Martingale Trading Bots: The Math That Guarantees Ruin
Martingale and averaging-down bots win small for months, then one losing streak wipes the whole account. The math of ruin, and why a fixed stop beats it.
A martingale bot shows you the prettiest equity curve you will ever see. A near straight line, up and to the right, week after week, with a win rate north of 90%. Then one afternoon the line drops vertically and the account is gone. That is not a malfunction. That is the strategy doing exactly what it was designed to do.
Martingale bots, and their cousins that average down into losers, are the most seductive category of automated trading and the most reliably fatal. The math is against you, and it hides the bill until the very end. Here is what is actually happening under the hood.
What a martingale actually is
The martingale is a betting rule from the roulette table. You bet one unit. Lose, and you double to two, then four, then eight, then sixteen. The moment you win, that single win recovers every prior loss and leaves you one unit ahead. On paper it looks like a machine that mints a small profit out of any losing streak.
In trading the same idea wears two costumes:
- The pure martingale. After a losing trade, the bot doubles position size on the next trade, so one winner claws back the whole string of losses.
- Averaging down. The bot keeps a losing position open and adds to it as price moves against it, lowering the average entry, so a small bounce reaches breakeven instead of a full recovery.
They are the same trade in different clothes. Both refuse to accept a defined loss, both bet larger the more wrong they already are, and both win almost all the time, right up until the one time that ends everything.
A strategy that wins 95% of the time can still have negative expected value and, given enough trades, a probability of ruin that climbs toward certainty. How often you win and whether you make money are two separate questions. Martingale bots are built to look wonderful on the first and are terrible on the second.
Why it wins so often
The high win rate is not a trick, it is arithmetic. If you never book a loss and keep adding size until price comes back, then most of the time it does, and you close green. A market that oscillates rewards this for weeks, and every recovered position is logged as a win. What the curve does not show is the size of the position currently underwater, or how much capital is committed to defending it. The profit is small, frequent and visible. The risk is large, rare and hidden. That asymmetry is the whole story.
The math of ruin, worked out
Take the friendliest case: a fair coin, no house edge, no commissions. You bet $1 and double after every loss. Here is what the next bet costs after a run of losers.
| Consecutive losses | Size of the next bet | Total staked to get there |
|---|---|---|
| 0 | $1 | $1 |
| 3 | $8 | $15 |
| 6 | $64 | $127 |
| 9 | $512 | $1,023 |
| 12 | $4,096 | $8,191 |
| 15 | $32,768 | $65,535 |
Fifteen straight losses sounds impossible, but over thousands of trades it is expected to show up eventually. This is the classic gambler's ruin problem, and it has a known answer: with a finite bankroll against an opponent (the market) that has effectively unlimited capital, continued play ends in ruin with probability approaching one. That is why it is worth reading risk of ruin as a coin flip before you trust any high win rate.
The expected value makes it worse. Each sequence risks the whole bankroll to win a single unit, so once you weight that rare catastrophic loss by its probability, the edge is zero at best on a fair coin and negative on any real market, where commissions, slippage and gaps take their cut. You are trading a large hidden tail risk for a small steady stream of wins, and over enough trades the tail collects.
The drawdown is invisible until it is terminal
In a normal strategy, a loss is realized the moment the trade closes and shows up in the drawdown figure. In a martingale, the losing positions stay open, so the reported closed-trade drawdown stays tiny while the real mark-to-market risk balloons behind it. The number you would use to judge a system is lying to you by construction. That is why what drawdown is matters here: a bot that hides its drawdown has hidden the only figure that tells you whether you can survive it.
Why scams love it
A martingale bot can show a spotless live track record for six or twelve months, and sellers use that as proof. The screenshot is real. The disaster is real too, it just has not printed yet, so it is not in the window they show you. Grid bots are the same family, a lattice of orders that quietly accumulates a bigger and bigger position against the trend, which is why so many grid bots blow up. A 90-something percent win rate, a tiny drawdown and no stop loss is almost always one of these, and it belongs on your list of trading bot red flags. Our guide to algorithmic trading covers what a real edge looks like.
Frequently asked questions
Do martingale bots ever work?
They work until they do not, and that phrasing is the honest one. A martingale produces months of small, steady gains while the market chops sideways, which is why the live results look convincing. The losing streak that ends the account is not a rare glitch, it is a normal feature of markets that the strategy has no defense against. Over a long enough sample it arrives, and one blow-up erases every prior win plus the account.
Is a 95% win rate not a good thing?
Not on its own, and it can be a warning sign. Win rate only tells you how often trades close green, not how much you lose when they do not. A strategy with a defined stop can win less than half the time and still make money because its winners are larger than its losers. A martingale inverts that: it wins constantly and small, then loses everything at once. Always ask what the losers cost, not just how often the winners land.
Is averaging down the same as a martingale?
It is the same idea running more slowly. Averaging down keeps a losing position open and adds to it as price moves against you, lowering your average entry so a small bounce reaches breakeven. Like the pure martingale, it refuses to accept a defined loss and commits more capital the more wrong it already is. The hidden risk grows the same way, and a strong trend against the position ends it the same way.
How do I spot a martingale bot being sold to me?
Look for a very high advertised win rate, a smooth equity curve with almost no drawdown, and no mention of a stop loss or a defined per-trade risk. Ask whether the bot ever adds to a losing position as price moves against it. If the answer is yes, or the seller cannot give you a fixed worst-case loss per trade, you are looking at a martingale or a grid, and the pretty track record is only measuring the time before the blow-up.
- William Feller, "An Introduction to Probability Theory and Its Applications," Volume I, chapter on the gambler's ruin problem.
- Lester E. Dubins and Leonard J. Savage, "How to Gamble If You Must: Inequalities for Stochastic Processes."
- CFTC, "Customer Advisory: Be Cautious of Trading Systems and Robots."
Educational content, not investment advice. Trading futures involves substantial risk of loss and is not suitable for every investor. Any reference to Algomyth results is hypothetical, simulated over historical data, and shown in full on the performance page; past performance does not guarantee future results.
Twenty years, drawdowns included
Every engine publishes its full history repriced at today's values, with the losing years and the deepest drawdowns printed as plainly as the profits. Verify first, decide later.
More from Judging a system
Judging a systemDo Trading Bots Really Work? What the Evidence SaysDo trading bots work? Yes at executing an edge without flinching, no at inventing one. What the evidence supports, and what the marketing quietly leaves out.Judging a systemWhy cheap trading bots cost the most, in actual dollarsA cheap bot that burns months and a slice of your account costs far more than its price. What cheap trading bots really cost, in dollars.
Judging a systemWhat a Trading Bot Can and Cannot Legally Promise YouA US trading bot cannot legally promise guaranteed returns or say it is risk-free. Where the CFTC, NFA and FTC draw the line, and what honest claims look like.