Most automated trading strategies don't fail because the logic is wrong - they fail because of a handful of operational mistakes that have nothing to do with the strategy itself. The seven below show up constantly, across every platform and every level of complexity: never testing on a demo account before going live, using a fixed-pip stop loss on every instrument regardless of volatility, risking a fixed lot size instead of a fixed percentage of the account, judging profitability purely from backtest numbers that don't model real spread and slippage, over-fitting a strategy until it only "works" on the exact chart it was built against, assuming every platform handles the same order the same way, and simply forgetting to flip the platform's global auto-trading switch before walking away. None of these are exotic - they are the mistakes that turn a genuinely sound rule set into a losing one, and every one of them is fixable in minutes once it's actually on your radar. These are mistakes made once trading is already automated - if you haven't automated anything yet, see why most traders never get that far for the barriers that stop people before this list even applies.
1. Skipping the Demo Account
A demo account is the only place to catch execution problems before they cost real money. A backtest runs on historical data - it can't show a rejected stop distance, a symbol name your broker spells differently, or a platform toggle left off. See backtesting vs. demo trading for what each one actually proves.
2. A Fixed-Pip Stop on Every Instrument
A stop distance that never changes protects a calm instrument and fails a volatile one - the same 20 pips can be too tight or too wide depending only on what's being traded, not on anything about the strategy itself. Sizing the stop from the instrument's own recent volatility (an ATR-based stop) fixes this automatically instead of guessing a number once and reusing it everywhere.
3. Risking a Fixed Lot Size Instead of a Fixed Percentage
A fixed lot size ignores the stop distance entirely, which means the same 0.10 lots can risk a very different dollar amount from one trade to the next. Position sizing and stop placement solve two different problems - the stop decides how far away the exit sits, sizing decides how large the trade is given that distance - and skipping one undoes the other. A related, more aggressive version of the same mistake is letting position size grow *after* a loss instead of staying fixed - see why martingale and grid trading strategies always blow up for what that looks like taken to its logical extreme.
4. Judging Everything by Backtest Numbers Alone
Most historical data used for backtesting doesn't model real spread widening, slippage, or requotes, so a backtest can confirm the logic is sound without proving it's actually tradeable. That last part only gets confirmed live.
5. Over-Fitting a Strategy to Its Own Backtest
A strategy tuned until it perfectly explains every swing on one historical chart is usually not describing the market - it's describing that one chart, and the gap shows up the moment new data arrives. Testing a few candidate parameters against each other, rather than endlessly optimizing one until the equity curve looks flawless, is what keeps a strategy's logic honest instead of memorized - how to spot a curve-fitted strategy covers the warning signs and the out-of-sample check.
6. Assuming Every Platform Trades an Order the Same Way
MT5's netting account mode can merge two separate orders on the same symbol into one blended position, changing how stop loss and take profit apply even though the strategy's own logic didn't change. MT4 has no netting mode at all. Neither behavior is a bug - not knowing which one your account uses is the actual mistake. See MT4 vs. MT5 vs. cTrader for the fuller picture.
| Netting (MT5 default) | Hedging / MT4 | |
|---|---|---|
| Two orders, same symbol | Merged into one position at a blended price | Stay fully independent |
| SL/TP after a 2nd order | Recalculated from the combined position | Each order keeps its own |
| Available on | MT5 (default mode) | MT4 (always), MT5 (hedging mode) |
7. Forgetting to Enable Auto Trading
An Expert Advisor sitting on a chart does nothing at all until the platform's global auto-trading switch is turned on - one toggle that silently overrides every rule attached to every chart, on every symbol. It's the last step in setup precisely because it's the easiest one to forget once everything else is already configured.
Common questions
What's the most common mistake in automated trading?
Skipping the demo account. A backtest runs on historical data and can't surface real execution problems - a rejected stop loss, a symbol name mismatch, a platform toggle left off - that only show up once a strategy is actually running against a live feed.
Do I need a demo account if my backtest already looks profitable?
Yes. A profitable backtest confirms the logic is sound on historical data - it says nothing about execution: whether the broker accepts the stop distance you configured, whether the symbol name resolves correctly, or whether spread behaves the way the backtest assumed. Those only surface live.
Why does my EA behave differently live than in backtesting?
Most historical data used for backtesting doesn't model real spread widening, slippage, or requotes - a backtest can show whether the logic is sound, but the real edge only gets confirmed once it's trading against a live feed with real execution conditions.
Does position sizing still matter if I already have a stop loss set?
Yes - they solve different problems. A stop loss decides how far away the exit sits; position sizing decides how large the trade is given that distance. A fixed lot size on every trade means the same lot can risk very different dollar amounts depending on how wide the stop happens to be.
What's the difference between netting and hedging for an automated strategy?
On a netting account (MT5's default), two separate orders on the same symbol merge into one blended position, which changes how stop loss and take profit apply even though the strategy's logic didn't change. On a hedging account, or on MT4 (which has no netting mode at all), each order stays fully independent.