Trading ADA and SOL back and forth
My first steps in pairs trading
Although I have had a cryptocurrency portfolio for quite a while, my investments have all been long-term. I only put money into projects I believed in. Never speculated on meme coins, never jumped on the hype train for some shiny new thing. At least, I used to.
I’ve never had the 1) time, 2) energy, and 3) expertise for day trading.
Recently, as my book approaches completion, I suddenly found myself with plenty of the first two. As for the third, well, that depends on the time and energy invested. Over the past few days, I’ve invested quite a bit of both. It’s Christmas break, after all.
This post is about my first steps of developing my first automated and profitable cryptocurrency trading strategy.
Disclaimer. I am not an expert in trading. This post is not investment advice. Do not follow it blindly. My goal is to document my experiences with trading and share my ideas. Especially the bad ones, because those are the ones we learn the most from.
What’s a dream scenario for a trader? Take a look at this (totally made-up) candlestick chart.
If the price movement of any crypto asset looked like this, trading would be ridiculously simple. Open a long position at the bottom. Open a short position at the top.
The bad news: this market doesn’t exist. The good news: this doesn’t stop me from looking for one. I’m a mathematician, so I possess an incredible amount of imagination, confidence, and naivety.
How can we find something like this?
It’s a well-known phenomenon in crypto circles that almost everything moves in tandem with Bitcoin, and beyond that, many assets also correlate with each other. Check out this chart where I overlaid the ADA:USDT and SOL:USDT price movements. (Since the two coins don’t operate on the same scale, I plotted the percentage change relative to the initial time point.)
They move together quite close, don't they?
This phenomenon gave me an idea, and I'm currently working on turning it into a profitable (and automated) strategy.
The idea in a nutshell:
We look for two cryptocurrencies, X and Y, so the price spread X − aY is stationary for some constant a. This is called cointegration, and while it sounds complex, it essentially means that X − aY behaves like white noise — bouncing around a certain value without drifting away from it.
Once we find such a pair, we calculate the X − aY spread and normalize it by removing its mean and variance. This gives us the so-called z-score, a common method for scaling time series.
When the z-score is high, we short Y and long X. When the z-score is low, we do just the opposite. We exit the positions when the z-score goes back to zero.
This is called pairs trading, and once I discovered the term itself, lots of lightbulbs went off. You know: once you have the name, you have the power. This is especially true for research.
I found the best material on the now-defunct Quantopian's YouTube channel, which included a video and a Jupyter Notebook that I'm using as a reference. (Here's the video and the Jupyter Notebook.)
The strategy is dead simple. However, I'm incredibly lazy and have no intention of sitting in front of my screen all day, watching trading signals and executing trades manually. So, right from the start, I've aimed to automate everything.
Currently, I'm working with PineScript, TradingView's custom scripting language. I've already created my own indicator to track the normalized spread. (You can check it out live here, and the PineScript source is at the end of this post. I won't lie; I did the entire thing with ChatGPT. Did I mention that I'm lazy?)
So, I have a signal; now I just need a bot. My plan is to set up a webhook via 3Commas to receive signals sent from TradingView. It sounds complicated, but I promise you: once I figure it all out, I'll write a hell of a tutorial post on the whole process.
But that still requires some work — and a ton of learning.
As promised, you'll find the spread z-score indicator below.
//@version=5
indicator("Spread Rolling z-Score", overlay=false)
// Input for asset symbols
asset1 = input.symbol("BINANCE:ADAUSDT", title="Asset 1")
asset2 = input.symbol("BINANCE:SOLUSDT", title="Asset 2")
// Fetching prices for selected assets
asset1_close = request.security(asset1, timeframe.period, close)
asset2_close = request.security(asset2, timeframe.period, close)
// Spread Calculation
spread = asset1_close - asset2_close
// Moving Average and Standard Deviation for Z-Score
window = input.int(30, title="Moving Average Window", minval=1)
ma_spread = ta.sma(spread, window)
stdev_spread = ta.stdev(spread, window)
z_score = (spread - ma_spread) / stdev_spread
// Plot Z-Score and thresholds as an oscillator in a separate pane
plot(z_score, color=color.yellow, linewidth=2, title="Z-Score")
threshold = input.float(2, "threshold (STD)", minval=0.0)
hline(threshold, "Upper Threshold", color=color.red, linestyle=hline.style_dotted)
hline(-threshold, "Lower Threshold", color=color.green, linestyle=hline.style_dotted)
hline(0, "Zero Line", color=color.gray, linestyle=hline.style_solid)
// Background shading for extreme Z-score regions
bgcolor(z_score > threshold ? color.new(color.red, 90) : na, offset=0)
bgcolor(z_score < -threshold ? color.new(color.green, 90) : na, offset=0)This is how you can add it to your TradingView terminal:
