Why is my EA worse live than in the backtest?
A live robot almost never matches its backtest: costs are wider live, fills slip, the price feed differs, and part of any backtest was luck. No fixed gap is normal; what matters is whether live results fall outside the range the test's own trades allow.
Check these first#
Ten places a gap between a test and live results comes from. You can check each one without Oathra.
- The spread. Was the test run on a fixed spread, and is the live one wider at the hours the robot trades?
- Commission. Was it in the test at all?
- Swap. Does the robot hold positions overnight, and did the test charge for it?
- Slippage. Are live fills worse than the price the robot asked for?
- Delay. How long does an order take to reach your broker from where the terminal runs?
- The price feed. Was the test run on your broker's own history, or on another source?
- The quality of the history. Was it tested on real ticks, or on prices built from bars?
- The settings. How many did you try before these? The more you tried, the more of the result was luck.
- The market. Has the pair's volatility or its trading range changed since the test ended?
- The account. Do the lot size, the leverage and the symbol's contract match what the test assumed?
How Oathra shows which case applies#
Oathra sets a linked robot's live numbers beside the backtest's, each with its range, and shows what changed and what to check first. Check your costs first.
Oathra raises one alarm, Below breakeven: the robot has been losing money for longer than chance explains. A robot that earns less than its backtest and still makes money gets no alarm.
The limit that applies#
Live tracking against the backtest works on MetaTrader 5 today. More platforms are coming.
This is not advice to keep or stop a robot.
Live tracking