Some of the most popular trading techniques, for bots and for people clicking by hand, rest on a single idea: when a trade goes against you, add more. The logic sounds reasonable. If price is cheaper now, buying more lowers your average entry, so a small bounce should bring you back to a profit. This lesson explains how the techniques work, why they feel like they succeed, and why they are so hard to sustain.
Three techniques, one idea
- Martingale. After each losing trade, the next trade is made bigger, often double. The aim is that one win recovers all earlier losses plus a small profit. The name comes from a betting system used in casino games.
- Grid trading. The trader or bot places buy orders (or sell orders) at fixed price steps, such as every 20 pips, and closes each one for a small profit on a bounce. As price keeps moving against the grid, more trades open.
- Averaging down (or "averaging in" to a loser). Adding to a losing position at a worse price, with the hope that a small recovery brings the whole position back to break even. Done without a plan, this is the manual version of the same behavior.
The common thread is that the position grows when the trade is wrong, which is the opposite of how risk is normally kept under control (see position sizing).
Why they feel like they work
- High win rate. Most cycles end in a small profit, so the record is full of winners. Many such systems win 80% to 90% of the time or more, until they do not.
- A smooth-looking equity curve. The account rises steadily with only small dips, which makes screenshots and backtests look superb.
- Losses stay hidden. Floating losses are not realized until the position is closed, so an open, growing loss does not show up in the "closed trades" record.
- It feels logical. Buying "cheaper" feels smart, and each win reinforces the habit.
The payoff profile is the problem. These methods collect many small gains and accept a rare, very large loss, often described as picking up pennies in front of a steamroller. A high win rate says little about whether the method is sound (see win rate and profit factor), because the size of the occasional loss matters as much as how often it happens.
The arithmetic of doubling
Take a simple martingale: start with a bet that risks $100 on a $10,000 account, double after each loss, and imagine a streak of losses in a row.
Doubling after every loss: how fast the damage grows
Illustrative: a $100 starting bet that doubles after each loss, with every trade assumed to lose in a row. Each bar is the total lost so far. By the 7th loss the total, $12,700, is more than the whole account, and the 7th bet could not even have been placed in full.
| Losses in a row | Next bet (× the first) | Total lost so far (× the first) | Chance at 50/50 odds |
|---|---|---|---|
| 1 | ×2 | 1 | 50% |
| 3 | ×8 | 7 | 12.5% |
| 5 | ×32 | 31 | about 3% |
| 7 | ×128 | 127 | about 0.8% |
| 10 | ×1,024 | 1,023 | about 0.1% |
A streak of 7 losses is unlikely on any single attempt, about 1 in 128 at even odds. But over 100 attempts, the chance of seeing at least one such streak is roughly 54%. Rare events become likely when you repeat the process long enough, and real trading has costs and imperfect win rates that make it worse.
Two more points follow from the numbers. First, changing your bet size does not change the odds of any single trade. A martingale does not create an edge. It only reshapes the results into many small wins and a rare large loss. Second, to survive a 10-loss streak you would need capital more than a thousand times your first bet, yet each successful cycle earns about one first bet. That is a very inefficient use of money.
Grid trading and averaging down: the same trap
A grid does not double the size, but each added trade still adds exposure in the direction that is already losing:
A grid in a trending market: small wins, then a growing pile
Illustrative: a grid bot buys at each step down and takes a small profit on every bounce. While price ranges, that keeps paying. When price trends down, buys pile up and the loss on the open trades grows with each one.
While price moves sideways, a grid keeps collecting small profits. When price instead trends, the open trades pile up on the wrong side, and the loss grows with each new step. Open positions held for a long time also pay or receive swap fees, which can add more cost to a large losing basket.
Averaging down by hand follows the same logic, with a human reason added: not wanting to admit the original trade was wrong. It is different from a plan to scale into a position in stages, where the total size and the total risk are decided in advance, with a stop-loss that applies to the whole position. The difference is not the act of adding. It is whether the maximum loss was fixed before the first trade.
Why these approaches are hard to sustain
- Capital is finite. Each added trade uses more margin. The account can reach a margin call or stop-out well before the market turns back.
- Brokers set limits. Most brokers cap position sizes and require margin, so the "next doubling" may simply not be allowed.
- Trends happen. Markets do move one way for long stretches, sometimes after central bank decisions or major news. Any method that depends on price always snapping back is exposed to those periods.
- Costs add up. Spreads, commissions and swaps are paid on every added trade, and they are paid most when the basket is largest.
- You see the survivors. Accounts that blew up rarely post screenshots, so the examples you find online tend to be the ones that have not failed yet.
- Psychology. The method asks you to add the most money at the moment you are doing worst. Even disciplined people tend to find that stressful, and a bot has no way to second-guess itself.
"Not sustainable" does not mean the account is guaranteed to fail next week. It means the risk builds quietly, so that a long, profitable-looking record can end in a single large loss. Compare this with approaches where each trade has a fixed, small risk and survival does not depend on a streak never arriving. See risk of ruin.
If you still want to study it
Some experienced traders use limited forms of averaging with strict safeguards. If you want to explore it, treat it as research, not income:
- Cap the number of adds and the total position size in advance.
- Use a hard stop-loss for the whole basket, set so that the worst case is a loss you could accept, such as a small percentage of the account.
- Size for the worst case, not the average case, using position sizing.
- Test through the worst trends in your data, with realistic costs, and see what the largest floating loss would have been.
- Practice on a demo account first and use very small size when going live.
- Be wary of any seller who presents a recovery method as low risk. See EA risks and scams.