Okay, so today I’m gonna talk about something I messed around with recently: Hurst exponent and something called Schultz method. Don’t ask me where I heard about it, just know I did, and figured, “Why not give it a shot?”
First off, I started by trying to figure out what the heck the Hurst exponent even is. From what I gathered, it’s basically a way to measure how persistent or random a time series is. Like, if it’s high (closer to 1), it means if it went up yesterday, it’s probably gonna go up today too. Low (closer to 0), it’s basically a coin flip.
I decided to use Python, ’cause that’s what I’m most comfortable hacking around with. I grabbed some random stock data from Yahoo Finance – nothing fancy, just some daily closing prices. Then, I found a Hurst exponent implementation online. There are a bunch of libraries out there, but I just grabbed a simple, straightforward function.
I ran the thing, got a number. Now what? Well, I played around with different stocks, different time periods. I noticed some stocks had consistently higher Hurst exponents than others. Made me wonder if I could use this to predict stuff. (Spoiler: I couldn’t, not reliably anyway).
Next up: Schultz method. This was a bit more confusing. From what I could understand, it’s a way to estimate the number of trends in a time series. Kinda like finding turning points, but with a fancy name. Again, I found some code online (thank you, internet!).
The Schultz method involved doing some calculations on the differences between data points. I won’t bore you with the details (mostly ’cause I barely understand them myself). I ran it on the same stock data. It gave me some numbers that I think represent the number of trends.
Here’s where it got interesting (or, at least, mildly amusing). I tried combining the Hurst exponent and the Schultz method. My idea was: if a stock has a high Hurst exponent and a lot of trends, maybe it’s a good candidate for something? I honestly don’t know what exactly. It was more like, “Let’s see what happens!”
I plotted the Hurst exponent against the trend count for a bunch of different stocks. Didn’t really see any clear patterns. Some stocks had high Hurst and low trends, some were the other way around, and some were just all over the place.
Ultimately, did I discover some amazing trading strategy? Nope. Did I get rich? Absolutely not. But I did learn a little bit about Hurst exponents and the Schultz method. More importantly, I got some hands-on experience messing with data and trying to make sense of it. And that, my friends, is always a win.
The code was messy, the results were inconclusive, but the process was fun. And who knows, maybe this will spark some other idea down the line.
