each rebalance. Uses no historical data at all — this module's own baseline, and what every other method is measured against.
Inverse-Volatility
w ∝ 1 / σAt each rebalance, weight is proportional to 1 / trailing annualized volatility, normalized across the basket. The calmest holdings get the most capital, the most volatile the least.
Critical Line Algorithm
mean–varianceSolves the classic mean-variance problem on the trailing covariance matrix — Maximum Sharpe, Minimum Volatility, or any point you pick along the efficient frontier — then holds those weights forward, out of sample.
Regime-Adaptive Allocation
auto-switches by regimeAt each monthly rebalance, picks exactly one existing rule based on this platform's own macro regime read: Maximum-Sharpe CLA in an aggressive/cyclical-bull regime, Minimum-Volatility CLA in a crisis regime, Equal-Weighted otherwise — never a blend.
This platform's own design, not from published research — a rule-switcher built on top of the other four methods, not a sixth independent one. View today's regime →
Hierarchical Risk Parity
correlation clusteringClusters the holdings by return correlation, then splits capital down the tree inversely to each cluster's volatility. Returns only — no fundamentals, no settings to tune — over a longer 3-year lookback than the other methods.
Build a basket, and every method above runs on it — free to switch between them any time.