I've been working on a CL day trading system for the last 4 years. Basically so far it's been a system that is over optimized and breaks down eventually and needs re optimization. However I think I've finally come to the point where I filtered out all the noise and I'm wondering if any other CL day traders can look at this equity curve chart and comment on it.

This is a 5 year EC chart :

pepelucho 

I've been working on a CL day trading system for the last 4 years. Basically so far it's been a system that is over optimized and breaks down eventually and needs re optimization. However I think I've finally come to the point where I filtered out all the noise and I'm wondering if any other CL day traders can look at this equity curve chart and comment on it. This is a 5 year EC chart :

It's not hard to generate an equity curve in backtesting like that.

But the real trick is completely out of sample walk forward testing (you can only test against it once, after that it is effectively "in sample").

Add to that, naturally, live forward testing in an actual market.

This has been covered in great detail in numerous webinars on nexusfi.com (formerly BMT). I would suggest starting with the ones from Kevin Davey in the archives, but you can also find many more from other presenters on the site. All the info you could need on how to properly design and test a system is there.

Mike

Big Mike 

It's not hard to generate an equity curve in backtesting like that. But the real trick is completely out of sample walk forward testing (you can only test against it once, after that it is effectively "in sample"). Add to that, naturally, live forward testing in an actual market. This has been covered in great detail in numerous webinars on NexusFi (formerly BMT). I would suggest starting with the ones from Kevin Davey in the archives, but you can also find many more from other presenters on the site. All the info you could need on how to properly design and test a system is there. Mike

right, like I've said, I've been working on this strategy for the last 4 years, and before that I was day trading since 1998. I have tried walk forward optimization testing and it didn't help. I do appreciate your input, however.

I was really hoping to get some feedback from people who have actively traded CL for many years, I get the feeling that there aren't too many successful CL day traders around tho.

pepelucho 

right, like I've said, I've been working on this strategy for the last 4 years, and before that I was day trading since 1998. I have tried walk forward optimization testing and it didn't help. I do appreciate your input, however. I was really hoping to get some feedback from people who have actively traded CL for many years, I get the feeling that there aren't too many successful CL day traders around tho.

I've been trading for nearly 20 years, and traded CL for several years in recent memory.

If you want feedback, you need to post more detail. A screenshot of a backtested equity curve isn't going to get a lot of input from those with real experience.

Mike

Big Mike 

I've been trading for nearly 20 years, and traded CL for several years in recent memory. If you want feedback, you need to post more detail. A screenshot of a backtested equity curve isn't going to get a lot of input from those with real experience. Mike

Good point, here are more details :

it's an intraday strategy and the strategy report info bellow is for trading 1 contract continuously, not ramping up contracts.
This is 60 months of data :

700 trades in four years is really easy to curve fit. What makes you think you didn't?

Sent from my LG Optimus G Pro

Big Mike 

700 trades in four years is really easy to curve fit. What makes you think you didn't? Sent from my LG Optimus G Pro

haha, I don't think I didn't curve fit, as a matter of fact I'm pretty sure that it's a curve fit.

This is like the 10th iteration of this system, so far all previous tries ended up too curved fit. The nice thing, however, is that when the curve fit has broken down in the past, it has done so in very well behaved manner. That being, it didn't have newer and bigger draw downs, it just simply didn't have as big a profit. But it was still profitable.

Anyhow, this is exactly the kind of constructive criticism I was looking for. Thanks.

pepelucho 

haha, I don't think I didn't curve fit, as a matter of fact I'm pretty sure that it's a curve fit. This is like the 10th iteration of this system, so far all previous tries ended up too curved fit. The nice thing, however, is that when the curve fit has broken down in the past, it has done so in very well behaved manner. That being, it didn't have newer and bigger draw downs, it just simply didn't have as big a profit. But it was still profitable. Anyhow, this is exactly the kind of constructive criticism I was looking for. Thanks.

Can be easy to prove. Test against QM or USO. If results are outside in multiple metrics, it's likely over fit.

Personally, I would test this against dozens of symbols and analyze the results. I also prefer to trade from a portfolio level instead of single instrument.

Watch the webinars in archives. Tons of info on this.

Sent from my LG Optimus G Pro

Also slightly vary timeframe. Change from 1000 tick chart to 1100 for example, assuming tick data. If results are wildly different, than likely curve fit.

Sent from my LG Optimus G Pro