Random Trading EA with success | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Is random really random? 231 replies

Random Thoughts on Trading 15 replies

"Random Entry Trader Exit" trading simulation 0 replies

Question about the LHF trading Simulator. Real ticks or random ticks? 1 reply

Random Trading 18 replies

Attachments: Random Trading EA with success
Exit Attachments

Random Trading EA with success

  • Poll

  • This EA is ..
  •  
  •  
  •  
  •  
  •  
  •  
  • #1
  • First Post: Edited 2:19pm Sep 8, 2014 12:06pm | Edited 2:19pm
Out of boredom I did this 72 liner, a very basic random trader.

Works like that:

- A buy trade hits either TP or SL
- Next is a sell ..
- Than a buy again etc

Currently I let this EA run a backtest and interestingly it works lol

BackTesting runs this way:

- Tickstory 4 Years EURUSD including correct comissions and swaps, close to 4GB file size limit
- Lot 0.1
- Optimize TP and SL from 1 to 500 each

Seems the backtest will run more 30 hours, best result so far: Net profit 5800 and DD 26%

Funny also if I optimize the EA similar to neuronal networks by optimizing TP/SL with the last 30 (or 60/90) days to trade forward the next 30 days it gives very good results. So if you optimize basically every day to prepare for trading the next 24 hours it could even be useful somehow.

Attached File(s)
File Type: mq4 RandomFun.mq4   2 KB | 1,730 downloads
  • #2
  • Sep 8, 2014 12:57pm Sep 8, 2014 12:57pm
Of course, stuff like this always interests me to no end. Thanks for publishing.

Did you account for Variable Spreads, or did you use a Fixed Value? Also, did you use interpolation for ticks or Real Tick Data? That will matter quite a bit. I don't see those variables in your back testing set-up. I use TrueFx for historical data (had problems with many others including Tickstory). Also, did you run any MAE/MFE analysis for each trade over the four (4) years? That should provide you with a path to some optimization as well.

Interesting results.
F-X-C-M | A Classic Retail Bucket Shop Scam Artist Who Manipulates Prices
  • #3
  • Sep 8, 2014 1:53pm Sep 8, 2014 1:53pm
I think spreads are fixed in MT4 during backtesting. Its coded in a way that in theory trades are entered only if the spread is 1 point. Doesn't apply in backtesting anyway afaik. Data is from Dukascopy as it's the main datasource in Tickstory. Once testing is finished I will upload it to Myfxbook so we see more details. Plus throw it on a Demo ECN account ..

However I tried to optimize this thing with a trailing stop but than it fails. Most trades will go negative quite a bit. The trades need room to swing until they hit their TP and lots of time ..


More 37 hours to go ..

Attached Image (click to enlarge)
Click to Enlarge Name: RandomFun.png Size: 27 KB


Strange that the number of trades is so low. Have expected ten thousands of trades like a zombie scalper or something
Forgot to mention, start capital is 1000 $.
  • #4
  • Sep 8, 2014 4:36pm Sep 8, 2014 4:36pm
You were wondering why the number of trades is so low. Are you aware that when you have a TP or SL of 200, based on $1000 starting account and 0.1 lot size, that it is equivalent to a 2000 pip TP or SL?

Also, your entries are NOT random. They are fixed and alternating following a certain path dictated by the data set you are using as well as the SL/TP amount. Change the signals' path by altering the starting value on the same data (int GLOBAL_Last = 1; // 0 Buy - 1 Sell) and you will see quite different results for those top optimized parameters.

Random entries would use something like a random number generator to determine long/short but you will find (if you test it) that you need some sort of edge, which random entries don't provide.
  • #5
  • Sep 9, 2014 12:00am Sep 9, 2014 12:00am
Quoting FXEZ
You were wondering why the number of trades is so low. Are you aware that when you have a TP or SL of 200, based on $1000 starting account and 0.1 lot size, that it is equivalent to a 2000 pip TP or SL? Also, your entries are NOT random. They are fixed and alternating following a certain path dictated by the data set you are using as well as the SL/TP amount. Change the signals' path by altering the starting value on the same data (int GLOBAL_Last = 1; // 0 Buy - 1 Sell) and you will see quite different results for those top optimized parameters....
This is interesting, but FXEZ is correct, not random because depends on outcome of previous trade, the only random thing is the date range. Did you try different date ranges? I did something similar in the past, and was surprised at an initial positive outcome, only to try a different date range which resulting in an equivalent negative outcome. Also, when you're finished optimising, you'll probably find the optimum will be a massive SL and a small TP.
  • #6
  • Sep 9, 2014 12:25am Sep 9, 2014 12:25am
what is the logic of its "random trading"? Care to explain ?
  • #7
  • Sep 9, 2014 12:25am Sep 9, 2014 12:25am
any TF or specific pairs?
i load it on my chart EU 15m nothing happen so far.
thanks
  • #8
  • Sep 9, 2014 2:20pm Sep 9, 2014 2:20pm
Quoting fredgold3434
any TF or specific pairs? i load it on my chart EU 15m nothing happen so far. thanks
If the EA doesn't work try to set the Spread to 1 in the tester.

Quoting moneybill
what is the logic of its "random trading"? Care to explain ?
Well there is no real logic behind this EA, its just a test of random trading. But as mentioned by the previous posters it may not be exactly random in fact.

Its still optimizing, but the best result of 5800 has not been beaten. TP 165, SL 135
  • #9
  • Sep 9, 2014 4:19pm Sep 9, 2014 4:19pm
Quoting NeuroBeep
{quote} If the EA doesn't work try to set the Spread to 1 in the tester. {quote} Well there is no real logic behind this EA, its just a test of random trading. But as mentioned by the previous posters it may not be exactly random in fact. Its still optimizing, but the best result of 5800 has not been beaten. TP 165, SL 135
i dont do tester i just load it on demo 15m EU chart.
it does not work at all.
regards
fred
  • #10
  • Sep 9, 2014 4:23pm Sep 9, 2014 4:23pm
The EA works only if the spread is 1 point (0.00001) which was done to reduce spread costs. You can remove this spread check in the code.
  • #11
  • Sep 9, 2014 4:30pm Sep 9, 2014 4:30pm
how can i remove it or maybe adjusted in the code?
thanks
fred
  • #12
  • Sep 9, 2014 4:44pm Sep 9, 2014 4:44pm
Here the version without Spread checks:

Attached File(s)
File Type: mq4 RandomFun_NoSpread.mq4   2 KB | 817 downloads
  • #13
  • Sep 9, 2014 4:53pm Sep 9, 2014 4:53pm
I stopped the optimizer and uploaded the best result here:

http://www.myfxbook.com/strategies/randomfun/70312
  • #14
  • Sep 9, 2014 5:03pm Sep 9, 2014 5:03pm
Quoting NeuroBeep
I think spreads are fixed in MT4 during backtesting.
Use Birt's Patch to make them Variable. You will also be able to set-up other more realistic variables as well.

Quoting NeuroBeep
Its coded in a way that in theory trades are entered only if the spread is 1 point. Doesn't apply in backtesting anyway afaik. Data is from Dukascopy as it's the main datasource in Tickstory.
Dukascopy, use to be good for back testing in MT4, but they started having problems later on - not sure why or what happened. Maybe they've gotten it corrected by now.(?)


Quoting NeuroBeep
Once testing is finished I will upload it to Myfxbook so we see more details. Plus throw it on a Demo ECN account .. However I tried to optimize this thing with a trailing stop but than it fails. Most trades will go negative quite a bit.
That's where the MAE analysis helps. It won't solve the problem for all draw scenarios but it can help to filter out some and increase performance. It can also help to shift entries and exits which can compensate for underlying losses that would have normally taken place. I can't design a system anymore without good MAE/MFE analysis.

Quoting NeuroBeep
...Strange that the number of trades is so low. Have expected ten thousands of trades like a zombie scalper or something...
Signal Availability. It is in the category of Most Overlooked Factors Affecting Performance Optimization. Difficult to get reliable revenue projections with unstable SA. One of those lessons systems developers will learn one way or another over time.

Good hunting with it!
F-X-C-M | A Classic Retail Bucket Shop Scam Artist Who Manipulates Prices
  • #15
  • Sep 9, 2014 5:05pm Sep 9, 2014 5:05pm
Two years of profits followed by two years of breakeven doesn't look very fun.
  • #16
  • Sep 9, 2014 6:38pm Sep 9, 2014 6:38pm
thanks
  • #17
  • Sep 9, 2014 6:46pm Sep 9, 2014 6:46pm
Your EA close all other pairs open trades,why?
for example if i put the EA in EU chart,then it closes all other pairs open orders,why?
regards
fred
  • #18
  • Sep 9, 2014 7:01pm Sep 9, 2014 7:01pm
The EA is very basic and assumes that there are no others trades in the account. Have a look at the code. You can avoid this using magic numbers for example.
  • #19
  • Sep 11, 2014 7:21am Sep 11, 2014 7:21am
Quoting NeuroBeep
{quote} If the EA doesn't work try to set the Spread to 1 in the tester. {quote} Well there is no real logic behind this EA, its just a test of random trading. But as mentioned by the previous posters it may not be exactly random in fact. Its still optimizing, but the best result of 5800 has not been beaten. TP 165, SL 135
'Ok thanks, so if this is random entry, what is the concept of the exit strategy?
  • #20
  • Sep 11, 2014 10:46am Sep 11, 2014 10:46am
Exit is when TP or SL is hit. This thing is basically just a proof of concept or .. disproof You need to runs this through the Optimizer first to determine working TP/SL levels for the timeframe you like to test.

It seems that this EA only works with constant optimizing, otherwise it will make boom soon. Handle with care ..
0 traders viewing now