Published trading strategies, rated on what happened after they were published.
Cupel backtests every strategy it can find the same way, gives each one a plain report card, and keeps the failures on view. Then it builds portfolios from the few that held up, to grow your savings or to pay you an income.
Thesis: An even split of shares put back to even once a year
The problem
The edge mostly belonged to the past.
Across the 510 published strategies that have years since publication to judge them on, the average edge over the market was −1.2% a year before the idea was published and −4.1% after. Most of what is sold to investors shows you the before. Cupel rates only on the after.
What goes wrong
Where this industry puts fantasy first.
Much of what is sold to investors sells hope first and evidence second, because hope sells. The habits are well known, and Cupel is built to refuse each of them.
✕The backtest picked after the fact
Try enough ideas on the same years and one will look brilliant by luck alone. Showing that one is selling the luck.
Cupel shows every one it tried, and rates each on years it could not have been fitted to.
✕Only the winners on display
Abandoned and failed ideas quietly disappear, so the survivors look normal.
Cupel lists them: 42,044 of 88,746 runs of strategies' own rules could not be run at all, each with its reason.
✕Returns with the costs left out
Leaving out what it costs to trade flatters every figure.
Every run pays its commissions, and the catalogue says what is still left out: the spread, market impact and taxes.
✕An average return instead of an income
An average says nothing about whether a portfolio can pay you for 30 years through its bad stretches.
Cupel states the withdrawal and the risk of running out together.
✕One rosy path
One projected line, with the bad outcomes left off the page.
Cupel shows the range of futures, the bad ones included, and the years a portfolio was never fitted to.
✕One number that hides the rest
A single score lets a high return stand in for evidence.
The report card has no total and no rank: each question gets its own answer.
✕Quiet revisions
Figures change, and nothing says when.
Every rating carries the date it was worked out: 2026-10-16.
Our mission
Cupel helps people realistically get the most from their savings: the most their portfolio can grow to, and the most it can pay them, year after year, without running out.
The important word is realistically. Savings have to last for decades, so every number here is one we can stand behind: ratings earned on the years after an idea was made public, portfolios built only from what held up, and the whole record, failures included, kept up to date.
What we keep
We throw out almost everything.
Every strategy is run, the failures included: 42,044 of 88,746 could not be run at all, and 58,541 of the runs rated are C, unproven, however good their curve looks. Portfolios are built from what is left.
The report card
Five questions, five answers. No total, no rank.
A strategy can have real evidence and still lose money, or protect you in a crash and earn little. So each question gets its own answer, with its workings on view.
Evidence
Is there an edge beyond what index funds and cash give, since the rule was published?
Outlook
What might it earn a year? A at or above the index, B the 60/40, C cash, D below cash, F below zero. The one rating that is a projection.
Risk
How much could you lose? Its worst fall from a peak, with the dates, and the regulators' risk level beside the index's.
Protection
What did it do in the index's worst months? A hedge, B shelter, C cushion, D falls with the market, F falls harder.
Fit
Does a fifth of it leave an index holding better off, at the index's own risk?
Most similar
How closely it moves with its nearest twin. Two strategies that move together are one bet, whatever they are called.
How Evidence is earned
A letter takes years to earn, and can be lost.
One published method, the same for every strategy. Only the years after publication count, measured against an index fund, a bond fund and cash. A rule run as a guess cannot rise above B.
Two in three early leads were gone five years on, so a C+ is not a small B. A rule with evidence against it is marked D from one year. Every rating is worked out again as runs finish, and dated: 2026-10-16.
Your own strategy
Describe it in your own words. Cupel tests it the same way.
Say what your strategy buys and when. Cupel puts it into the rule its engine runs, reads the rule back to you, and tells you if it already holds the same idea. Then it is tested in both currencies with trading costs and given the same report card as everything else here. The result is published, failures included, and it starts at C: its evidence begins the day you submit it, and a letter takes years to earn.
Portfolios
Built to grow, or built for income.
A portfolio is built for one job at a time, from strategies that held up. The examples are made-up households in the United States: those still saving are built to grow, those retired to pay an income.
Built to grow
What it could become, with the bad case first: growth compounds a backtest's luck over every year.
Built for income
The yearly amount it could have paid for 30 years, rising with inflation, with a 5% chance of running out stated plainly.
The clever version is never shown alone. Every portfolio is scored again on years it was not fitted to, beside a plain equal split. Today the plain split did better in 4 of these 4 examples, and each page says so.
Who it is for
The research is free to read. Your accounts stay yours.
People managing their own money
Read the catalogue and its report cards free. Cupel Portfolios builds your portfolio for growth or for income on your own computer, optimized to the settings you choose: your accounts and balances are never sent to Cupel. Download it free, or sign up for a Cupel account to get data through the API.
Researchers and platforms
The catalogue's data through the API, the same for every caller, under a commercial licence.
Why we do this
We built it for our own money.
Cupel began as the tool we use to decide where our own savings go, and it still is. A rating that flatters a strategy would mislead us before it misled you. That is why the failures stay on the page, and why nothing is ranked or totalled into a single score.
We are just a small group of fintech experts. The financial data all of this runs on is expensive, and offering Cupel to others is how we hope to help pay for it.
If you use the free side and find it useful, pitching in helps us cover the data costs and expand what we cover.
Support the data on Polar ↗See which strategies held up.
Every strategy, its rule, its trades and its report card, with the failures beside the rest.
Every figure is from a backtest: a rule applied to past prices. Past results do not say what a strategy will do next.
A portfolio built to grow states a range of what it could become; read the bad case first.
A portfolio built for income states a pre-tax research projection, not an account-by-account retirement plan and not a measured probability of running out.
A portfolio is the best one the stated search found, not a proven optimum. Outlook is a projection; every other rating describes the record.
The example portfolios use made-up accounts: illustrations, not a recommendation for anyone. Cupel covers Canada and the United States, and no other country.