How To Evaluate A Trading Bot

Evaluating bots is never easy, especially so for a beginner. Follow the steps below so you can determine if a bot is something that warrants purchasing.

+Here are qualifying questions on each bot that I need answered.
+Performance and drawdown are the two biggest deal breakers to begin with as long as they look good from a few different angles, and if those are okay then the other metrics will help influence the decision making process on a cumulative basis. 
Please go to www.myfxbook.com and find a bot to follow along.  Some developers will expose everything about the bot, others will show only a little.  FXBlue is another popular bot reporting tool just not as popular and a little harder to navigate
+If a developer cannot provide a link to the bot performance, then I dont even bother looking further.

Myfxbook General Data and Stats

Top Header Row

  1. Top left – Real or Demo account, name of broker, leverage, trading software platform
    1. Broker – the better reputation of the broker the better.
    2. Demo or Real account – Better to have a real account
    3. Leverage – this affects the drawdown reading.  More leverage the better for trading, but not all brokers offer more than the 1:200 needed for most bots  
  2. Track Record & Trading Privileges – these should be green and checked off.  (Red flag if they are not)
  3. Time in service – like to see 6 months or more of history, 1 year or more is better of course.
    4. Custom Analysis – click and hit ‘all’.  Some developers will hide bad results they don’t want you to see with a date range on the chart.  Sneaky.

Left Side Box

  1. Gain – Not to be trusted.  This is a myfxbook calculation issue and not the bot.
  2. Daily return – accurate and can be relied upon
  3. Monthly return – this is the most accurate on myfxbook IMO and tracks my monthly return when I calculate manually.  
  4. Drawdown – this is the overall deepest drawdown the bot has experience. Anything above 30% needs to be looked at carefully, how frequent that happens is important.  Anything below 10% is a solid 
  5. Balance – current balance of funds
  6. Equity – balance less open trades at current price
  7. Profit – overall profit the bot has made
  8. Interest – interest paid or received by holding trades overnight (Swap fees)

Deposits & Withdrawals – I don’t want to see too many deposits, and I want to know when additional deposits happened after the initial funding.  If a bot goes into drawdown, then the developer may add more funds to make sure the bot is in a good spot, this should not happen in the first place, but is okay if not very frequent.

Right Side Chart

  1. Growth Line (orange line on chart) – Nice smooth line is better.  Not all bots will have a smooth line, but as long as the losses are not big and general trend is up to the right
  2. Equity Growth Line (yellow line on chart) – When the equity growth line dips below the growth line, that means the bot has open trades at a loss. That is okay sometimes, just not so often.

Drawdown

(Displaying the Drawdown tab)

  1. Drawdown (very important ) – One of the most important factors in a bot’s performance.  This chart shows the end of day drawdown reading for the bot.  
  2. You really want to see a consistent low value.   
  3. Anything below 10% is very good with the right leverage, once a month or so above 10-20% is okay.  Anything above 30% and I want to know why or that it has not happened very often.
  4. Your account can survive above 30% but of course many factors come into play.  Riding too close to the sun above 30%, you do not want to see that all the time. 
  5. You don’t want to see a sustained period of drawdown at a high level.  It means there are open trades at a loss. 

Trading Statistics

  1. Overall profit performance statistics by period for the bot.  Not much to explain.  However, the yearly figure is not to be trusted, this is a myfxbook issue. Better to add the individual monthly returns.

 

Advanced Statistics

  1. Longs Won/Loss Won – want to see these in the 70-80% combined range.  Less is possible for a good bot, but I like above 70%.  You can have a profitable bot with 50% win rate though, really depends on the bot strategy
  2. Profit Factor – above 1.5 is my general standard. Above 3 and I start getting a little suspect.   
  3. Standard Deviation – this is a measure of risk.  Readings are all over the map but the lower the better. 
  4. Sharpe Ratio – the higher the better.  You can still have a good bot with a low sharpe ratio but it’s a good number to look at when evaluating bots.  
  5. Average Trade Length – lower the better, it can give you an idea of how the bot strategy.

Trading Activity

Note: Trading Activity/History – Sometimes these tables are left open by the developer and will show everything including deposits. Take a good look if you can.

 

  1. Open Trades – not available to view very often as folks will use this to copy-trade a bot.
  2. History – I want to see that the ‘Lots’ traded by the bot are consistent for the bot strategy,  trades that are out of the ordinary, most likely a lot higher size, need to be understood as they could be manual trades.  No additional deposits by the developer after the initial deposit is preferred as that means the bot was in trouble and needed additional funds.   Withdrawals are okay of course.

Monthly Analytics

  1. Monthly Analytics – look at the month over month returns – This is very accurate.  Like to see a consistent return within a range. Click on a month, that will show you what securities are being traded. 
  2. WARNING:   If the bot has a very large out of the ordinary return in the beginning months of the bot, which is not repeated in any other month subsequently, then you need to understand why.  Look at the trading activity for that month if you can and the lot sizes being traded are consistent, maybe manual trades were placed.  What some developers will also do is run many versions of the same bot and keep the winner, if recent monthly returns are lower than the first few months, that’s not good.  They could have also had an aggressive setting in the beginning to get good overall statistics, but now run it at lower risk to keep the bot safe.

Other Bots

Other bots linked to the same account.  Click on them and you can take a look at their performance.  Nice to see this developer testing a bot on different risk levels.

Backtests

Some developers love to throw backtest results at you and say ‘buy my bot’. Not a chance. There are few good reasons to not rely on a backtest unless you know the person who did the backtest and trust them, which is probably not the case. 

 

  1. A bot can be optimized for the historical market data that it uses to perform the backtest. You would never really know.
  2. I consider backtests as getting you in the ballpark, but are not a true indication to be relied upon for future performance. You need a forward test on demo or real equity to truely get an understanding of the bot.
  3. You will see a lot of the same metrics in a backtest report that I talk about in evaluating a bot, but data quality is the most important thing, it should be 99% or better.
  4. A 10 year backtest does nothing for me as markets change.  I generally want to see 1-2 years of backtest results, depending on the timeframe the bot runs on.