I have spent some time studying python and am picking up machine learning concepts and thought to try algorithmic trading to make fast money - but quickly realized that a lot of hard work goes into making a good algorithmic trading strategy and program. However, the problem of how to write a good trading algorithm with Machine Learning techniques has grabbed me and now I want to work seriously towards developing one. So this post is to ask for what resources you used to start out, how you went about writing your first successful trading algorithm, and perhaps which mistakes you would advise someone else to avoid making. If anyone has a link to a similar post in the past that would be very helpful also.
just how many people do you think have successful algorithms? i started with fortran doing weather forecasting but was too young for punch cards but i did make a many of holiday wreaths from them when my aunt would bring them home.
Key is to identify which periods or days or times have a surge in orderflow. That's all you need. Its data mining and sometimes it just comes from experience with a market.
with respect different opinion here. i have been at institutions where i was the surge in order flow. a surge in order flow is not a guarantee of market movement. momentum readings of price itself is really all you need. volume has no bearing on price direction but price direction does and momentum best tracks that. i never understood why people waste any time what so ever with volume when price is what you trade.
it's all about drawdown, 4 tick risk with profit factor of 1.5 the sky is the limit. i have been thinking 1000 lot x 6 trades a week x 52 weeks x 1 point profit = 15+ mil a yr. and that's on one market with no fixed ratio.