The $10,000 notional account

One account size is open, on two markets: prediction rounds, which this page describes, and US stocks, which is the same account with the same numbers on a different instrument. Larger sizes ship when the market you trade is deep enough to support them.

#What you pay and what you get

The challenge capital is the same as the seat. That is deliberate. You are evaluated at the size you will actually trade, so a pass means something about the seat rather than about a scaled-down rehearsal.

It is also fixed by the account rather than being a snapshot of whatever your practice balance happened to be that day. Snapshotting a drifting balance meant every challenge started from a different number, so a drawdown limit and a profit target were not comparable between two traders or between two runs by the same trader.

$10,000 is a notional, and it is worth being exact about what that means. It is the size your limits, your drawdown floor and your profit target are all measured against, and it is the number the seat is labelled with. It is not a claim that ten thousand dollars of cash sits idle in your seat account waiting to be lost: Fortell funds the seat with what the limits below can actually put at risk, and the most a seat can lose is the $250 of Fortell's own cash reserved against it. The first-loss buffer is where that money is, and Checking it yourself shows you how to read the account.

#The four bars you have to clear

All four, at the same time. There is no partial credit.

In money: +0.5% of $10,000 is +$50, and 2.5% of $10,000 is $250. The profit bar is deliberately smaller than the loss budget: you have to earn less than you are allowed to lose.

The last row is the one that behaves differently from the other four, so it is worth a sentence on its own. It is a bar to clear, not a rule to breach: going over it does not end your run and does not fail you, exactly like being short of scored calls. You keep trading and it falls as your other days earn. It falls by making profit on another day and never by losing on one, because a losing day shrinks the total and pushes the share up. A day here is the same day your loss cap resets on, and a day's figure counts positions that closed in it, so an open position contributes nothing until it settles. Until a run is in profit overall there is nothing to concentrate and this row is met.

How a challenge is graded explains what counts as a scored call, how the forecast score is calculated, and how drawdown is measured. Read it before you assume any of these mean what they mean at another firm, because two of them do not.

#The limits you trade under

The same limits apply during the challenge and on the funded seat. That is the point: a funded trader must not be able to do the thing the evaluation forbade.

The risk rails explains what each one is for and exactly when it refuses you.

#The profit split, if you pass

The pool takes no share of what a funded trader makes. That is the whole of it: your profit is split between you and Fortell, and the capital standing behind the seat is not paid out of it.

That is a fact about this account rather than a rule the protocol enforces. The split is part of the terms a challenge is sold under: it is fixed at purchase, written into the seat account when it is funded, and checked against the published record. An account added later may split differently without touching either of these, and neither a table edit nor a redeploy can move a split a seat already holds.

Getting paid covers when you can take it.

#Why these numbers and not others

Nothing above is arbitrary. Each number falls out of another one, and it is worth seeing which, because it tells you what would have to change for the deal to change.

#The first-loss cover is the maximum you can lose

$10,000 x 2.5% = $250.

That is not a coincidence, it is the definition. Your seat's drawdown floor sits at its starting size minus its maximum drawdown, and the software refuses any trade that could take you below it. So the very worst a seat can lose is exactly the cover Fortell reserved for it, and Fortell's own money absorbs all of it.

This is what turns "backer capital is untouched" from a promise into arithmetic. It only works while the cover is the whole of the maximum: reserve less than a seat can lose and the remainder does not go away, it lands on backers. An earlier version of Fortell reserved less than the maximum, and the code still carries the note recording what that cost.

A smaller percentage is not a looser rule. Read the drawdown against the money, not against the label. 2.5% of a $10,000 notional is $250 of Fortell's own cash at risk per seat, which is more than the older, smaller accounts put up, not less. The percentage fell because the label rose.

#The daily loss limit is half the maximum

$125 is exactly half of $250. Reach it and the run is over, so the rule is that you have to lose your whole budget twice as slowly as you are allowed to lose it once. That ratio is the standard one: the largest prop firms set the daily at half the total, and this is that.

On the funded seat it is half of what you START THE DAY WITH, not a fixed $125. There is no fixed-dollar daily field on a seat at all. The program takes the balance the seat is flat at when the day opens and allows half of it to be lost, so the figure trails your account in both directions: $250 on the morning you are funded gives $125, $350 after banking $100 gives $175, and $150 after losing $100 gives $75. The ratio above is what stays constant; the dollars are what move. Half of what you actually hold is the same rule at every size, which is what makes it a pace rather than a number - and the cost is that a bad day charges you twice, once in the loss and once in a smaller allowance the next morning.

During the challenge you see the flat dollar figure. It is one ratio expressed in whichever unit each layer can work in.

#The size limits are priced against the loss budget, and bounded by the venue

• One round can be at most $50.

• Everything open at once can be at most $200.

• So one round is never more than a quarter of your total open exposure.

The round cap is set by what the profit target asks of you. At $50, the +$150 you have to make is five net winning calls. At the $20 this page used to carry it was twenty of them, on a venue whose rounds are coin flips with a spread to cross, which is a different challenge wearing the same words. The cap is what decides whether the target is a forecast or an endurance test.

The venue is what stops it going higher, and it is a price limit rather than a size limit: a full-size order fills across most of the range you are allowed to enter at, but not at the very cheap end, where the venue has to put up far more than you do to create your position.

The open cap is a backstop and is meant to be one. It is deliberately set above what the daily loss limit allows on a flat day, because a limit that binds before the daily limit does would silently replace it. An earlier version of this page had it at $100, which was tighter than the daily allowance and therefore the only limit that ever refused an order: the daily loss limit you were sold could not bind, and neither could the one extra trade it exists to let you take. The risk rails says what each one refuses and when.

You can therefore hold at most four full-size rounds at the same time, and if every one of them went to zero simultaneously the loss would be $200, which is 80% of the first-loss cover. The limits bound the swings the cover has to absorb, rather than trusting you not to concentrate.

The daily loss limit is tighter than that and normally reaches you first: it reserves the worst case on what you already hold, so starting a day flat you can have three full-size rounds open, or several smaller ones adding up to $175. Four at once needs a day already $25 or more in profit. Both size limits count what is open, so closing a position hands its allowance straight back and re-entering the same round later is an ordinary thing to do.

#The entry fee against the cover it commits

$79 commits $250 of Fortell's own cash.

79 / 250 = 31.6%

That is the pass rate at which Fortell breaks even, if every single funded trader blew their entire maximum drawdown. The margin against it is real but it is not enormous, which is why the measured pass rate below is published with its conditions rather than as a headline.

#How hard is this, actually

Fortell models it rather than guessing, and publishes the result with the conditions that produced it, because a bare percentage from this model is not usable.

A trader with no edge at all, playing the rules as well as they can be played, passes 10.28% of the time. The same trader trading blind passes 0.04%. That is 25,000 simulated 14-day challenges behind each figure, measured on 2026-08-24 against the rails that ship, on 1d buckets. It is the rate under those rules on that clock and nothing wider: when a rail moves, the figure is measured again rather than carried forward.

Those two numbers are easy to confuse and they mean opposite things. 10.28% is the CEILING: what someone with no forecasting skill at all can reach by playing the rules perfectly - sizing to every published limit and stopping the instant they qualify. 0.04% is the FLOOR: the same zero skill, trading blind. Neither is a prediction about you, and 10.28% does not mean "about one in ten people pass". It means that luck alone, played optimally, gets you there roughly one run in ten - which is the number we have to size our own money against, not a forecast of the people who buy.

Read the ceiling as a band, not a point: this figure resolves to 0.93 points. It is not calculated. It is the best play a search over trading strategies found, and a search is a draw: run it again from different starting points and it lands somewhere else. Ten independent runs of that search landed between 9.62% and 10.54%, and that spread is the resolution above. It is published because we once reported a 0.70 point move in this number as a rule change when 0.68 of it was the search. A difference bigger than the resolution is something real. A difference smaller than it is our instrument, and it is evidence for nothing in either direction - including for the idea that a rule which moved this number by nothing was worth nothing.

The band is not symmetric, and the direction matters. The ceiling is a lower bound on a maximum: more searching can only raise it, never lower it, so a run finding something worse than 10.28% is the search failing to find the best play rather than evidence the real ceiling is lower. Four of the ten replicate searches did find better - 10.54% at the best of them - so the honest reading is that 10.28% is what our published method finds and the real ceiling is at least a little above it.

The profit target was lowered from $150 to $50 on 2026-08-18 and this figure barely moved, which is not the same as the change being small. It was 10.41% before and is 10.28% now, a gap far inside the 0.93-point resolution above. Hold the trading strategy fixed instead of letting the model re-optimise, and the same trader goes from 4.50% to 9.98% - the easier target slightly more than doubles the pass rate of anyone who does not change how they trade. The headline moves by nothing because the model, given the harder target, simply finds a different way to play that gets almost all the way back. A real trader who does not re-plan gets the larger effect.

Read the headline as a ceiling, not an estimate. It assumes a trader who knows the rules, sizes to them, and stops the instant every criterion is met, which the grader makes permanent from day 7 onward. A trader who keeps going after qualifying does worse. The 0.04% floor is a floor rather than an estimate for the opposite reason: it ignores the maximum buy size the product tells you, so a real client that reads that number places orders this model loses.

The honest reading is that both of those brackets sit under the break-even rate above, and that the gap is the margin the business runs on - though at 10.28% against a 31.6% break-even the margin is about threefold, not enormous. Given real forecasting skill the criteria separate: at two cents of edge per call the same trader passes 12.72% of the time; at five cents, 98.60%.

Read the edge figures against the venue's own spread, which is about four and a half cents. Two cents of edge does not cover the cost of crossing a book that wide, which is why the two-cent column is only a little better than luck; five cents does, and the curve jumps. An edge figure quoted without the spread it was measured against is not comparable.

This figure has been published wrongly ten times in this project's history, each time stated confidently. It is now generated from a single measured run rather than typed, and a build gate refuses to ship the site if the number on the page and the number in the record disagree. Do not quote it without its conditions, and do not compare it to a figure measured under different rules.

#The window is 14 days, and a pass can come earlier

Your challenge runs 14 days. If you meet all five criteria and 7 days have elapsed, you pass then rather than waiting out the clock. A pass, once recorded, is permanent.

A breach works the other way and is immediate. Blowing the drawdown limit or the daily loss limit ends the challenge the moment it is detected, not at the end of the window. That is covered in How a challenge is graded.