Lorentzian Classification

Question 1: What is Lorentzian Classification?

A tool that learns from history: it describes today's market with a few indicators, finds the past moments that looked most alike, and predicts up or down based on what those similar moments did next.

Question 2: Why 'Lorentzian' distance?

That's just the maths it uses to measure how 'similar' two moments are. This particular method is better at ignoring weird one-off spikes (like big news days), so its matches tend to be more reliable on price data.

Question 3: How do I trade the signals?

Take its up/down calls in the direction of the bigger trend, and keep its built-in filters switched on so it doesn't fire against the trend.

Question 4: What are the key settings?

How many similar past moments it compares against, which indicators it uses to describe the market (and their lengths), and a couple of trend filters.

Question 5: Is it really ML?

Yes - it's a genuine 'learn from similar past examples' model. But 'similar in the past' is never a guarantee; signals still fail, especially when the market's character suddenly changes.