[Experience] Spread Across 7 Firms, Blown Up Together: Same-Time, Same-Direction Accounts Are One Big Position, Re-Entering to Win It Back Becomes Averaging Down and Martingale, and Dragging a Real Account In Wiped Out the Year's Profit [November 2026]
Note: The account figures in this article come from the operator's own records in the prop-memo P&L tracker. Account numbers, trading times and strategy details are not included. The simulation uses synthetic data (coin flips) and does not guarantee future results. This is not investment advice.
TL;DR
- In one week of November 2026, almost all of the operator's prop accounts fell together. Two failed on max loss and one got a warning for its floating-loss limit. Only two accounts stayed in profit
- Two causes
- The accounts were spread across 7 firms, but every one held the same symbol, in the same direction, at the same time. Each account risked only 2–3%, within the rules, but added up it was about 15% of an account per stop-loss. The diversification was only on the surface
- After the losses, the operator wanted to win it back and re-entered in the same direction by hand. Stopping out and re-entering the same way, on several accounts at once, is effectively averaging down plus martingale
- In a synthetic Monte Carlo (7 accounts, zero edge), the average number of passes was 1.9 whether trades were staggered or aligned. What differed was the chance of losing every account: 10% staggered vs. 73% aligned
- The same re-entries were made on a real, own-capital account, and most of the year's profit is gone. "Losses stop at the fee" does not apply to a real account
- A lot size you have raised is hard to bring back down. Pass with the raised size and you carry it into the funded account, which eats into the value of the account you just won
- The money was being made from the gap between the fee and the reward for passing, yet fixating on the market itself handed that profit back
- Prop expected value depends only on the fee and the value of the account when you pass. That is why raising diversification works better than raising lot size
What happened
The setup the operator called "v5" put 2–3 "waves" (trades opened and closed at fixed times) on indices and gold into each of 14 accounts. The firms were all different.
But there were only a few kinds of waves. Waves that opened the same symbol in the same direction on the same day and hour sat in accounts at several firms.
| Same-time, same-direction position | Accounts holding it | Combined risk per stop-loss (% of a $100K account) |
|---|---|---|
| Gold long | 7 accounts | about 15% |
| Index long (two waves an hour apart on the same day) | 9 accounts | about 17% (both waves) |
| Index long (another weekday) | 7 accounts | about 17% (both waves) |
Account by account, each was 2–3%, within the rules. But because they all moved together, it was really the same as risking 15% per trade on a single $100K account. No firm's rules see this. Only you can.
Then the market went the other way, and every account dropped together.
Re-entering to win it back becomes averaging down and martingale
The real trigger came after that.
Looking at the damaged accounts, the operator wanted to win it back and added gold longs by hand. When they stopped out, they went back in the same direction. Doing this across several accounts meant:
- Within one account, "stop out → re-enter the same way" = averaging down
- Across all accounts at once, the combined lot size grew with every loss = martingale
The result: Fintokei Phase 2 (overall loss limit 11%) and Moneta 2-Step (10%) failed on max loss, and FTM issued a soft breach because "open trades exceeded the maximum allowed P&L level" (a 1% floating-loss limit).
It doesn't matter if each firm allows averaging down or martingale. Every re-entry in the same direction re-placed the same bet with the lot size of all accounts combined.
The worst part: including a real account
The same-direction re-entries were also made on a real, own-capital account. As a result, most of the profit built up this year is gone.
On a prop account, the loss stops at the fee. A real account has no such floor. "Take as much risk as the rules allow" only works in prop because the loss is capped at the fee. The moment a real account joins the same-direction position, that premise disappears, and one losing trade brings both prop failures and real losses at once.
Looking back, the money was coming not from the market but from the gap between the challenge fee and the reward for passing. Buy an account cheaply, collect the reward when it passes. That gap was the source of the profit, yet partway through the operator fixated on the market itself, kept betting that "this drop can be won back," and gave the accumulated profit back to the market.
Don't bring prop-style sizing into a real account. That was the most expensive lesson here.
Raised lot sizes are hard to lower
One more thing learned the hard way: a lot size raised to win it back is very hard to bring down.
- Lowering it means "winning it back takes longer," so there's never a reason to lower it
- After a win, it feels like "this size was right," so it stays
- After a loss, you want it back, so it goes up again
In other words, lot size only moves up, like a ratchet that won't turn backward.
In prop, if you pass with the raised size, you enter the funded stage at that size. The funded account is exactly the value you win by passing. Carry an oversized lot into it and you cut the value of passing. Some firms treat a big gap between evaluation and funded sizing as a problem in itself, so "I'll just lower it once funded" isn't a clean way out either.
Simulation: same expected value either way, very different chance of total wipeout
Checked with a coin-flip Monte Carlo (20,000 runs).
Assumptions: 7 accounts. Each passes at +8% and fails at −10% max loss (static) or −5% in a day. Up to 3 trades a day, 120 trades max, 2.5% risk per trade, RR 1:1, cost 0.05R. "Staggered" means the 7 accounts win or lose independently; "aligned" means all 7 win or lose together (same time, same symbol, same direction).
Zero edge (50% win rate)
| Setup | Avg. passes | All fail | At least one passes | Passes (10th pct / median / 90th pct) |
|---|---|---|---|---|
| Staggered (independent) | 1.91 | 10.4% | 89.6% | 0 / 2 / 3 |
| Aligned | 1.92 | 72.6% | 27.4% | 0 / 0 / 7 |
| Aligned + double after a loss, reset after a win | 2.29 | 67.3% | 32.7% | 0 / 0 / 7 |
| Aligned + 1.5x after a loss, never reset | 2.61 | 62.7% | 37.3% | 0 / 0 / 7 |
55% win rate
| Setup | Avg. passes | All fail | At least one passes |
|---|---|---|---|
| Staggered (independent) | 2.77 | 3.0% | 97.0% |
| Aligned | 2.79 | 60.1% | 39.9% |
| Aligned + double after a loss, reset after a win | 2.94 | 58.0% | 42.0% |
| Aligned + 1.5x after a loss, never reset | 3.27 | 53.3% | 46.7% |
Three takeaways:
- The average number of passes is almost the same whether staggered or aligned (1.91 vs. 1.92). How you place trades doesn't change expected value
- The chance of losing everything is completely different: 10% staggered, 73% aligned. Seven aligned accounts are not "seven attempts" but "one attempt at seven times the size." The median is 0 passes and the 90th percentile is 7 — all pass or all fail
- Raising lot size does add a few passes in the model. Since prop losses stop at the fee, taking risk within one account makes sense. But the chance of losing everything stays above 60%. Raising size while staying aligned doesn't change the all-or-nothing outcome. And the model doesn't count the cost of carrying the raised size into the funded account
Why "raise diversification"
Prop expected value comes down to:
Expected value = probability of passing × value of the account when you pass − fee
Losses stop at the fee, and profit only comes after passing.
- Raising lot size may lift the pass rate a little, but it tends to eat into the value of the account when you pass (the size you carry into funded). And with aligned accounts, the chance of losing everything barely falls
- Raising diversification (staggering time, symbol and direction across accounts) keeps expected value the same while sharply cutting the chance of losing everything. If you don't get wiped out, you don't get the urge to win it back or the cost of rebuying everything at once
This failure was exactly the opposite. Diversification was the lifeline, but partway through the operator merged the waves into what was effectively one, and after losing, chased it with bigger size.
In Is Diversification Really a Free Lunch?, tested with 40 real-data strategies, putting 10 accounts on one strategy wiped out 58.4% of the time versus 1.8% when split across 10 strategies. This article is the record of actually doing the "concentrated" version.
What changes from now on
- Stagger time, symbol, weekday and direction across accounts. Don't put the same wave at several firms
- Add up the same-time, same-direction risk across all accounts. Put it in a table, and if any time slot is too large, change the placement
- After a stop-loss, don't re-enter in the same direction. The urge to win it back is the entrance to averaging down and martingale
- Within one account, take risk up to the rule limits. What you raise is the risk of one account, not a lot size that moves every account at once
- Don't put a real account on the same direction and time as prop accounts. Don't bring prop-style sizing into an account whose losses have no floor
FAQ
Q. Isn't spreading across firms already diversification? A. No. Even at different firms, holding the same symbol in the same direction at the same time means every account wins or loses together. It's only diversification when wins and losses come at different times.
Q. If the firm allows averaging down or martingale, isn't it fine? A. Within one account's rules, yes. But re-entering on several accounts at once inflates the combined lot size somewhere no firm's rules are watching. That's how every account fell together here.
Q. Doesn't raising lot size increase the pass rate? A. Within one account, yes. Since losses stop at the fee, taking risk up to the rule limits makes sense. But raising it while all accounts stay aligned barely lowers the chance of losing everything, and the raised size tends to be carried straight into funded. Raise it within one account only, and stagger the accounts.
Assumptions and limits
- Wins and losses are random (coin flips). Real price moves, gaps and slippage are not included
- "Aligned" is the extreme case where outcomes match exactly (correlation 1). A real mix of index and gold longs isn't correlation 1, but same time and same direction gets it close
- Funded-stage performance after passing is not calculated (only the number of passes is counted)
- Script:
scripts/sim-correlated-accounts.py
Related articles
Written by
Hosono P | the prop firm strategist
I buy challenges with my own money and record everything through to the payout. Recorded payouts: ¥6.1M in total from Fintokei, Fundora and Funded7, plus $14,297 from The5ers and RZR Funded (as of November 2026). Author of the semi-discretionary EA "ELDRA".