LOVINIDA - CL automated strategy

[vit007]

Hello fellow traders. I just finish working on my new automated strategy and before I put it live I want your opinion --- what week spots it has and what parameters can be improved . Its NT7 base , 2 contracts and best for major active futures.
Thanks in advance
Vit

 

[FaceOff]

I would recommend testing over a multi-year period, rather than 2 months. You will also need a strategy for handing roll-overs, since this will happen once a month.

 

[vit007]

I would recommend testing over a multi-year period, rather than 2 months. You will also need a strategy for handing roll-overs, since this will happen once a month.

The reason I pick last 2 month because CL market was neutral . The last year market was more like exception so its hard to rely on past results.
Any way 2014 till now looks like this ,with 2 contracts..

 

[vit007]

The FDAX is the craziest one ...

 

[FaceOff]

For me, whenever I see a chart like that, it looks like curve fitting. And not sure how you can tell if a market is neutral or not until after the fact. Turning an automated system on/off depending on market conditions can result in missing the most profitable trades.

 

[jelite]

Hello fellow traders. I just finish working on my new automated strategy and before I put it live I want your opinion --- what week spots it has and what parameters can be improved . Its NT7 base , 2 contracts and best for major active futures.
Thanks in advance
Vit

Vit, I only looked at your first attachment. You are claiming about 17 ticks profit (on 2 contracts I assume), that's almost 8 ticks per trade net alpha with an average of more than 6 trades per day. So essentially you are saying you can capture net alpha of 50 ticks per day on CL in an automated way. That alone is a clear indication that you must have a serious error in your backtest (or it's a total curve fit). Such results are probably only achievable on an hft market making level. Sorry, it must be wrong.

 

[lucysparabola]

Sharpe on CL is >3x better than on Dax...

 

[vit007]

Vit, I only looked at your first attachment. You are claiming about 17 ticks profit (on 2 contracts I assume), that's almost 8 ticks per trade net alpha with an average of more than 6 trades per day. So essentially you are saying you can capture net alpha of 50 ticks per day on CL in an automated way. That alone is a clear indication that you must have a serious error in your backtest (or it's a total curve fit). Such results are probably only achievable on an hft market making level. Sorry, it must be wrong.

I downloaded new NT7 test it and the results are the same . Any idea how I can test it differently???

 

[vit007]

If strategy shows profit on multiple markets (with some minor parameters change) its should do OK live . Ive start demo it just 3 day ago and planning to connect to Collective2 .
Here some for TF ..

 

[FaceOff]

I papertraded my CL system for 3 years prior to going live with real money. That's how long it took to work out all the glitches and remove every trace of curve fitting. Just looking at a chart is curve fitting! In order to have a profitable system you need to work harder and be smarter than 99% of the other market participants, otherwise you will loose money in the long-term. You are not going to suddenly discover a long-term profitable trading system without an immense (ie. years) amount of effort.