Anybody is interested in collaboration of Matlab trading system?

mizhael · Jan 11, 2010

I am not looking for a copy of Matlab. I have both Matlabs at company and at home.

I like both R and Matlab. Both have large user bases and active communities.

R is terrific for doing some statistics. However the debugging of R is horrible - it has no step-by-step debugging. Therefore making sure the R code is correct itself is not an easy task when the project gets large.

Matlab has less stats functions; but its step-by-step debugging is great and productive. I would like to mainly use Matlab but call R from within Matlab if needed.

I am looking for collaboration and discussion of using Matlab to explore and backtest trading strategies. Through collaboration, we could save some time.

Things on our list:

1. Find a good Matlab backtest/exploration platform for trading strategies (if any), to serve our base; or at least look around in other trading platforms to see what features are neccessary for our platform;

2. Find good open source technical analysis tools in Matlab, or in other languages that are easily incorporatable into our platform.

any more thoughts?

bwana · Oct 6, 2010

I'm interested. I've got a few neural networks setup, some that incorporate Renko boxes as well.

What kind of collaboration are you looking at?

mizhael · Oct 6, 2010

Quote from bwana:

I'm interested. I've got a few neural networks setup, some that incorporate Renko boxes as well.

What kind of collaboration are you looking at?

Collaborate on backtest of strategies ...

and share the outcomes together...

any success stories using NNs?

bwana · Oct 6, 2010

Not really, i'm running off weekly so progress is low. Highest drawdown was ~23% while i had it running a few months ago. Tried it for a total of 7 or 8 weeks and it was stagnant.

The first nn i tried, i used all OHLCV of all 32 pairs from MT4 as inputs with the target being the 'time + 1' close of GBPJPY.

The second used only /JPY crosses, same target.

The third / fourth used indicators of all 32 pairs / only /JPY crosses to forecast 'time + 1' of GBPJPY.

the fifth is similar to the third in that it uses indicators of all 32 pairs with a renko box of 3.0 (300 pips) for the close of GBPJPY at 'time + 1'.

I realize that this might be clear as mud so please let me know where i should elaborate.

Have you had progress with nn?

WinstonTJ · Oct 7, 2010

Quote from mizhael:

Collaborate on backtest of strategies ...

and share the outcomes together...

any success stories using NNs?

Do you still have a group of people working on strategies or has that shrunk down to just you?

I'm in NYC as well. Where do you have an office and where do you keep your hardware? Do you have tick data or any historical data? Who would you be looking to execute through?

JoePaterno · Oct 7, 2010

mizhael is not necessarily someone you are going to get any value out of. Aside from occasionally trying to probe for females on ET, his posts indicate it is more likely trying to tap your ideas.

mizhael · Oct 8, 2010

Quote from JoePaterno:

mizhael is not necessarily someone you are going to get any value out of. Aside from occasionally trying to probe for females on ET, his posts indicate it is more likely trying to tap your ideas.

ooops, I begin to have enemies on ET... which is quite good.

well I am trying to get people engaged in discussions... I think there is nothing wrong with that.

mizhael · Oct 8, 2010

Quote from JoePaterno:

mizhael is not necessarily someone you are going to get any value out of. Aside from occasionally trying to probe for females on ET, his posts indicate it is more likely trying to tap your ideas.

btw, JoePotato:

I curiously looked into your past posts and wanted to see if your posts added any value...

you know how disappointed I was?

-----------------------

Okay I will stop here; the message I wanted to get across is:

don't be that judgmental or mean - it's all subjective...

mizhael · Oct 8, 2010

Quote from bwana:

Not really, i'm running off weekly so progress is low. Highest drawdown was ~23% while i had it running a few months ago. Tried it for a total of 7 or 8 weeks and it was stagnant.

The first nn i tried, i used all OHLCV of all 32 pairs from MT4 as inputs with the target being the 'time + 1' close of GBPJPY.

The second used only /JPY crosses, same target.

The third / fourth used indicators of all 32 pairs / only /JPY crosses to forecast 'time + 1' of GBPJPY.

the fifth is similar to the third in that it uses indicators of all 32 pairs with a renko box of 3.0 (300 pips) for the close of GBPJPY at 'time + 1'.

I realize that this might be clear as mud so please let me know where i should elaborate.

Have you had progress with nn?

Nope, I had never got a SharpeRatio>0.6 with NN...

mizhael · Oct 8, 2010

Quote from WinstonTJ:

Do you still have a group of people working on strategies or has that shrunk down to just you?

I'm in NYC as well. Where do you have an office and where do you keep your hardware? Do you have tick data or any historical data? Who would you be looking to execute through?

PMed you