Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
I'm just getting started with trading so figured now is a good time to start a journal.
A bit about me: I have a PhD in signal processing and optimization and spent the last 10 years in industry as a software engineer developing signal processing and machine learning algorithms. Hopefully this will make me a little more disciplined when it comes to OOS testing and cross-validation. I'm very comfortable on the programming and algorithms side but have very little experience trading.
I've had good success with property investing but that's a very different ballgame I'm sure. It's also hard to tell with property how much was just luck because the number of transactions is so low. On the plus side though I'm well capitalized as a result.
Ideally I'd like to transition to trading as a job probably for the same reasons as many others: being independent and getting out of the corporate world. I figure there is 5% or less chance of this happening which is OK with me. If it becomes a hobby that helps me generate some pocket money that's fine to. Or maybe it just doesn't work at all. It would be nice to find a way to gradually improve through consistent effort and chart a path to success rather than just giving up though.
So on to the plan. Right now I'm reading through Ernie Chan's algorithmic trading and setting up my own back testing environment in Python (although zipline / Quantopian / IB look pretty interesting too as an off the shelf solution). I'm planning to trade only once per day or so, running the algorithms nightly and then entering trades by hand, at least initially. Low frequency should keep transaction costs low and hopefully decrease the number of opportunities more experienced traders have to eat into me with more precise timing and better market knowledge. I really lack in my ability to read price action. It's something I'll need to get a better feel for. Any references on this would be much appreciated.
Ideally I'd like to identify 100 or so algorithms, short list and implement 20, back test them over many different time periods, then forward test them with small size trades. If one ended up being moderately profitable I'd be pretty happy.
The real challenge now is identifying a pipeline of algorithms to try out. I have a preference towards correlation / stat arb and machine learning models for mean reversion but am frankly open to anything. I have a strong bias against Elliott waves, square of nines and other esoteric forms of TA though.
Anyway, I'm looking forward to getting to know some of you better and setting out on this journey.
As you want to identify quite a big number of algorithms to trade with, i believe you have to increase your trading frequency (and number of tradable instruments also) because then only you will come to know about real life behaviour of the algorithms.
In practical terms all statistical testing will only help you to sort the good algorithms from the bad ones(from historical perspective only!);although future utility of any algorithm is still very much subject to chance (especially so in the financial markets).
Positive expectation from a histirically proven algorithm/strategy is as valid as the hope that..."history repeats ....".
All the best.
According to your research, have you categorized the market conditions. Or do you simply distinguish between range bound vs out of range/balance conditions?
There are many different conditions
One need to understand the strength and the weakness of each algorithm and when to use it ...
You can compare it to a doctor, making a diagnose and then subscribing the right medicine
market condition depends on many factors and a trader should have an arsenal of tools
some tools are for certain conditions
You need to have a controller that knows when to 'enable' and when to 'disable' an algorithm.
There are many more conditions that can trigger/halt a algorithm than 'range vs out of range/balance conditions'
think about correlation, statistical, ...
Kevin Davey is the Founder and CEO of KJ Trading Systems and will be monitoring this thread so that he may answer any questions that you post here relating to his products or services, primarily focused on algorithmic trading systems.
Please keep in …
I am not an algorithmic trader, and so I couldn't say whether these would help much or not.... But Kevin ( @kevinkdog ) is a frequent poster and a successful trader, and, to my non-algo eye, makes useful contributions and seems to know what he is talking about.
So that might be something to look at. There will be some others also, but these just came to mind.
You could also just go up to the Search box in the upper right corner of every page and try some relevant search terms. You might have better luck opting to use Advanced Search (basic Search just looks for matching thread titles, Advanced looks for text in the posts.)
There are fewer algorithmic traders here than there are discretionary, but some searching may find you something that may be of value to you.
Can you list the conditions you have worked out so far in your own tests? Are these conditions part of the algo that you have developed or are these conditions only identifiable by a human?