LessonThis is a lesson restated in new words. Any figure in it is an example from the source, not this desk's record.
A proprietary-trading challenge can be a fee business wearing a trading costume. In the lecture’s own attempt, an $89 “fast” simulated test required a 10 percent gain with a 5 percent maximum loss. The account failed. A single logged loss was $4,140. The firm lost nothing on that failure, because the trading was simulated and the fee was real. Scaled to 100,000 traders each with a generous 50/50 chance, the same structure collected $8.9 million in fees and, in that illustration, netted about $2.72 million after real-capital losses and gain splits. A roulette comparison in the same talk netted about $2.51 million.
Why it matters
The customer experiences a chart, a target, and a diary of setups. The firm experiences a filter. Under the 50/50 assumption, about 66,515 of the 100,000 fail the first stage and about 33,485 continue. A second similar cut removes about half of those who remain, leaving on the order of 16,000 to be considered for real capital and about 16,000 gone. Tens of thousands of people have been removed before the firm has to trust them with money that can actually be lost.
The idea in plain words
Under a generous 50/50 assumption, about 66,515 of 100,000 entrants fail a first stage and about 33,485 continue. A second similar cut leaves on the order of 16,000 people to consider for real capital. The firm has removed most of the crowd before it has to trust them with money that can actually be lost. Fees in that sketch are $8.9 million.
After funding, the drawdown rule still ends accounts, so real losses begin, and the firm keeps a share of the rare large winner. The lecture’s combined illustration still shows about $2.72 million of firm gain, close to a roulette sketch that netted about $2.51 million. A winner who made $60 million, with the firm owing a 10 percent share in the talk’s hypothetical, would hit that bottom line if the firm paid. The uglier branch is a terminal that never sends the trades to a real market and then cannot pay.
The law of total expectation makes the same point as an inequality. Simulated failures contribute the fee, not a trading loss. Using the sketch’s roughly 66 percent loser rate, one piece is $89 times 0.66, and about 34 percent pass. The firm cannot budget the unknown gains of real winners, so the safety sum keeps the expected real-capital loss and ignores the fantasy win. The expression has to stay positive. The knobs the firm actually holds are the fee and the drawdown cap. Set both tightly enough and the average stays positive. That is a house edge. Refusing payouts is the version that was never a casino with a cashier, only a fee.
How to use it
Before you pay, separate simulated stages from live capital in writing. Ask who loses if you lose, and who pays if you win. If the first answer is “only you” and the second answer is vague, you are buying a ticket.
What usually goes wrong
Treating a failed simulator as shared pain. Journaling a “fair value gap” as if the fee were tuition in a craft. Assuming a gain split will be honored by a firm whose economics improve if it is not. And copying a payout screenshot that cannot be tied to a real fill.
Questions people ask
What did the $89 test require?
A 10 percent gain target and a 5 percent loss limit, on a simulation. The lecture’s account failed. One logged loss was $4,140, and the firm’s trading loss on that failure was zero.
Where do the $8.9 million and $2.72 million come from?
They are the lecture’s scaled illustration: 100,000 entrants at $89, a 50/50 pass assumption, and a net after real losses and gain shares. They are not an audited industry total.
Can a challenge firm be built to pay winners?
In the lecture’s inequality, yes, if fees and drawdown caps keep the average firm result positive and the firm actually has the capital. A terminal that never trades, and then blocks payouts, is the version that cannot.