Nice! Maybe you could post some real-(or near real)time forecasts so we can see how well it actually performs. Finding patterns ex post facto is one thing, and interesting; but using them to trade profitably is another. Thanks
I will add a function to my program to twitter it's transactions ... I'll post the user id when I test it.
Interesting! I'm new to this forum - just found it. I have been working on my own for quite a while, developing software to experiment with ideas, so it's great to finally read about other's endeavours. I've developed my own ensemble voting system using neural networks and more recently genetic programming - I don't think I'm quite ready to hook it up to an automated trader yet though, or pack in the day job (C# dev). I started a blog https://theartificiallyintelligenti...5/12/13/algorithms-ai-and-volatility-trading/ but don't think any one has read it!
The program has been doing trades (real) for months (as a trade assistant, providing recommendations and single button buy/sell/stop loss transactions). Finished the first production version last week (Monday) with the full automation. Now it is running on its own. It is returning about 10% currently, working on Version 2 now.
10% in what time frame? Annual? Looks not like that much for that many work and effort behind the system, isn't it? How much capital does it trade?
What is the nature of the patterns your setup is looking for? I've seen several approaches such as Toby Crabel's "Day Trading with Short Term Price Patterns and Opening Range Breakout" where he did statistical studies of 2, 3, 4, and 5 bar patterns. Also, they're doing this sort of thing very rigorously at asirikuy.com