https://www.quantstart.com/articles/Forecasting-Financial-Time-Series-Part-1. I'll leave this here if you are interested.
Trading Catechism
- Thread starter nitro
- Start date
Code from Ernie Chan's book.
https://www.quantopian.com/posts/so...ail&utm_term=0_7cca55778a-7ff2c43c24-99379153
https://www.quantopian.com/posts/so...ail&utm_term=0_7cca55778a-7ff2c43c24-99379153
I am interested, Thanks Gambit.
Gambit those are both interesting. I have seen them before and both of those guys, especially Ernie Chan, should be credited with being one of the guys that genuinely tries to bring these techniques to people that couldn't code these ideas on their own.
I own Mathematica and Matlab, and I have R installed. One thing that I would always want to do (in the absence of Test Driven Code which the Python code should be, imo) is go to the original code to make sure the ports are agreeing with the original. As pointed out by all the posts below on that site, there are all sorts of disagreements from original to port. The site points to Matlab code, but when I try to go there, it gives a log in prompt?
I own Mathematica and Matlab, and I have R installed. One thing that I would always want to do (in the absence of Test Driven Code which the Python code should be, imo) is go to the original code to make sure the ports are agreeing with the original. As pointed out by all the posts below on that site, there are all sorts of disagreements from original to port. The site points to Matlab code, but when I try to go there, it gives a log in prompt?
Nitro, are you talking about one of my links? Both my links worked. Let me backtrack and see if I posted another link.
Gambit, I am asking about the original Matlab code to Chan's book.
You might have to buy the book nitro. I have a copy but I'm travelling now so it isn't here. Perhaps, I can scan some of the matlab code and PM it to you sometime. That way you can compare.
@nitro Great thread and interesting thoughts.
I'd like to breakdown and clarify your original premise, in minimising directional risk. I assume you do this by employing various mean-reversion AND trend following strategies together.
In the trend following case, yes the trades may be directional but the diversification across many asset classes ensures overall risk is directionless. Likewise mean-reversion strategies utilising co-integration techniques to form stationary series ensures there is no directional risk. Additionally, cross-sectional momentum strategies are directionless and market neutral, thus again limited directional risk.
In each case vol (using realised and implied) based position sizing is employed to control risk at the asset class level.
Is this correct? Also, I assume measuring and managing risk across the entire portfolio all of these strategies is the key to profitability and longevity?
I'd like to breakdown and clarify your original premise, in minimising directional risk. I assume you do this by employing various mean-reversion AND trend following strategies together.
In the trend following case, yes the trades may be directional but the diversification across many asset classes ensures overall risk is directionless. Likewise mean-reversion strategies utilising co-integration techniques to form stationary series ensures there is no directional risk. Additionally, cross-sectional momentum strategies are directionless and market neutral, thus again limited directional risk.
In each case vol (using realised and implied) based position sizing is employed to control risk at the asset class level.
Is this correct? Also, I assume measuring and managing risk across the entire portfolio all of these strategies is the key to profitability and longevity?