doh
Junior member
- Messages
- 26
- Likes
- 4
This trade is now (just) in profit. I let it run until the NN gives an exit signal - might still be a loser, but I don't second guess.
This trade is now (just) in profit. I let it run until the NN gives an exit signal - might still be a loser, but I don't second guess.
doh
Junior member
- Messages
- 26
- Likes
- 4
NNs
I've seen quite a lot of discussion about NNs and just to add my bit - it's all about the inputs. Good inputs and a dishwasher will deliver a decent system. Fussing about NN internals is largely a waste of time in my opinion.
PS Don't be discouraged by the previous loss - even with 75% wins you can get three or four losses in a row (but you can also get fifteen winners in a row).
I've seen quite a lot of discussion about NNs and just to add my bit - it's all about the inputs. Good inputs and a dishwasher will deliver a decent system. Fussing about NN internals is largely a waste of time in my opinion.
PS Don't be discouraged by the previous loss - even with 75% wins you can get three or four losses in a row (but you can also get fifteen winners in a row).
DionysusToast
Legendary member
- Messages
- 5,965
- Likes
- 1,501
What happens if you get 10 losers in a row with this system?
Will you pull the plug before it gets to that point?
Will you pull the plug before it gets to that point?
doh
Junior member
- Messages
- 26
- Likes
- 4
I would pull the plug at 5 losses in a row. Fortunately I don't just trade the DAX - I might have up to 20 trades concurrent at any one time. Sounds like a lot of risk, but if they are poorly correlated instruments the risk is pretty much the same as a single trade. The S&P for example is currently long and has compensated for the DAX loss. I know what drawdowns are acceptable for any particular NN and if DD goes beyond that I retrain the network.
So at the current time I am long GBPUSD, short USDCHF, long Apple (despite earnings), long Alcoa, short AUDCAD, long S&P. There is of course always some level of correlation - but it's weak enough to live with.
Of course it may all go belly up - but so far this year I have done quite well - but I don't bet the farm on it.
If it does work long term then it really will prove to me that NNs are not the defunct things many people believe them to be, but that some imagination where inputs are concerned is all that is missing (and a good output - don't try to predict price).
Dick Lexic
Veteren member
- Messages
- 4,954
- Likes
- 124
Years ago I had a screen saver called MoneyBee, it was a neural network kinda thing and as far as i can remember it used the power of my computer and many other computers throughout the world to generate signals for the Dow Jones and other indices. I have no idea what happened to it and my searches have drawn a blank...have you ever came across this doh?
doh
Junior member
- Messages
- 26
- Likes
- 4
No I'm afraid not - sounds interesting though. If it was successful someone probably bought it or kept it to themsleves.
J
jcl365
Have you tried a decision tree instead of a neural network? I'm currently experimenting with them and found that a decision tree is not the holy grail either, but seems to be more useful than a neural network.
Similar threads
