Looking for Traders who would like to test out my ML based trading systems

[MLTraders]

Hi Guys,
I struggle to find traders who are interested in ML trading. I have developed a generic ML framework and applied it to trading. Currently I use CTrader.
Not many traders
I have also developed trading algos from first mathematical principals.
Please email me if you would like to evaluate my tools on demo or live accounts.
Sorry I am not trying to spam anyone.

Best Regards,
Alistair

 

[Willzy]

Hey Alistair,

I too use ML for my trading, I have mostly focused on stocks and used the basic methodology in HBandy's book QTA. I've had good success thus far.

What ML methods are you working with?

Regards
Matt

 

[MLTraders]

So sorry for the very late reply! I am still testing my bots, Ctrader had a big change to the their framework that stuffed me up for a long time.
I have built a generic ML framework based on microsofts latest tools. I can build models in minutes on GB of data. My backtest results are incredible but I am still struggling with live trading. I could use some help testing my bots if anyone is interested.
Here are some recent backtest results. I can build models on any timeframe, range, hk, tik and for any market on CTrader.

 

[martyjames]

Sounds interesting. Are the systems mean reversion primarily?

 

[MLTraders]

Mate once you give me a dataset of features the ML can predict any of its values or you can massage the data to say if the market will go up or down and use the data to predict that. This dataset of features is called a vector. I have around 100 indicators and many custom/best practise indicators from the CTrader community. If you are new to ML it only works to predict one of the variables of your set. However this ML has the advantage that any of the features can also be vectors and I use this. So its actually mult-dimensional ML. But the ML is not using a time series prediction method, that is really only for very seasonal type phenomena. The ML can do any classical type of ML such as regression or categorisation. Currently for trading I have only implemented those types of models, but I could also do recommendation, outlier models. I can even do image models and I don't need to use transfer learning which is very powerful as most python image models use that.

 

[MLTraders]

Why the Government will continue to target super and the pension

An aging population will drive policy decisions.

www.morningstar.com.au

I think you need at least 2 mill in the bank to get 50K interest a year but who can live on that? Everyone probably needs to supplement their incomes such as through trading.

 

[divs4ever]

Why the Government will continue to target super and the pension

An aging population will drive policy decisions.

www.morningstar.com.au

I think you need at least 2 mill in the bank to get 50K interest a year but who can live on that? Everyone probably needs to supplement their incomes such as through trading.

well i could probably live on $50 K a year ( no car , no kids , no bevy of beautiful babes )

but how long is the money safe in the bank ( from corporate failure or dipping from Government departments ) the government only guarantees the first $250,000 (per ADI )

 

[MLTraders]

I don't think I could have to move to Thailand, I have kids and extended family OS to look after.

 

[qldfrog]

Hi ML traded,
Not trading now as short of fund until farm is sold but will get back to you once done

 

[MLTraders]

Thanks mate, well maybe you can just test on demos now?

 

[MLTraders]

I can also send anyone screen shots of my recent backtest and live trading. I need to get a new server at the moment.

 

[investtrader]

What is ML trading? I could google it but couldn't be bothered

 

[MLTraders]

No problem, machine learning is really what most AI is. ML is about letting a machine use statistical methods to predict future values of data values based on previous values or to break data values down into categories. Most classical ML methods are decision tree based, more modern ML often use neural network NN methods. However NN methods would not be good for trading, they are good for phenomena that obeys consistent laws. Also classical ML methods such as decision trees are more robust.

 

[MLTraders]

My system uses these classical ML methods that are well accepted but as the code is all C# models can be built much faster and the results are actually better so others tell me.

 

[investtrader]

Thanks for the description. Does Jim Simons use these models?

 

[MLTraders]

Yes of course they would be, he was probably one of those practical early experimenters in ML who had success while the academics were floundering, it was called the AI winter from the 1990s till about 2010. But of course they use their knowledge of markets as well and insider knowledge and probably all these hedge funds colude but they can't be caught

 

[qldfrog]

Yes of course they would be, he was probably one of those practical early experimenters in ML who had success while the academics were floundering, it was called the AI winter from the 1990s till about 2010. But of course they use their knowledge of markets as well and insider knowledge and probably all these hedge funds colude but they can't be caught

Ahh the AI winter,
The frog graduated in AI in 1991
Talk about timing....
Luckily, I had also much training in "normal" computer sciences...
There is much which can be done nowadays so good luck @MLTraders