The ACD Method

[justrading]

Can I offer some advice? Go to DataCamp. It's $6 a month. They basically teach you R and Python doing actual real exercises. I found learning this stuff from scratch is the most painful thing in the world. In DataCamp, they have a course called financial trading in R or Python. The way the course works is they give you the code and you actually do the exercises and learn from doing. So you will learn how to download the stock data, create charts, build an algo, test the algo and run the performance. It's honestly the best short cut. A lot of the courses are even free so you can try it out before you pay the $6 a month but honestly $6 a month is what, a sandwich? It's nothing and you can quit when you want. And they are adding new courses every day.

https://www.datacamp.com/

Thanks Mav, this is great. Yeah, I reckon I can squeeze 6 bucks a month, I pay that for CyberGhost to evade the government censors, another 6 won't hurt when it's all for a good cause.

Your suggestion is spot on, as an adult learning by doing works best for me. I could soak everything up like a sponge when I was a kid, alas those days are long gone.

 

[justrading]

Everything involving programmatically coding something relies on basic coding and statistics. Python is not a secret weapon here (it can be done with any language) it's just that a lot of the "quant" types have gravitated towards it for whatever reason and there are existing libraries available.

All the basic time series calculations are trivial and any basic language won't present a barrier. It's the statistical portion where it's going to get more difficult.

I picked Python because my short research told me it is close to R but perhaps a little easier to learn for someone who has no programming knowledge. Plus it has applications beyond basic data as I mentioned in my long reply to eurusdzn.

 

[jj90]

@justrading If you want simple data manipulation and nothing more, R might be a better way to go. I went with Python because it was a full GPL so I was future proofing but if you just want to run numbers easily(there are many libraries for this) have a look at R.

 

[justrading]

@justrading If you want simple data manipulation and nothing more, R might be a better way to go. I went with Python because it was a full GPL so I was future proofing but if you just want to run numbers easily(there are many libraries for this) have a look at R.

Thanks, will check it out.

 

[copyplus]

Mav or anyone else, can you give me some advice or throw me a bone when it comes to number line derivatives? I read so much about people having success with them, and Mav saying a while back that he trades off derivatives only without caring about the raw scores... so I am puzzled as to why I can't find anything useful in this area. I've tried so many ideas on rate of change or momentum type formulas over multiple lookback periods but keep coming up empty.

I don't know what I am missing or not understanding. Any advice or examples?

 

[justrading]

Mav or anyone else, can you give me some advice or throw me a bone when it comes to number line derivatives? I read so much about people having success with them, and Mav saying a while back that he trades off derivatives only without caring about the raw scores... so I am puzzled as to why I can't find anything useful in this area. I've tried so many ideas on rate of change or momentum type formulas over multiple lookback periods but keep coming up empty.

I don't know what I am missing or not understanding. Any advice or examples?

I posted this a while back, I still hold that view.

https://www.elitetrader.com/et/threads/the-acd-method.170318/page-1299#post-4427159

Truth be told I only get excited with raw scores nowadays when the 5D NL confirms, because when nothing else is showing, that is often the sign of a nice move. Otherwise the derivatives flag the moves.

Only other advice I can offer is don't be late to the party. If you've missed it, don't chase it. That first rate of change is where you start to look for your entry.

 

[copyplus]

Thanks for the reply.

I have looked at the 5 day change but it seems to move around too much. I can't find a signal in there. The 5 day has a mean-reverting quality to it, so if it's been strong then it's probably due to quiet down. A +10 can quickly revert back to a 0 just from normal mean reversion and not imply any weakness. A +4 score can become a -6 on a pullback and then bounce right back. It seems like a lot of noise to me.

Interested in what others think...

 

[sss12]

@Maverick74 or anyone else. I'm trying to incorporate some of the principles of the ACD method to my swing trading (or visa versa). I use ATM options with 7-18 DTE on only those ETF's with the most liquid options that are showing some trend and/or vol.

I started looking at the high low range of the PAST 5 days as my OR and setting my A levels with approx +\- 25% of ATR from there. OR on current day just too tight. Any thoughts or comments ? thanks

 

[baggerlord]

@Maverick74 or anyone else. I'm trying to incorporate some of the principles of the ACD method to my swing trading (or visa versa). I use ATM options with 7-18 DTE on only those ETF's with the most liquid options that are showing some trend and/or vol.

I started looking at the high low range of the PAST 5 days as my OR and setting my A levels with approx +\- 25% of ATR from there. OR on current day just too tight. Any thoughts or comments ? thanks

Sounds kind of like a weekly level. Why not just use that? I think its great to try out new ideas like that though.

 

[baggerlord]

Also, I personally only use daily levels to generate NL scores. I have one daytrade method that trades off daily levels in forex I am autotrading right now but I suspect it will only be useful as a means to get a better than random entry when I want to open a longer term position.

An approach you might wish to consider is to watch a 5 day NL and close positions early if it looks like they might be shifting against you.