- This topic has 3 replies, 1 voice, and was last updated 6 days, 7 hours ago by
.
-
Posts
-
I’m having huge differences with real accounts after creating robots in FSB pro. Trying to set up the historical data is a difficult job. Can you provide some historical data files from 5%ers, FTMO & Funded Trading PLus so I can match my robots to their platform better?
The best way to achieve the same results in FSB Pro and MetaTrader is to have the same data.
The easiest way is to export data files from your FTMO account and import them into FSB Pro. Good or bad, these are the data you are going to trade.
To Add to this, we don’t provide data from these sources. As Miro says, if you use mt5 to download the broker data and then export that data from mt5 and import into FSBPro. To test whats going on try to make sure the broker data and fsbpro data is identical and then then things should match, huge differences due to different data sources could be a sign of overfitting.
I’ve tested almost 864 EAs over time, and one thing I’ve learned is that a high backtest profit doesn’t automatically mean an EA is good.
Out of all the EAs I’ve tested, there are only 2 I’m comfortable recommending.
What made them different?
• Predefined SL & TP
• No Martingale
• No dangerous Grid
• Proper risk management
• Controlled drawdown
• Reasonable risk-to-reward
• Consistent performance across different market conditions
• Realistic backtesting
• Forward/live testing
• Spread & slippage protection
• News filter
• Trading session filter
• Lot-size control
• Daily loss protection
• Maximum trade limits
• No unlimited averaging down
• Stable equity curve
• Doesn’t depend on one lucky periodI stopped judging EAs by “how much profit did the backtest make?”
The better question is:
“How does it behave when the market stops going its way?”
That’s where most of the 864 EAs I tested failed.
The two I kept weren’t necessarily the ones with the biggest returns. They were the ones that showed the best balance between profitability, consistency, and risk control.
