Automated tradingUpdated 3 min readBy the FXNOD team
Why do trading bots make losing trades?
Trading bots make losing trades because no rule can know the next price. Every strategy, good or bad, loses some of its trades. A bot is doing its job when each trade matches its rules, win or lose. The useful question is not why a trade lost but whether the wins, over hundreds of trades, are large enough to pay for the losses.
Losses are part of every strategy
A strategy with a real edge might win 55 trades in 100. It still loses 45, and those losses arrive in clumps, not evenly spaced. Judging a bot by its last five trades is like judging a coin by its last five flips.
Six reasons a bot keeps losing
- The market is random at that scale. On a synthetic index, each tick is generated independently of the one before. Patterns in the last few digits or ticks do not make the next one more or less likely.
- The payout is against you. If a contract pays less than it risks on an even chance, a rule with no predictive power loses slowly but surely.
- The rule was fitted to the past. Settings tuned until a past period looked perfect describe that period and nothing else.
- The market changed character. A trend-following rule loses in a sideways market, and a range rule loses in a trend.
- The stake sizing magnified an ordinary streak. The entries were no worse than usual, but the stake had grown.
- The sample is too small. Ten trades cannot tell a good strategy from a bad one.
The gambler's fallacy, automated
Many popular bots wait for a streak and then bet against it: five reds, so black is due. On an independent random process nothing is ever due. After five even digits the chance that the next is even is the same as it always was. Such a rule is not harmful by itself, since it is no worse than any other entry on a random market, but it gives no advantage, and combining it with a growing stake is where the damage comes from.
How to tell bad luck from a bad strategy
Check the trades against the rules
If the bot entered where the rules say it should, the software is fine.
Count at least a hundred trades
Fewer than that and you are reading noise.
Work out the break-even win rate
Divide the average loss by the sum of the average win and the average loss. If an average win is 0.90 and an average loss is 1.00, the bot must win more than 52.6% of the time.
Compare
If the measured win rate sits below break-even across several hundred trades, it is the strategy.
What to do about it
Do not raise the stake to win it back, and do not adjust settings after each loss. Stop the bot, return to demo, change one setting, and test again. If no version clears break-even over a few hundred trades, the honest conclusion is that the idea has no edge.
Questions people ask
- Is my bot broken if it loses several trades in a row?
- Usually not. Runs of losses are a normal feature of any strategy. On an even-chance contract, five losses in a row happens about once in every 32 starting points.
- Can a bot predict the next tick?
- No. On synthetic indices each tick is generated at random, so nothing predicts it. On real markets, short-term moves are close to random as well.
- Why did my bot win on demo and lose on real?
- Most often the demo test was too short to show a losing stretch, or the stake was scaled to a demo balance far larger than the real one.
Run a bot that keeps its limits
FXNOD's bots run on its servers and will not start without a stop loss, so your limits hold when your device is off. Build one in Bot Forge or start a ready-made one in Auto Hub, on your Deriv demo account first.
Get started