The EV Bros Are Half Right

There's a genre of prop-firm content making the rounds that goes like this: stop babying your funded accounts. A challenge costs $100. The payout is $2,000. If you have an edge, the mathematically correct play is to size huge, pass in three trades, extract the maximum payout, and treat blown accounts as the cost of inventory. Run 100 challenges and the spreadsheet prints six figures.

I spent ten years on institutional rates desks, and I want to do something unusual with this argument: take it seriously. Because about half of it is genuinely right — more right than the play-it-safe crowd wants to admit. The problem is that the other half is wrong in a way the spreadsheet is specifically designed to hide.

Here's the honest audit.

What the EV framing gets right

First: accounts are not pets. The EV crowd's best insight is psychological, not mathematical. Traders get attached to funded accounts the way they get attached to open positions, and attachment produces the same errors — trading scared, cutting winners to protect a balance, treating the account's survival as the goal instead of the cash flow it exists to produce. On a bank desk, nobody is sentimental about a book. The book is a vehicle. Detachment is correct.

Second: time in market is a risk channel, and fewer trades can mean less risk. This one sounds backwards and isn't. Every additional trade is another opportunity to break your own rules, revenge-trade a loss, or meet a regime change. A trader who needs 300 trades to hit a target is exposed to 300 chances of being human. If the edge is real, compressing the path is defensible. The play-it-safe crowd treats small size as automatically virtuous. It isn't — sometimes it's just slow.

Third: the asymmetry is real. Risking $100 for a shot at $2,000 with a genuine edge is, in fact, a good bet. No argument.

If the pitch stopped there, I'd co-sign it. It doesn't stop there. It proceeds to a spreadsheet where $980 becomes $600,000 a year, and to get there it has to cheat three times.

Cheat #1: Your win rate is a sample, not a fact

Every one of these models starts the same way: "assume a 65% win rate." Then the number gets treated as a physical constant — plugged into pass-probability formulas, streak tables, hundred-account projections, all computed to the decimal.

But nobody has a 65% win rate. They have a win rate estimate from a finite sample. If your 65% comes from 60 trades, the honest range around it is wide — the true rate could plausibly sit anywhere from the mid-50s to the mid-70s. That's not pedantry; it's the whole game, because the strategy's aggressive sizing makes the outcome brutally sensitive to exactly this number. At the sizing these articles recommend, a true rate of 57% doesn't mean "slightly worse results." It means the pass probability craters and the hundred-challenge projection flips sign.

The spreadsheet shows you one future, computed from a number known to one decimal place that you actually know to within ten points. A desk quant would hand that model back and tell you to come back with error bars.

Ask yourself one question: how many trades is your win rate based on? If the answer is under a few hundred, in one market regime, you don't know your edge — you know a rumor about it. Sizing at half your drawdown per trade on a rumor is not aggression. It's rounding error cosplay as strategy.

Cheat #2: A hundred challenges, one bet

The Monte Carlo sections in these articles simulate 100 challenges as 100 independent coin flips. Sum the outcomes, watch the law of large numbers smooth everything into profit.

Except the flips aren't independent. It's one trader, running one strategy, in one market regime, across all hundred accounts — often simultaneously. When the regime shifts and the edge degrades, it degrades everywhere at once. When the trader tilts, every account gets the tilted version. The portfolio-of-challenges framing borrows diversification math while holding a single concentrated position: you.

Institutional risk desks obsess over exactly this — correlated exposure wearing a diversified costume is how banks blow up, not how they make money. Fifty accounts running the same edge is not fifty bets. It's one bet with fifty ways to feel it.

Cheat #3: The model assumes a trader who doesn't exist

Here's the quiet assumption underneath every table: flawless execution. The model trader risks half the drawdown per trade, loses, sizes correctly again, loses again, and calmly executes trade three by the book — because the spreadsheet says the streak probability is only 12.25%.

I sat next to professional traders — screened, trained, salaried, supervised — for a decade, and I can tell you what happens to human execution at maximum size after two losses: it degrades. Sizing that consumes 40–50% of your total risk budget per trade doesn't just raise variance. It puts every trade in the exact psychological conditions under which rule-breaking happens — which invalidates the win rate the whole model is built on. The 65% was measured on trades taken calmly at survivable size. It does not transfer to trades taken at existential size, and the model has no term for that transfer loss.

This is why banks don't let traders size anywhere near this way, even with proven edges and someone else's capital. Not because desks are timid — because they've watched the execution term fail at scale for decades, and they size to protect the edge from the person executing it. The structure isn't overhead on the strategy. It's what keeps the strategy's statistics true.

What a desk would actually do with this edge

Take the same trader the articles imagine — real edge, positive expectancy, prop challenges available at $100 a seat. Here's the institutional version of the play:

Size to survive being wrong about your own numbers. Not to the spreadsheet's win rate — to the bottom of its honest confidence band. Compress the trade count where structure genuinely supports it, but never past the size where your execution quality starts to move. Detach from accounts emotionally, exactly as the EV crowd says — and then let that detachment cut both ways: an account is also not worth breaking your process to resurrect. Track your actual per-setup expectancy from your own logged trades, in enough sample to mean something, and re-run the sizing math as the estimate tightens.

That version is slower than the fantasy. It also has the advantage of still existing in month four.

The concession

I'll give the EV bros the last thing honestly: their disrespect for account-preservation-as-religion is earned. Grinding $50 days for eight months to protect a funded account you're too scared to trade is also a failure to understand expected value — it just fails politely. The correct posture really is closer to a business owner than a curator.

But a business owner with one product, one supplier, and one customer doesn't call it a portfolio. They call it what it is: concentrated risk that had better be sized like it.

Run your own numbers before you run anyone's spreadsheet. Your win rate with its real sample size, your drawdown structure, your honest bands — that math takes two minutes and it's the difference between an edge and a story about one.

---

Fourdesk's free tools do this math without the fantasy: the expectancy calculator shows what your stats project — bands, losing streaks, and all — and the drawdown calculator sizes your next trade against the room you actually have left. The journal is free.