Hi Kenneth,
Thanks for sharing this little trading system. My first thought was this result with two averages and one oscillator must be curve fitted! But after playing around I have to admit that you have found a promising setup. The simplest things are the best:-). I tried to make it better and the only little improvement I found so far was the usage of a percentage stop/take profit.
Best regards,
Reiner
@Despair,
I have removed the “averaging down” part of your code (personal choice based on my experiences), and applied only the MM to my strategy…The difference is awesome… but somehow scares me too … I feel that one morning my account may be blown up !
However – You have been doing an excellent job Despair if I may !
Thank you.
@Inertia: I also run the code with deactivated averaging down. Just set averagedown=0 and it is off. To the MM. I don’t think it is so awfully risky, I run it in most of my live strategies. I think the major risk is that you lose accumulated profit faster. To really damage your account it would need that the strategy makes some profit over some time to increase the position size and then there must come a VERY bad day. If the losses come little more slowly with several trades the strategy is automatically reducing position size again. So I think the risk is bearable.
@bertholomeo: I uploaded the version with MM. In this code you find the MM. It is even marked out as “fixed fraction MM”. Have a look.
@inertia: I thought again about the risk of the MM and not even a VERY bad day like i wrote before would any harm since we never trade without stop loss. So I think the risk of blowing up your account are low. The position sizing is only active when there is profit. By adding MM you will get a much more volatile equity curve. How volatile depends on the parameter delta. Here you can adjust how fast the position size is increased.
REINER: do you use fixes % with the original trailinstop?
Hi Kenneth,
Yes, the version above is with fix %values for stop/take profit and your original trailingstop setup. I optimized all parameters as well.
I also tried % value for the trailing mechanism. This version delivers the best profit but drawdown is also higher.
Your idea is good and I have added it in my life portfolio!
Thanks, Reiner
Hi Reiner,
When you optimize the parameters.How long lookbackperiod did you choose?
Hi Despair,
I played around with 200k candles. The difference to Kenneth original version isn’t really significant.
Did you try other instruments?
Just a general question to optimizing. Is it really superior to optimize the parameters on all data we have? If we have run a WFA with an in-sample period of say 8 months and than 3 months out of sample. Then we are lucky and have a strategy that seems to walk forward nicely. Wouldn’t it be than the most reasonable to re optimize the parameters with the recent last 8months as insample period and run this new (fresh) parameters live?
I thought this is how one uses WFA. Optimizing over all data should give better backtest but worse out of sample performance. Or did I understand something wrong? @Nicolas if you read this, please comment.
At least this is what I do. But I got a little bit unsure because I see you optimize over the whole period.
I tried some other assets. The only which gave good results was London Gasoil but we don’t have a lot of historical data for this asset so the testperiod gives only about 50 trades. Another “problem” with london gasohol is that even the mini-contract is worth 50USD/pip. So already a stop of 20 pips gives you 1000 USD DD.
Also tried chicago wheat. The result was not hopeless but same problem here with limited testdata.
REINER:can you please share your update on the strategy. So other dont need to use 6 houres optimisation 🙂 would be much apriciated 🙂
Hi Kenneth,
Sorry for the misunderstanding but I already attached the code here. The optimization is “old style with 200k” and no WFA because of lack of time.
Strategy can be optimised for us crude and only long on the dax.
Yes, I had also good results for CL but CL and LCO are so tightly correlated that it is pointless to run both. At least if one has diversification in mind.
Someone would be interested to produce on prorealtime a strategy similar to that in attached link (breakout on gbpusd) ?
http://www.myfxbook.com/members/autotrade/lucky-pound/1045415
I see no code or trading rules?
