Discussion on Leo's Neural Networks

GraHal · 2019-09-05T09:42:20+00:00

Leo did a loads of commendable work on Neural Networks and was kind enough to share with us while he traveled the lonely path … I think nobody could help because nobody understoood it all (?) … least of all me! 🙂 Anyway, Leo found limitations on the PRT pathway and is now forging ahead with new tracks and fighting Pythons along the way! 🙂 @Leo left us with below … Leo wrote: here is the best approach I got Neural Network v2.2.1 https://www.prorealcode.com/topic/neural-networks-programming-with-prorealtime/page/8/ I made v2.2.1 into a workable strategy which I offer for further development and sharing on this Thread. We’d love to see you around in the future Leo … hope the Pythons don’t squeeze the life out of you, as I’m sure you must have felt like with PRT at times!? 🙂 Best Wishes GraHal PS hope all’s good with your baby, wife and family.

Viewing 6 posts - 1 through 6 (of 6 total)

Author

Posts
09/05/2019 at 10:42 AM #106572 quote

GraHal
Participant

Legend

Leo did a loads of commendable work on Neural Networks and was kind enough to share with us while he traveled the lonely path … I think nobody could help because nobody understoood it all (?) … least of all me! 🙂

Anyway, Leo found limitations on the PRT pathway and is now forging ahead with new tracks and fighting Pythons along the way! 🙂

@Leo left us with below …

Leo wrote:

here is the best approach I got Neural Network v2.2.1

https://www.prorealcode.com/topic/neural-networks-programming-with-prorealtime/page/8/

I made v2.2.1 into a workable strategy which I offer for further development and sharing on this Thread.

We’d love to see you around in the future Leo … hope the Pythons don’t squeeze the life out of you, as I’m sure you must have felt like with PRT at times!? 🙂

Best Wishes
GraHal

PS hope all’s good with your baby, wife and family.

Leo.jpg Leo-2.jpg LeoNEURNET-v2.2.1-DJI-30M-v1.itf

09/05/2019 at 12:01 PM #106597 quote
GraHal
Participant

Legend
Above .itf needs below as Line 3 … trade was rejected at 2000 preload bars. but trade just executed at 10,000 on Real data.
```
DEFPARAM PRELOADBARS = 10000
```
09/07/2019 at 11:19 AM #106762 quote

Roger
Participant

Veteran

Hi Grahal, all this work on neural networks is very impressive, thanks to you and Leo. I have tried to backtest LeoNEURNET-v2.2.1-DJI-30M-v1 but when I try to test it on live and to launch the code for live trading (with a small size of contract) PRT says that I need to replace variables by values. Have you met the same problem ? Thanks in advance.

09/07/2019 at 11:35 AM #106763 quote

GraHal
Participant

Legend

RGKP wrote:

thanks to you and Leo.

That should be a BIG BIG thanks to Leo and maybe just maybe a very small thanks to me! 🙂

All I’ve done is manage to turn Leo’s Neural Network Indicator into trading Systems and I hoped others may have done a better job than me, but it appears none have been inspired enough to have a go? I guess because Leo’s code is so bespoke and personal?

RGKP wrote:

I need to replace variables by values. Have you met the same problem

I guess you are trying to run my .itf EXACTLY as I provided above?? If Yes, then you need to read up on how to use the Optimiser.

PS I dont know why that appears as bold font … I did not select bold and I am not shouting! 🙂

09/07/2019 at 12:30 PM #106768 quote

GraHal
Participant

Legend

GraHal wrote:

If Yes, then you need to read up on how to use the Optimiser.

I had to get off quick earlier as I could tell that PRT was hogging my RAM and I may have lost all my work if / when PRT crashed!

You need to enter the values after optimisation into the variable as shown in the code by Line number / variable name and then delete the variable from the Optimiser and then you can run the System to execute Live trades.

But best you read up a bit on how to use the Optimiser … follow the link below.

https://www.prorealtime.com/en/pdf/probacktest.pdf

09/07/2019 at 1:05 PM #106770 quote

Roger
Participant

Veteran

I guess I have not made my homeworks 🙂 Sorry for that I am a fresh newbie in Pro Real Code (even if I m used to backtest strategies for a while, with other softwares). I m going to learn it and get back here if I can help for any improvement/idea. Thanks.

GraHal thanked this post
Author

Posts