Neural networks are pretty magical, they can create features that are predictive of price by virtue of some high-dimensional pattern recognition. BUT, it's not easy to create good neural networks in trading, and I've read a lot of papers on neural networks in trading and boy are they practically impossible to parse. I want to keep my articles on neural networks practical, and you should walk away from this thinking: "Huh, that's all there is to it, doesn't seem too difficult to try." Here's how to implement a convolutional neural network that can predict short-term price direction from limit order book data with > 80% accuracy. As usual, happy to share articles with anyone who comments and retweets on THIS post.

There are signals that just "makes sense", and when you find them, you need to do almost no kind of backtesting and fitting. This is one such example. Once you understand this signal, you will go, "That totally makes sense, why didn't I think of that?!" Most people don't understand what it means to be "creative". Every signal must be momentum or mean reversion or *shudders* pairs trading. I hope to share with everyone that good signals can be found by thinking deeply about behavioral effects and existing algorithmic signals. Random comments and retweets will get this article for free.