Learning Machines, Reading Chains
Why I am skipping toy problems and throwing ML into crypto's chaos
Most people learn machine learning by predicting house prices or classifying cats.
That’s boring. That’s safe. That’s useless.
The world doesn’t need another Kaggle notebook that guesses the cost of a three-bedroom in San Francisco.
What it does need are models that can make sense of the chaos on-chain i.e. wallets, tokens, NFTs, contracts, the rawest dataset humanity has ever created.
That’s where I’m starting.
I’ve joined Analytics Sages with one goal: build ML models on crypto data and deploy them in the wild.
No toy problems.
No sanitized datasets.
Just the mess of real transactions, volatile markets, and human greed written in code.
Each phase is a fight
First, exploring crypto datasets and squeezing insights out of noise.
Then, predicting token prices with regression.
Next, classifying wallets: good trader or bad trader, diamond hands or paper hands.
After that, deploying models as APIs and stress-testing them with ensembles.
And finally, deep learning on-chain chaos, NFTs, memecoins, volatility, all of it.
This isn’t your “textbook ML”.
It’s experimental.
It’s reckless and it might fail.
But that’s the point.
Because the future of ML isn’t in perfect datasets and neat Jupyter notebooks. It’s in messy, adversarial, real-world systems like blockchains. Where fraud, speculation, and innovation collide every second.
I’ll write it all here. The wins, the bugs, the models that overfit in seconds, the insights that might actually matter.
If you’re tired of the polished, predictable ML content that clogs your feed, I am personally welcoming you.
This Substack is for people who’d rather build a scam detector than a spam filter. Who care more about predicting rug pulls than predicting tomorrow’s weather. Who want to learn in public, not hide behind “case studies.”
I don’t promise clean conclusions. I promise experiments.
And maybe, by the end of this, deployed models that actually survives the madness of crypto.
Learning machines. Reading chains.
Let’s see how far this goes.
Community: https://x.com/AnalyticSages
Mentor: https://x.com/apostleoffin
This isn’t just my journey. It’s ours. Join in, experiment, and let’s learn in public together.