How Ember picks

How much goes into stocks

When you need the money, how much of a loss you could live with, and whether you'd lean on this money in an emergency each set a limit on the share of your money in stocks. Ember uses the lowest of the three.

Ember averages your two answers about losses and rounds down to a whole 5%. If you answer only the first one, it uses that one. Each loss answer points to the stock share whose sharp-crash estimate, explained below, comes closest to that loss.

Taking money out each year keeps at least 8% in cash. That covers about two years of withdrawals at 4% a year, so a drop doesn't force you to sell stocks.

Whatever isn't in stocks is split between bonds and cash. The sooner you need the money, the more of that part stays in cash. Choosing Some for companies outside the US puts 20% of the stock money in an international fund, and More puts 40% there.

Funds or individual companies

Your answer to “Funds or individual companies?” decides how the US stock money is split. Funds only puts all of it in a broad fund. Stocks only puts all of it in individual companies. International, bonds and cash always use funds.

• Base US fund: VTI, or the screened ESGV when you avoid any industry.
• International: VXUS, or the screened VSGX when you avoid any industry.
• Bonds: BSV for money needed within five years and BND otherwise. When you avoid any industry, VGSH and VGIT take their place, since they hold only US Treasuries.
• Cash: SGOV, a fund of Treasury bills that mature within three months.
• Focus funds: when you pick a focus, 40% of the US fund money is split among up to three focus funds (SCHD, VUG, VTV, VB, QUAL and USMV). They're skipped when you avoid industries, because they may hold them.
• Sector funds: with funds only, the sectors you choose get 25% of the US fund money, through Vanguard sector funds.

How companies are scored

Each company is ranked twice on the figures below, once against the other companies in its sector and once against all 1,490. A rank is the share of companies it beats, from 0 to 1. Missing figures count as average.

The two ranks are blended differently for each measure. Comparing within a sector stops whole sectors from winning a measure by nature. Banks always trade at low prices compared with their earnings, so without it a value portfolio would be mostly banks. Someone after dividends or calm prices wants them wherever they are, so those measures lean on the whole list.

A company's score adds up these parts:

• 45% for how well it matches the focuses you picked. With no focus picked, this part is half quality and half company size.
• 25% for how close its beta is to a target that depends on how much of your portfolio is in stocks. This part reaches zero 0.7 above the target, or 1.4 below it, because a company that swings less than the target only makes the portfolio calmer.
• 20% for quality, in every portfolio.
• 10% for company size, which favors larger companies unless you chose smaller ones.
• A bonus worth 15% of a full score for companies in sectors you chose.
• A small nudge worth up to 4% of a full score either way. It's what Shuffle changes, and it stays the same until you shuffle again.

Choosing Steadier prices caps the target at 0.80. Choosing Dividends leaves out companies that don't pay one. Portfolios with 40% or less in stocks also skip companies that lost money over the last year.

Companies whose price fell more than 75% over the last year are left out of every portfolio. Their figures are usually out of date, after a company splits up or runs into trouble, and a yield worked out from the old dividend looks far too high.

Keeping picks spread out

• No more than 25% of the picks come from one sector, or 40% for a sector you chose. All the sectors you chose together get no more than 60%.
• The picks cover at least 5 sectors when there are 10 or more of them.
• No more than 2 come from one industry, or 3 when Ember picks 15 or more.
• No company gets more than 10% of the portfolio. Ember adds companies when it needs to.
• Higher scores get somewhat more money. The top pick gets up to about 1.7 times as much as the last one.

What avoiding an industry removes

Each company carries an official industry code (GICS). Avoiding an industry removes every company with a matching code, and switches the funds as described above.

The numbers on your results

• In a sharp crash multiplies the portfolio's beta by one third. It assumes US stocks fall by about a third, as they did in early 2020. Beta measures how much a holding has moved with the market. A beta of 1.2 has tended to move 20% more than the market. Real crashes can be worse.
• Dividends from your companies uses each company's current yearly dividend. Fund payouts aren't included, because the data doesn't cover them.
• Stocks over the last year is the price change of the stock holdings over 52 weeks, without dividends.

Where the data comes from

The company list and industry codes come from Wikipedia's lists of the S&P 500, S&P MidCap 400 and S&P SmallCap 600. Prices and company figures come from Finnhub and are from October 23, 2026. The site doesn't fetch prices while you use it, so every result uses the same snapshot.

What Ember leaves out

• Taxes, fees, account types and the rest of your finances beyond the questions in the builder.
• Rebalancing. A real portfolio drifts as prices move and needs adjusting over time.
• Predictions. Every figure describes the past, and none of it promises anything about the future.

Ready to try it? Build a portfolio