Funded death rates 14.2%–75.2% among profiles one evaluation grades within 2%

An evaluation is a race between two lines. Reach the target and it ends well; touch the floor and it ends. Both lines are terminal, and every decision inside an evaluation is about which one you arrive at first.

A funded account keeps the floor and deletes the target. What it puts in the target's place is not another line but a set of conditions on taking money out: an amount of profit that has to be standing before a payout can be requested, a schedule saying how often requests are allowed, a quantity of trading required before the first one, and a split on whatever is finally withdrawn.

None of those is a finish line. There is no state a funded account can arrive at that ends it well. It can be extracted from, repeatedly, until it can't — and the whole of the difference between the two phases follows from that one structural fact.

How a floor is set, and whether it climbs behind the account as the account climbs, is not this page's subject. The line that closes a trailing-drawdown account covers the mechanic, and the tick-by-tick version records, in a single sentence and then leaves alone, that the freeze most traders quote is a property of the funded account and not of the evaluation. This page begins where that sentence stops. Here the floor is a number the reader sets; what is being examined is everything built around it.

Money inside a funded account is not money yet. It is a claim that survives exactly as long as the account does, and every rule about payouts is a rule about how long that claim has to stay standing.

The question can be answered rather than argued. Take a large set of trader profiles — every combination of a winning month, a losing month and a mix of the two across a wide grid — and score each one twice. Once against an evaluation: reach a target of 8% of account size before touching a floor of 6%, inside twelve months. Once against a funded year that keeps the same floor and replaces the target with payout rules. Then ask whether the first score predicts the second.

It does, in the aggregate, and it would be dishonest to pretend otherwise. Across 547 profiles the rank correlation between the evaluation score and the funded-year death rate is -0.72: profiles that pass more often do die less often. A pass is evidence.

What it is not is a measurement. Among the 11 profiles this evaluation grades between 69% and 71% — a spread of 2%, well inside the noise of any real sample of trades — the share of funded accounts closed inside a year runs from 14.2% to 75.2%.

Two traders the evaluation cannot tell apart can hold funded accounts that die at 14.2% and 75.2%. The grade is informative about the population and close to useless about the individual.

The spread has a direction. Hold one trader's shape completely fixed — the ratio between a winning month and a losing one, and how often each occurs — and change only how much of the account rides on it. Nothing about the strategy moves. Only the scale does.

From a winning month of 4% of account size upward, the evaluation stops responding. Its pass rate covers a range of 8.4% across that whole stretch, and it does not fall. Over the same stretch the funded account's death rate goes from 25.9% to 75.2%.

The mechanism is the missing line. In an evaluation, trading larger carries you faster toward the target and faster toward the floor, and past a certain size those two effects very nearly cancel: you are more likely to hit something, and the ratio of which thing barely moves. In a funded account there is no target for size to carry you toward. The only line left is the one underneath, so the same increase that was neutral becomes one-directional.

An evaluation cannot price size, because size moves a trader toward both of its lines at once. A funded account has one line, and prices it exactly.

This is a sharper claim than the familiar one about habits — that a trader gets used to pushing for a target and carries the reflex across. The reflex may well exist. What the arithmetic shows is that it was never tested. The evaluation did not reward the size and did not punish it either; it was indifferent, and indifference is what a trader carries into the account that is not.

An account's equity begins at its opening balance and moves in exactly two ways: the trading changes it, and withdrawals remove from it. Nothing else touches it. So the profit the trading produced over a year has to equal what was withdrawn plus whatever is left inside, and what was withdrawn splits into the trader's share and the firm's. Four parts, and they are not an estimate — they are the only destinations that exist.

At the profile and rules this page declares, the trading produced 9.29% of account size across the twelve months. 11.01% reached the trader's bank. The firm's share of what was withdrawn came to 1.22%, and 0.51% was still sitting inside accounts that survived to the end.

Those three come to more than the total. They have to be brought back by the fourth: 3.46% of equity surrendered by the accounts that hit the floor — because an account is permitted to finish below the balance it started with, and that shortfall is not the trader's to make good.

More money reached the trader than the trading produced, by 1.72% of account size. The floor that closes 47.8% of these accounts is also what pays for the difference, and what it pays with is the firm's capital.

This is the property that makes a funded account a different instrument rather than a harder evaluation. In an evaluation the trader's own money — the fee — is what is at risk, and the floor only ever takes. In a funded account the money at risk below the floor belongs to the firm, and for the trader the floor changes sign.

Two things this does not say. It is the arithmetic of one account considered alone, and it excludes entirely what the account cost to obtain: the fees, the failed attempts and the months, none of which is small. And it excludes every rule that caps a withdrawal rather than delaying it — consistency tests applied at payout, ceilings on a single request, scaling plans. Those rules exist, they all reduce what comes out, and none of them is in this model.

The rules can now be priced one at a time, because the instrument below answers the whole question for any setting of them. It is not a simulation with a seed: it walks the distribution of account states forward through the year and reports the exact average.

One control does double duty, and not as a convenience. The profit a trader has to leave standing is both the gate on a payout and the cushion the account has left afterwards, because with a floor pinned at the opening balance those are the same quantity. The firm transcribed above states it outright: a request for the whole balance closes the account, since the balance then meets the loss limit.

Three results are worth moving the controls to reproduce, because two of them are the opposite of what "payout friction" suggests.

The cushion is nearly free. Requiring nothing to be left standing — the trader draws the account back to its opening balance at every opportunity — banks 11.74% and leaves 24.8% of accounts alive at the end of the year. Requiring 2% of account size to stay behind banks 11.72% and leaves 52.2% alive. Survival more than doubles; the bank balance moves by 0.02%.

The waiting period is nearly free too. Moving the wait before a first payout from one month to six changes what reaches the bank from 11.01% to 10.95% and raises survival from 52.2% to 61%. The profit a minimum-days rule forces a trader to leave in the account is the same profit that keeps the account off its floor.

The schedule is the one that costs. Moving payouts from monthly to every three months takes the bank balance from 11.01% to 10.31% while raising survival from 52.2% to 60.3%. Unlike the other two it is a genuine transfer: the account is safer and the trader is poorer.

A payout rule that makes a trader hold profit is not friction. It buys survival with money that was going to stay in the account anyway. A payout rule that makes a trader wait to ask is friction, and it is the only one of the three that is.

Every number here comes from a model with a stated grain, and the grain is coarse in ways that matter. Each of the following is a reason to treat the death rates as a lower bound and the bank balances as an upper one.

• The floor is tested at month ends only. A real limit is tested continuously, and an account that dips through its floor mid-month and recovers by the thirty-first is dead in life and alive here. Real death rates are higher than these, and higher by an amount this model cannot estimate.
• A month has three sizes. Real months come in every size. Because a losing month here is always the same depth, the floor is crossed in whole units of it, which is why survival improves in steps rather than smoothly as the cushion grows. The steps are the model's resolution, not a property of accounts.
• No rule caps a withdrawal. Consistency tests applied at payout, ceilings on a single request, and scaling plans all reduce what comes out and none of them is modelled. Their absence inflates every bank figure on this page.
• The account is free. Nothing here is charged for obtaining the account or replacing it once it dies. The whole cost of getting funded sits outside this model, and it is the term that decides whether the enterprise is worth entering at all.
• One account, one year. A trader who loses an account and buys another is running a different problem from the one modelled, and so is a trader whose account scales.
• The evaluation is a comparator. It exists on this page to be scored against the same floor and the same clock as the funded year. It is not a model of any firm's programme and no pass rate here belongs to anyone.

And one thing the model is silent on by construction: whether a payout, once qualified for, is actually paid. That is a question about a counterparty, not about a floor, and no amount of arithmetic on this page speaks to it.

Everything argued above is derived on this page from the model this page states, and every number in it comes from the generator that ships beside it. The single exception is the rule set transcribed in the first clause, which carries its source and the date it was read.

Companion reads: which rule actually ends an evaluation, for the phase this one comes after; how much the floor mechanic alone moves a pass probability, for what the floor does when it is the variable rather than a setting; and the terms this page uses without stopping to define.

If the arithmetic rewards trading larger once funded, why do firms publish rules against it?

Because the arithmetic is the firm's problem, not the trader's. An account whose downside stops at a floor the firm funds is a call option on the firm's capital, and the value of a call option rises with volatility. Every published payout rule reduces that value: a threshold keeps profit inside where it can still be lost, a schedule delays the moment a claim becomes cash, a consistency test at payout refuses the trader who made it all in one session, a cap limits a single request. Read as friction they look arbitrary. Read as the price of the option they are the same rule written five ways.

Is this saying a funded account is free money?

No. It says that one funded account, considered in isolation from what it cost, has a payoff whose downside belongs to somebody else. What it costs to obtain — the fee, the attempts that failed, the months — sits entirely outside this model, and those terms are what decide whether the whole enterprise is worth entering. This page prices the account; what it costs to reach one is a separate question this page does not address.

Why is one part of the decomposition negative?

Because an account is allowed to end below the balance it opened with. Equity moves only by trading and by withdrawal, so what the trading produced must equal what was withdrawn plus what remains inside. For an account that hit its floor, what remains inside is a shortfall rather than a balance, and it enters the sum with the sign it has. That is also the whole finding: the shortfall is the firm's, which is why the trader's share can exceed the profit the trading made.

Why enumerate every year instead of simulating a sample of them?

Because the whole space fits. Three outcomes a month over twelve months is a little over half a million distinct years, so the exact expectation is a finite sum rather than an estimate. Nothing on this page has a confidence interval, a seed or a path count, and the two implementations that ship with it — one walking years one at a time, one propagating a distribution of account states — must agree exactly rather than approximately.