Prop-firm challenges at zero edge

A trader who picks direction with a coin flip, under the rules prop firms publish, reaches a funded account in 0% to 28% of accounts depending on how they trade. For a population like that, each dollar paid in challenge fees comes back as a median of about $0.20 in payouts. Most of the pass rates the firms publish need no trading edge to explain.

The coin-flip trader

I coded the published rules of seven firm-and-plan combinations (profit targets, loss limits and the other written conditions) and ran a trader with no edge through them: direction by coin flip, costs included, across a grid of risk levels and trading frequencies.

Across that grid, the share of accounts that reached the funded stage ran from 0% to 28%. The median cell of each firm-plan model ran from 1.7% to 12.4%. A single cell depends on instrument, risk, cost and trades per day, and some price paths are FX proxies for futures; the range is the honest summary.

What a fee dollar buys

In the same zero-edge population, the payout returned per dollar of fees had a median cell of $0.19–0.20, and up to $1.85 for some trading styles. These are averages per cell; the tail for a single account was not reported.

For the buyer, that is the arithmetic of a lottery ticket: a small chance of a large payout, a negative expectation on average.

The rules restrict variance

The cheapest way to pass a barrier with no edge is high variance: big risk on few days, many accounts at once, one-directional bets. The firms' written rules restrict exactly that: account rolling, one-directional betting, all-or-nothing trades, and "best day" or consistency rules.

In the simulation, a best-day or consistency rule cut the pass rate of a high-risk one-day strategy by a factor of 2.3 to 10 (gold, 1:2 trades, two trades a day, 3% or 5% risk). The rules are shown here next to the math; nothing here says anyone broke them.

Published pass rates

Some firms and data providers publish outcomes:

• Topstep, 2025: 16.8% of Combines passed (counted per account); 51.8% of participants passed at least once (per person); 0.71% of funded traders were called to a live account.
• FPFX Tech, 2024, across 300,000+ accounts at 10 firms (reported through a trade publication, so a secondary source): 14% passed, 7% of all accounts received a payout, and the average payout was 4% of account size.

Per-account and per-person rates are not comparable; Topstep gives both in one sentence and they differ by a factor of three.

Set against the zero-edge simulation, all but one of the published rates I compared fall inside the range a zero-edge population produces. Topstep's 16.8% is inside the range, though above 87% of the grid cells. One firm's 2025 test pass rate, 36.22%, lies above every zero-edge configuration I tried. That one is unexplained: it could be real trader skill, the proxy price path, how the firm counts, or how I read its rules.

The comparison is not blind, and the zero-edge ranges are wide. The fair sentence is that these rates need no edge to explain, not that the traders have none.

Where the payouts come from

Belgium's FSMA (7 March 2024) and Italy's CONSOB (8 July 2024) published warnings describing these challenges as simulated or demo accounts sold as paid exams. If the funded account trades no real position, a payout is a transfer out of the fees collected.

Whether such a structure needs a constant flow of new customers is a condition, not a label. If fees are collected upfront and payouts stay below a threshold share of fees (one, minus operating costs, minus the referral share), the cash never runs out even if new sign-ups stop. If payouts exceed fees, growth becomes necessary: at a payout-to-fee ratio of 1.5, the cash balance in my model turned negative by month 14 even with 30% annual growth.

The SEC's investor site defines a Ponzi scheme as an investment fraud that pays existing investors with money from new investors and needs new money to survive. The condition above tells you which side of that line a structure would sit on; without a firm's actual fees, payouts and costs, it does not tell you where any firm sits, and this post does not place any firm on either side.

Method: published firm rules as of September 2026, simulated with fixed seeds on gold and EUR/USD price paths. Code and outputs are in a private research repository and available on request.