Where does your portfolio sit on the risk-return curve?

The frontier, for your holdings

Tradune samples thousands of possible weightings of the assets you own and keeps, for each band of volatility, the mix that returned the most — so you see the shape of the trade-off your own holdings offered instead of guessing at it.

Minimum-variance and maximum-Sharpe, marked

The least-volatile mix (minimum variance) and the best return per unit of volatility above the risk-free rate (maximum Sharpe) are highlighted on the curve, with your current portfolio plotted alongside so the gap is visible at a glance.

A map, held loosely

Optimizers are exquisitely sensitive to inputs you can’t truly know and will happily over-concentrate. The frontier here is built from what already happened over one window; another window gives another curve. Tradune shows it as a lens for understanding the trade-off — a tool for better questions, not an instruction.

How is the efficient frontier computed?

The same return definition the risk page uses drives every figure here, so the two pages agree on the same weights. Background: the efficient frontier, without the matrix algebra.

1. 1

   Your holdings and a window you choose

   The inputs are your current positions at market value and their daily closing prices over 1, 2 or 5 years. The window you pick is a ceiling: it shortens to the dates on which every included holding traded, and the page shows the dates actually used.

2. 2

   Price returns, with gaps shown rather than hidden

   Returns come from raw closes, so dividends are excluded. A holding with fewer than 20 daily returns, or no price history, is left out and listed with its share of your portfolio, instead of being quietly renormalised away.

3. 3

   A one-year floor

   At least two holdings and about a year (250 aligned trading days) are required. The search ranks portfolios on an annualized return, and annualizing a few weeks produces numbers nobody earned, so below the floor the page refuses rather than guesses.

4. 4

   6,000 sampled portfolios, long-only

   From the daily returns Tradune builds an annualized mean vector and covariance matrix, then samples 6,000 weight mixes spread evenly over every way of splitting 100% across your holdings — no short positions, no leverage, no per-position cap. The sampling is seeded, so the same inputs give the same chart.

5. 5

   The curve and the two marked points

   Samples are grouped into 40 volatility bands and the highest-return mix in each band becomes a frontier point; points beaten by a lower-risk band are dropped. Minimum variance is the lowest-volatility sample; maximum Sharpe is the sample with the highest (return − risk-free rate) ÷ volatility, using one app-wide risk-free rate shown on the page. Displayed returns are geometric — what those constant weights would have compounded to — while the arithmetic mean the algebra uses is reported separately.

How do I read the optimizer’s output?

A hypothetical three-stock result, with Sharpe ratios worked out at an assumed 4.2% risk-free rate so you can check the arithmetic.

In this hypothetical, a different mix of the same three stocks returned more with less volatility than the current weights, so the current portfolio plots below the curve. That describes the past window only: change the window and the maximum-Sharpe weights can change a lot, which is why the page treats them as a reference point, not a target.

To see how much of today’s risk comes from each position, pair this with portfolio risk analysis.