OptiNod Academy
Martingale — A Betting Progression That Stacks Your Point of Ruin Into a Single Slot of the Account
The smooth, rising equity curve of a Martingale does not raise your win rate. It is just the result of deferring losses and piling them into the one losing streak where the bet has grown largest. This piece lays out how a structure that grows the bet on every loss leads to ruin in the market's tail distribution, and frames the Anti-Martingale setup — the same progression flipped to the profit side — with a proven BTC example.
> The smooth, rising equity curve of a Martingale is a bill for deferred losses stacked into a single slot.
The Martingale is a betting progression that came out of 18th-century French gambling houses. In a game with a 50% win rate, like a coin toss, if you double your bet on every loss, a single win recovers all your accumulated losses plus one unit of profit at once. You grow the bet 1, 2, 4, 8, and a single win pays you 16 — enough to cover the 15 lost so far and leave 1. It is clean mathematically, but it only holds on the premise that both your balance and your bet have no limit.
The problem is that this premise breaks down completely in the market. Traders carry this progression over and apply it as averaging down. When the price falls below where you bought, you buy twice as much to pull the average entry down, so that even a small bounce takes the whole position past break-even. The longer the losing stretch runs, the more the bet grows exponentially, but in most sideways and gentle declines a bounce does eventually come and recovers break-even. That is why the equity curve rises smoothly for a while.
What this piece wants to point out is what that rising curve actually is. A Martingale does not raise your win rate. It only slices losses up and defers them, then takes on all those deferred losses at once in the single losing streak where the bet has grown largest. While small wins pile up, the losses stay deferred and accumulate. The one losing streak where the bet is largest bills that accumulated total — principal included — all at once.
The Rising Curve Is the Result of Deferring Losses Into One Place
A Martingale does in fact win often. As long as you keep doubling the bet, even a single small bounce recovers the accumulated losses. The win rate itself can run past 90%. But each of those 90% of wins carries a very small profit per win, and in the single loss where the bet has grown largest, you lose all the profit earned so far plus the principal at once.
The mechanism is simple. If you double the bet each time, the longer a losing streak runs, the more the next bet needed to recover swells, as follows.
| Consecutive losses | Cumulative bet | Next bet needed to recover |
|---|---|---|
| 3 losses | 7 units | 8 units |
| 4 losses | 15 units | 16 units |
| 5 losses | 31 units | 32 units |
| 6 losses | 63 units | 64 units |
| 7 losses | 127 units | 128 units |
After 5 losses the cumulative bet is 31 units, and to recover you have to stake 32 units on the sixth; if the balance does not stretch that far, the account empties out right there. No matter how high the win rate, once the bet crosses the limit of the balance first, the trades end there. If you treat 1 unit as 1% of the account here, the sixth bet after 5 losses becomes 32% of the account. The unit on the gambling table is exactly the position weight in real trading. This structure is exactly the same as the core idea explained in Risk of Ruin. In a game with an expectancy of zero or below, growing the bet only defers the point of ruin — it arrives with probability 1.
Look at the FTX collapse window of November 8, 2022, and it is clear where this single slot gets billed. BTC fell that day from an open of $20,591 to a low of $17,167, and the next day, November 9, it broke further down to a low of $15,588. An account that started averaging down in the $19,000s and had been lowering its average entry would have experienced break-even recovery several times right up until then. But in a window that fell nearly 25% in two days, the bet was already at its maximum, and in that one drop it lost all the small wins built up so far at once. What to look at on the chart is not how far the average entry has come down. More important is whether the inventory now laid out has left enough balance to survive the next one.
The Market Differs From the 50% Game the Martingale Assumes
The Martingale assumes a game where each attempt is independent and the win rate is fixed. A coin has no memory of how many heads came up before; the previous results do not affect the next toss. The market is not like that. Price moves in connected trends, and a decline invites a bigger decline in a Liquidations Cascade. That is exactly why the point at which averaging down has grown the bet largest happens to overlap with the point at which the decline accelerates.
On May 19, 2021, BTC crashed about 30% in a single day, from an open of $42,850 to a low of $30,000. On the yen-carry liquidation day of August 5, 2024, it fell from an open of $58,161 to a low of $49,000. In both cases the prior few days looked like a gentle correction, and the averaging-down account would have been holding on while lowering its average entry. Then a single daily bar billed all the bets built up so far at once. In a coin game a 30% run of losses almost never happens, but in the market's tail distribution it comes around regularly. The Martingale's math assumes a normal distribution, but the place where the big loss occurs is always the tail outside that normal distribution.
This is where judgment divides. Using a Martingale on a strongly trending asset means that when your entry is wrong, you grow the bet and put more into an even more unfavorable spot. If your reason for entering has broken down (market structure break), stopping further buys and closing out is the normal response.
Anti-Martingale Flips the Same Curve to Put the Big One on the Profit Side
The Anti-Martingale reverses the direction of the progression. You grow the bet when you win and cut it when you lose. In the stretch where the trend is going your way, you add to the position to aim for one big win, and in the stretch where you are wrong, the bet is small, so the loss is sliced into small pieces too. The equity curve takes the shape of many small loss-teeth, then one big up-stretch that lifts the whole thing.
This progression works because it uses the market's tail distribution in exactly the opposite way to the Martingale. The Martingale arranges the bet so you lose everything in the tail, while the Anti-Martingale arranges it so you earn the most in the tail. The pyramiding and scaled entries that trend-following strategies use are the real-world form of the Anti-Martingale. When a one-directional trend runs on — like the long climb toward BTC's March 2024 ATH of $73,777 — an account that grew the bet every time it profited makes most of its profit and loss in that one big trend.
In exchange, the Anti-Martingale has a low win rate. Small losses are frequent, and if the big trend does not come, the balance slowly grinds down. It is a progression that is psychologically hard to hold, which is also why people instinctively gravitate toward the Martingale. The core of the Anti-Martingale is that, in exchange for enduring a low win rate, it removes the risk of ruin in a single event.
Setup: Turning the Martingale Impulse Into an Anti-Martingale Structure
If completely banning averaging down is hard, replace it with rules that flip the betting progression to the profit side and block growth on the loss side. Take a trend-following entry on the BTC 4-hour chart as an example.
- Entry: Enter with 1% of the account as the first unit. On a loss, fix additional entries at 0, and make it a rule to forbid adding inventory that lowers the average entry.
- Profit-taking / management: When price reaches +2% versus the entry, add 0.5%; at +4%, add another 0.5% to pyramid. Always keep each added unit smaller than or equal to the previous unit. When price reaches +6% versus the entry, close 50% of the holding, and hold the rest until the 4-hour close breaks below
EMA50. - Stop: Set it at whichever is nearer: 1.5x
ATR(14)below the first entry price, or a break of the prior swing low. Cap the total loss at 1.5% of the account. - Position ceiling: Even after all additional entries are filled, total exposure does not exceed 2% of the account (1% + 0.5% + 0.5%). The 1.5% stop limit is the maximum allowed loss across this entire 2% exposure, and since additional entries are only attached after the first unit is already in the profit zone, they do not clash with the stop line.
- Invalidation: If the 4-hour close breaks below
EMA50, stop additional entries immediately and manage only the holding to its stop line. - Losing-streak cut-off: After 3 consecutive stops, set new entries to 0 for the day, and resume the next day with only 1% of the balance.
The core of these rules is to attach additional bets only when you are in profit, and to remove losses entirely from the basis for buying more. It makes the point where the bet grows largest overlap with the point where the trend is going your way.
Checklist: Inspecting Martingale Exposure on the Equity Curve
Martingale-style risk can be spotted before you enter by looking at the equity curve and position size together. Check the following.
- [ ] Was the recent add triggered by a price drop?
- [ ] Is the current average entry unit larger than the first entry unit?
- [ ] Assuming 5 consecutive losses, does the final bet stay within the balance limit?
- [ ] Is the equity curve rising smoothly upward with no loss-teeth at all?
- [ ] If the single largest position is liquidated, is the maximum drawdown at an unrecoverable level?
If a price drop invites more buying, the average unit is larger than the first unit, and the final bet on 5 consecutive losses exceeds the balance, then the risk of ruin is effectively 1. A curve with no loss-teeth at all is a sign that losses are being deferred, and if the structure lets a single event recover the whole thing, the progression has to change.
Place a smooth 90%-win-rate curve next to a 40%-win-rate toothed curve, and when one big decline finally hits, which side protects the balance is what divides them. As long as you look at a curve that only rises and believe the strategy is winning, you are mistaking unrealized losses for profit.
