Are the best systems also the systems that fail most spectacularly? | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Top 100 best Trading Systems: post good Trading Systems here 62 replies

Which system is the best? Or which combinations of systems work the best? 46 replies

Mathematical reasons why MOST DayTraders & scalpers fail? 197 replies

Do your trading systems also work on stock indices? 11 replies

Systems, Systems and Systems... 1 reply

Are the best systems also the systems that fail most spectacularly?

  • #21
  • Aug 31, 2009 6:35pm Aug 31, 2009 6:35pm
Well I can see I've shut everybody up. For those of you purists who think I'm taking a "short cut"... This is the result of thousands of hours of thinking, failing, trying again, and never giving up!

Catch me if you can!!!
  • #22
  • Aug 31, 2009 6:39pm Aug 31, 2009 6:39pm
Quoting saaedb
Why are you wasting asking people questions that they can't answer. Demo test the bot on live data for a couple months and see how it works out. And then when you're comfertable with it, use proper money management and try it out on a real live accunt.

I'm demo trading this these days... waiting for it to bite the dust . Login details for it. (it's my own home brewed bot)

Login : 2406107
Password : wfi5xl
You don't know how right you are, mate. I've more work ahead of me now messing around with DLLS and trying to integrate this system with Metatrader. carefully monitored demo trading will follow.
  • #23
  • Aug 31, 2009 6:45pm Aug 31, 2009 6:45pm
Quoting FrinkFX
You don't know how right you are, mate. I've more work ahead of me now messing around with DLLS and trying to integrate this system with Metatrader. carefully monitored demo trading will follow.

Why did you build it in C++? Why not in MQL? I remember my hell days with DLLs. With C# out there, I'm glad those are over .
  • #24
  • Aug 31, 2009 6:46pm Aug 31, 2009 6:46pm
Quoting FrinkFX
You don't know how right you are, mate. I've more work ahead of me now messing around with DLLS and trying to integrate this system with Metatrader. carefully monitored demo trading will follow.

It's got to beat messing around with MT4 horrible code, using C or something ??

There isn't a long term winning EA out there sadly, if there way we'd all be using it!!


Good luck I fear you'll need it!!
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #25
  • Edited 7:18pm Aug 31, 2009 7:02pm | Edited 7:18pm
Quoting saaedb
Why did you build it in C++? Why not in MQL? I remember my hell days with DLLs. With C# out there, I'm glad those are over .
I needed the speed of C. The system simulates thousands of trades per second and requires huge computing power. MQL or even C# can't match the speed of C.

also the system doens't look into the future (even though it predicts it fairly well), doesn't hedge or use martingale or huge stoplosses any other dumb MM techniques. It doesn't need ultra-fast execution and can handle slippage. It doesn't care about news events/session times or anything. It places 1 trade at a time and wins more often than it loses.

edit: I tried it on 2 more pairs: GBPUSD and USDCAD and it STILL made profit. I then turned the spread up to...wait for it...7

and it STILL made profit!

I know most people will ignore me now because I'm making what seem to be outrageous claims and that suits me perfectly because after all this hard work I need some time to brag, but the evidence is there on the screen in front of me!
  • #26
  • Sep 1, 2009 12:45am Sep 1, 2009 12:45am
hi
would you mind telling us how fast it took to back test that time period for one Ccy Pair and also are you using tick data or OHLC?

Thanks
  • #27
  • Sep 1, 2009 4:18am Sep 1, 2009 4:18am
Quoting FrinkFX
I heard something on bloomberg about
some guy who did this, it wasn't forex trading, I think it was government bonds or something
You call that evidence? good luck anyway
BTC Return This Year: na
  • #28
  • Sep 1, 2009 10:47am Sep 1, 2009 10:47am
Quoting Spearion
hi
would you mind telling us how fast it took to back test that time period for one Ccy Pair and also are you using tick data or OHLC?

Thanks
Just using 1min data. Probably would work better if I used tick data though.
tests didn't take too long. about 30 seconds for each month, using 50 simulations per bar. If I use say 1000 simulations per bar it goes slower but makes more profit. Realistically I could perform several million of these simulations in one minute, as it would be in forward-testing. However the profit levels off at about 1000 simulations
  • #29
  • Sep 1, 2009 12:45pm Sep 1, 2009 12:45pm
Quoting M.A.C.Doug
absolute b.s, I challenge you to provide evidence of a single case of this occurring

Go Google 'Automated Trading Championship', enough said.


Frink, the reason a lot of the 'near 100%' systems fail big is because they have outrageous risk reward ratios. They have 100 pip or more stops and scalp a few pips at a time on the winners, so one loss can wipe out many wins.

If you've success over years worth of data on a high frequency system and your stops are reasonable, the you might be OK, IMO.

Good luck.
  • #30
  • Sep 1, 2009 12:49pm Sep 1, 2009 12:49pm
Im confused....

If a system is 'one of the best systems' then it wouldnt 'fail spectacularly'.
Living the adventure in my head.
  • #31
  • Sep 2, 2009 6:21am Sep 2, 2009 6:21am
Right so it takes 30 secs do 50 sims per 1 month of 1 minute bars and you then say you could do 1 million of these in a minute? Is that for two months?

Another thing going by your stats your 50 sims would take 30 secs x 48 months to do 4 years of testing which = 24 minutes. This is not fast for C++ and OHLC bars .... even tick data should be way faster than this.

Or have I missed something?
  • #32
  • Sep 2, 2009 10:26am Sep 2, 2009 10:26am
Quoting Spearion
Right so it takes 30 secs do 50 sims per 1 month of 1 minute bars and you then say you could do 1 million of these in a minute? Is that for two months?

Another thing going by your stats your 50 sims would take 30 secs x 48 months to do 4 years of testing which = 24 minutes. This is not fast for C++ and OHLC bars .... even tick data should be way faster than this.

Or have I missed something?
I've since made it faster but the program is doing more than just simulating trades

I tested the simulatetrade method itself and it simulated over a millon per second
  • #33
  • Sep 2, 2009 10:34am Sep 2, 2009 10:34am
So your system simulates millions of trades say or loads per 1M data block being OHLC, in different ways then calculates what to do if anything.


Not seeing this, but interested as a programmer so keep going.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #34
  • Sep 2, 2009 10:43am Sep 2, 2009 10:43am
Quoting Leonlorenzo
Im confused....

If a system is 'one of the best systems' then it wouldnt 'fail spectacularly'.
Leon,

The past has shown that 'extreme money generating models' like the one based on the Black & Scholes option formula used by Long Term Capital Management and the more recent CDO based models also rely on extreme leverage and market exposure.

Now, these 'systems' have as a basis a very sophisticated construction that in the ideal environment generate out of proportion (high) returns in a symbiotic fashion, i.e many people profit at once from being involved in the ladder of transactions (in the CDO market, the mortgage industry as well as the institutions that package the CDO's, down to the sub-prime mortgage receipient).

When something unexpected happens in the chain or an assumed value changes in the formula, it can lead to a massive implosion of the whole system (in this case, the CDO formula relied on rising - or at least steady - house prices).

So, while on one hand traders as well as other people earned much better than average while the housing market kept going, it failed spectacularly because of the involvement of so many parts depending on each other and ultimately depending on what was thought to be an irreversibly rising housing market.

regards
daytrading
Enter Signature
  • #35
  • Sep 2, 2009 10:53am Sep 2, 2009 10:53am
Quoting Turveyd
So your system simulates millions of trades say or loads per 1M data block being OHLC, in different ways then calculates what to do if anything.


Not seeing this, but interested as a programmer so keep going.
Well I'll go as far as I can without giving away the secret of how the system works. I made some refinements and improvements to it yesterday and heres my findings:

1. I chose 9 random currency pairs
2. I got the typical spreads for these pairs from FXCM.com (google "forex typical spreads")
3. I tested the system from 1 Jul 09 - 2 Sep 09 on each of these 9 pairs
4. It made profit on all 9 pairs, typically about 150-200 pips per pair

I then tested it on EURUSD from 1999-2009 (the early data is full of holes but it made huge profits on those years, remained profitable up to recent months)

I looked at some of the trades it made on my Metatader screen, just to see if they looked "feasable" and they did.

my conclusion: It seems to be a good "one size fits all" system , I have my fingers crossed, but I'm not getting excited until I test it out on the real world.

I've made an idiot of myself before on this forum claiming to have an incredible system and got shot down. I may start a trading journal and/or broadcast my MT4 stats, when I have the system up-and-running in Metatrader. Hopefully then I'll get some respect from all the non-technical traders who seem to hog all the respect.
  • #36
  • Sep 2, 2009 11:12am Sep 2, 2009 11:12am
Quoting FrinkFX
my conclusion: It seems to be a good "one size fits all" system , I have my fingers crossed, but I'm not getting excited until I test it out on the real world.
you are already as excited as it gets...

Quoting FrinkFX
Hopefully then I'll get some respect from all the non-technical traders who seem to hog all the respect.
forget about getting respect. forget about bragging. forget trying to show off. this are not the traits of a succesfull trader and all are things that will inflate your ego to the point where there is nothing but trouble...
Stubbornly persistent
  • #37
  • Sep 2, 2009 8:40pm Sep 2, 2009 8:40pm
so what are you simulating per pair? Is it different parameters for your trade triggers? Are you picking the best results of these 1mil simulations and hoping that that will be your system?

Just curious ...
  • #38
  • Sep 2, 2009 8:55pm Sep 2, 2009 8:55pm
Quoting daytrading
I would not dismiss this as 'b.s.'.

I know a fund in the Cayman Islands who exclusively trades black box algorithmic programs.

We all know that nothing has an unlimited life span - let alone profitability span, but fully computerized trading models have been and still are a major part of modern trading.

regards
daytrading

They also were a huge part of the collapse of everything as well. (Automated selling and buying when certain conditions are met.)

Let's not forget that part either.
  • #39
  • Sep 2, 2009 8:59pm Sep 2, 2009 8:59pm
Quoting FrinkFX
Well I can see I've shut everybody up. For those of you purists who think I'm taking a "short cut"... This is the result of thousands of hours of thinking, failing, trying again, and never giving up!

Catch me if you can!!!
I don't think you are taking a short cut, I think you are looking for something to take the hardest part of trading out of the equation - you. Congrats on your algo. If you are getting results, stop demoing and start trading. Simple as that.

As far as catching you, I am already profitable. And trading live. So it's the other way around.

Good luck,

Chris
  • #40
  • Sep 2, 2009 9:07pm Sep 2, 2009 9:07pm
Quoting M.A.C.Doug
there is no such thing as an infallable system or ea. there are market participants with literally billions of dollars at their disposal for research and developement into auto trading and they still manage to lose money. If such a thing were possible there would be no market. Your system is no different, the only garrantee it comes with is that it will fail. If you are anything like me you will not accept this and continue to waste your time trying to get a machine to do your trading for you, convinced from your useless tests that it is possible....
Well said, MACDoug. Unfortunatley, your sage advice will fall mostly on deaf ears. It is against human nature to accept advice that goes against one's belief system. It is a sad flaw that we must all make our own mistakes, lose our own money, and draw our own conclusions. No one wants to hear "You can't do it". We live in an age of catch phrases like "Believe in yourself", "Follow your dream", and "Don't limit your potential". We are conditioned to believe that if you just want something bad enough, it can be yours. Damn the realities...Full speed ahead. At some point, when we lose enough money, we hear that still, small voice in the back of our head.... and it is saying "Wake Up, Neo!"

Regards
0 traders viewing now