Why Averaging Down and Martingale Blow Up Accounts

Martingale wins nine times out of ten and takes everything on the tenth. That’s not a flaw you can tune away — it’s the entire structure of the strategy.

The seductive logic of adding to a loser

Adding to a losing position feels smart because most of the time it works. Price dips, you double up at a better average, price ticks back, and you exit green on a trade that was underwater minutes ago. You feel like you rescued a loser through conviction.

The problem is what that feedback loop teaches. Every successful save reinforces the behavior and hides the tail. You’re not being paid for edge — you’re being paid for absorbing risk you can’t see until it arrives. The strategy has a high hit rate and a catastrophic, rare loss, which is exactly the profile that fools people the longest.

How martingale grows exposure geometrically

Classic martingale doubles size after every loss so that one win recovers everything plus a unit. The exposure doesn’t grow linearly — it compounds:

Lose once, double. Lose twice, quadruple. Lose three times, you’re at 8×.
After a modest losing streak your position is many multiples of where you started, and it’s your largest position at the exact moment the trade is going worst.
The required capital to “just win one more” explodes long before the streak feels unusual.

Losing streaks that look impossible are ordinary over enough trades. A run of adverse moves that a normal strategy shrugs off will, under doubling, walk your position size straight into your account limit. The streak doesn’t have to be long — it has to be one longer than your capital can fund.

The single trade that erases months of gains

Here’s the asymmetry that ends accounts. A martingale equity curve grinds up in small, satisfying increments — then drops vertically. Months of green ticks are structurally funded by the one red cliff that hasn’t happened yet.

Think about it in R-multiples: a well-run trade risks a defined 1R to make some multiple of R. A martingale sequence has no fixed R — the “R” of the final, largest leg dwarfs every gain that came before it. You’re not stacking small edges; you’re selling insurance against a move that will eventually come, and paying out the entire policy at once. On a funded account, that single leg doesn’t just wipe the gains, it breaches the drawdown and ends the account before you get a chance to “win it back.”

Averaging down vs scaling into strength

It’s worth separating two things people lump together:

Averaging down adds size as the trade moves against you. Your average entry chases the price down, your exposure rises while your thesis is being disproven, and your worst case grows.
Scaling into strength adds size as the trade moves for you, on confirmation, with the stop trailed so total open risk stays bounded.

They feel similar — both involve adding — but their risk profiles are opposites. One grows your position into weakness; the other grows it into evidence. If you want to add contracts, add them to the trades the market is already validating, not the ones it’s rejecting.

Why a positive-edge trader still shouldn’t martingale

Here’s the counterintuitive part: even a genuinely profitable strategy is worse under martingale, not better. A positive edge means favorable outcomes over many independent bets at controlled size. Martingale destroys the “controlled size” assumption — it correlates your largest bet with your worst run and turns independent trades into one giant leveraged wager on a streak not happening.

Edge compounds only if you survive to keep pressing it. Any sizing scheme with a nonzero chance of total ruin has, over enough trades, a probability of ruin that trends toward certainty. Bankroll survival is a precondition for edge, not a consequence of it. The math doesn’t care that your setup is good if your position sizing eventually bets the whole account.

Run a losing streak through a drawdown recovery calculator and the point lands hard: recovering from a deep hole requires an outsized gain, and martingale is a machine for manufacturing deep holes.

Anti-martingale: the safer way to add size

Flip the rule. Anti-martingale increases size when winning and decreases when losing — press when the account is proving the edge, shrink when it isn’t:

Risk a fixed, small fraction per trade so your worst case is always defined and always smaller after a loss, not larger.
Add to positions only after they’ve moved in your favor, with the stop trailed so the whole structure can’t cost you more than your initial risk.
Let winning streaks — funded by the market, not by doubling — do the compounding.

Size every entry off a fixed risk unit with a position size calculator rather than reacting to the last outcome. This is why Shibiki enforces per-trade and account risk limits at the broker: it structurally blocks the “just one more double to get it back” impulse, because the limit is a hard ceiling the platform won’t let you cross. Live edge health then tells you whether your actual strategy is working — so you can press size on evidence, not on a losing streak you’re trying to outrun.