I have coded a profitable EA, now what? | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

playing around and coded a profitable indicator and EA 11 replies

CAN such a thing be coded?... 17 replies

I have a great system that I want an EA coded for 35 replies

HELP! Need a simple EA coded 1 reply

Color Coded Market Watch 1 reply

Attachments: I have coded a profitable EA, now what?
Exit Attachments

I have coded a profitable EA, now what?

  • #1
  • First Post: Aug 4, 2016 2:40pm Aug 4, 2016 2:40pm
Hello there.
After months of trading manually I diverted my attention on developing my own EA.
I lost money on my way to finding a profitable system and now I am refraining to use the EA I have made. I would like to share with you the results.
The system is very simple and take advantage of the intraday price action. It works as a charm on EURUSD.
I ran the test over 5 years period with a spread of 0.4 for the worst case scenario (ICMarkets han an average of 0.1). I used the tick-data from ducaskopy so the tick modelled is extremely accurate.

What is the next step, selling to a edge found bank perhaps? How do I approach them, anyone had that experience?
Attached Image(s) (click to enlarge)
Click to Enlarge Name: pip_freeze.jpg Size: 341 KB
Click to Enlarge Name: pip_freeze p.gif Size: 24 KB
  • #2
  • Aug 4, 2016 4:30pm Aug 4, 2016 4:30pm
Quoting archishard
Hello there. After months of trading manually I diverted my attention on developing my own EA. I lost money on my way to finding a profitable system and now I am refraining to use the EA I have made. I would like to share with you the results. What is the next step, selling to a edge found bank perhaps? How do I approach them, anyone had that experience? {image} {image}
make forward tests, at its best, let the EA run on a live account and you can see the real results
(.)(.) think first ...
  • #3
  • Aug 4, 2016 5:38pm Aug 4, 2016 5:38pm
Yes, you need forward tests. Then One year of live tests Minimum. If it really is that good. It seems you would make far more managing others money. Use it for your self. And approach someone like fundseeder.com. Start your own hedge fund. That's the quickest way to billions. After you make plenty. Don't forget to hit your broker up for better spread.
You can lead a man to knowledge, but you can't make him think.
  • #4
  • Aug 4, 2016 11:52pm Aug 4, 2016 11:52pm
this EA might not work in a live account
practising coitus interruptus
  • #5
  • Edited Aug 6, 2016 1:09am Aug 5, 2016 2:16am | Edited Aug 6, 2016 1:09am
C'mon, it just opens random trades (hence the number of trades) with 1 pip of spread. What the hell are you trying to do ?
  • #6
  • Edited 4:31am Aug 5, 2016 2:28am | Edited 4:31am
You must have 6 to 12 months test done on a live account of more than 100K

If you dont have it test on what ever you can deposit and sell it to retail traders. for $197 and you will become a millionaire in 2 -3 years But you will need to have some marketing skills. I know traders that have spent 1000,s on useless ea,s off the web.
20 Pips a day keeps centerlink away
  • #7
  • Aug 5, 2016 9:38pm Aug 5, 2016 9:38pm
hi,
I suggest you re-run your back test. The modeling quality is n/a. Result should not be considered. Good luck
  • #8
  • Edited 12:57am Aug 6, 2016 12:47am | Edited 12:57am
Quoting archishard
I ran the test over 5 years period with a spread of 0.4 for the worst case scenario (ICMarkets han an average of 0.1). I used the tick-data from ducaskopy so the tick modelled is extremely accurate. What is the next step, selling to a edge found bank perhaps? How do I approach them, anyone had that experience? {image} {image}
You have what could be considered a scalping EA (avg 200 trade a month?). Don't forgot to factor commission per trade and slippage. You may have controlled some of the fills/slippage in your EA code (don't know) but your MT4 report does not factor in commissions, which will most likely alter your equity/free equity and lot sizes (and thus results) if incremental lot sizing is used. The main things is the fills/slippage, your one trade on the ledger shows a stop of 4 pips. With trading live ECNs you can't send a SL with the trade and it must be modified after the fill (the demo does not reflect this fact, although it looks like you have it coded to modify after the fill). Although you can have the EA modify it quickly after the fill, in a fast market, you may not be able to get those close stops in. In a standard DD environment, you will not pay commissions, but you won't get those tight fills either. Increase the spread to simulate commissions paid, increase your SL and then see what kind of results you get. By the way, a bank or HF is not going to talk to you unless you have 3rd party verifiable results and experience and in most cases are licensed, and of course in many cases they will try to pry away your algo either overtly or subtlety. Of course, if your EA kicks ass in a live environment privately, it's like printing money so why would you sell a goose that lays gold eggs, is a question mark.
  • #9
  • Aug 6, 2016 1:18am Aug 6, 2016 1:18am
135 665 / 5 years = 27133 / 12 months = 2261 trades / month
  • #10
  • Aug 6, 2016 1:24am Aug 6, 2016 1:24am
I`m not going to judge the back-tested results... (almost 18mil% increase)

Back-tested data shouldn't be considered. Even though it is the first step the results are highly miss-leading. With that said here are the steps that should be considered after the development of EA and back-test.

1. buy a VPS
2. run a live demo account on your VPS for at least 6 months.
3. Compare the trades to your back-tested trades. Run a performance check every month.
4. After 6months run another test on a small live/real account.
5. Performance check every month for at least another 6months.
6. if you get to this step you can open a bottle of Macallan 25.

Many people think that with EAs it removes the emotional aspect of trading that many people struggle with. I disagree with this myth. Trading through EAs has a similar impact on your emotions than manual trading. Many people developing EAs will agree.

Good luck on your journey. This is a rewarding business if you have the patience and persistence.
  • #11
  • Aug 6, 2016 2:26am Aug 6, 2016 2:26am
Quoting HeyYou
135 665 / 5 years = 27133 / 12 months = 2261 trades / month
No kidding. I misread it by a digit. Definitely scalping. HFT. Definitely slippage and commissions/spreads will affect outcome. Definitely only legit ECNs with commissions charged will allow trading it.
  • #12
  • Aug 6, 2016 3:24am Aug 6, 2016 3:24am
You can shorten this up really easy - instead of wasting time testing this code live and wondering, why it is not working, simply simulate a "part" of the slippage and bad fills you WILL encounter by raising your backtesting spread up to 20 points (2 pips). I am pretty sure it is going to tell a different story then.

Taking real spreads into consideration while testing something shows a lack of experience on your side.

Especially ICmarkets slipps like hell, I am scalping with them live.
  • #13
  • Aug 6, 2016 3:34am Aug 6, 2016 3:34am
market execution model of mt4 will kill this scalper...

backtests assumes INSTANT execution....sadly there is no such thing in real life...

i have amazing bt's on my had drive

as i say...

holy grail on backtest ....holy shit in forward test...live

best of luck...dnt celebrate yet
practising coitus interruptus
  • #14
  • Aug 6, 2016 3:56am Aug 6, 2016 3:56am
IF backtested with 20 pips on EU/GU etc., SL greater than 200 pips and using open prices then it's 95% reliable.


Order execution is practically instant in live accounts and slippage can also be positive... by the way you shouldn't get slipped under 1 lot I guess, that would indicate very poor liquidity.
  • #15
  • Aug 6, 2016 3:59am Aug 6, 2016 3:59am
Quoting HeyYou
IF backtested with 20 pips on EU/GU etc., SL greater than 200 pips and using open prices then it's 95% reliable. Order execution is practically instant in live accounts and slippage can also be positive... by the way you shouldn't get slipped under 1 lot I guess, that would indicate very poor liquidity.
market execution kills tight trailing...which this EA has...

broker will quote at market price despite EA requests
practising coitus interruptus
  • #16
  • Aug 6, 2016 4:12am Aug 6, 2016 4:12am
Quoting avibe
{quote} market execution kills tight trailing...which this EA has... broker will quote at market price despite EA requests

I don't even consider that backtest... totally garbage.
  • #17
  • Aug 6, 2016 4:53am Aug 6, 2016 4:53am
Quoting HeyYou
IF backtested with 20 pips on EU/GU etc., SL greater than 200 pips and using open prices then it's 95% reliable. Order execution is practically instant in live accounts and slippage can also be positive... by the way you shouldn't get slipped under 1 lot I guess, that would indicate very poor liquidity.
I forgot to mention, when backtesting using open prices the SL/TP must be much more than 200 pips on timeframes higher than the M1.
  • #18
  • Aug 6, 2016 6:44am Aug 6, 2016 6:44am
If you really have IT, just trade it and make huge money. Simple.



Quoting archishard
Hello there. After months of trading manually I diverted my attention on developing my own EA. I lost money on my way to finding a profitable system and now I am refraining to use the EA I have made. I would like to share with you the results. The system is very simple and take advantage of the intraday price action. It works as a charm on EURUSD. I ran the test over 5 years period with a spread of 0.4 for the worst case scenario (ICMarkets han an average of 0.1). I used the tick-data from ducaskopy so the tick modelled is extremely accurate. What...
  • #19
  • Aug 6, 2016 7:04am Aug 6, 2016 7:04am
Quoting archishard
Hello there. After months of trading manually I diverted my attention on developing my own EA. I lost money on my way to finding a profitable system and now I am refraining to use the EA I have made. I would like to share with you the results. The system is very simple and take advantage of the intraday price action. It works as a charm on EURUSD. I ran the test over 5 years period with a spread of 0.4 for the worst case scenario (ICMarkets han an average of 0.1). I used the tick-data from ducaskopy so the tick modelled is extremely accurate. What...

I can do precisely that too you want to buy my EA for $18 mill? lol
Because I'm kind i'll sell it to you for only $9mill arseshard
  • #20
  • Aug 7, 2016 1:11am Aug 7, 2016 1:11am
This is starting to skew off topic from the original post but,

Quoting HeyYou
{quote} I forgot to mention, when backtesting using open prices the SL/TP must be much more than 200 pips on timeframes higher than the M1.
Backtesting open prices is faster but...

SL/TP even if more than 200 pips (that's a price of +/- 2.00 Yen, EUR, USD whatever) can be unreliable a few times a year during INTENSE price moves and volatility, especially on an H4, H1, 30m or even 15m and 5m although very rarely will price move more than 2.00 before the decimal on a one minute, it does happen at times and can definitely happen within a 4 hour bar. So testing on open vs. control or tick only saves backtesting time. Besides, really who trades with a 200 PIP stop? Very few trade with a 100 pip stop. Is having a 200 PIP stop considered trading or investing? (rhetorical question).

Anyway that scalping monster is trading with a 4 pip stop. No way to get over 2000 trades a month trading with a 200 pip stop. Almost impossible unless stacking lots of small positions, even so would hit max open position limit set.

Anyway testing with "open" data, not realistic at all, but definitely simple, clean and a timesaver. Whereas real "trading" is often messy and requires lots of patience. If you have actually successfully made a transition to consistently profitable live trading based on only an "open price data" backtest, hats off to you.
0 traders viewing now