SuperQuant

Applying 20 years of software engineering experience to the financial markets. Sharing my journey to becoming a systematic trader with code & lessons.

[SuperQuant @0xSuperQuant]

Hi, I'm 0xSuperQuant.

I've spent 20 years as a software engineer building complex, large-scale systems.

Now, I'm publicly documenting my entire journey to becoming a retail systematic trader. 🧵👇

[SuperQuant @0xSuperQuant]

@quant_xbt With 20 years of software engineering experience, and just starting this quant journey, I find this an incredibly valuable reminder.

The temptation to jump straight to complex ML models is huge. I'm forcing myself to go deep on the basics of market structure first. Thanks.

[SuperQuant @0xSuperQuant]

@PythonPr Solid roadmap for the 'ML' fundamentals! For the 'Engineer' part, I'd add a crucial step for MLOps — deployment (Docker, Kubernetes), model monitoring, versioning (like DVC/MLflow), and CI/CD pipelines.

[SuperQuant @0xSuperQuant]

@systematicls Great point. The mathematical identity in the code is clean, but the underlying execution assumptions are not symmetrical.

'I can buy at the close' (Signal A)
—
'I can sell at the close' (Signal -A)

...are two totally different propositions in a real market.

[SuperQuant @0xSuperQuant]

@imtommitchell 100% this. Thanks for sharing.

As a 20-yr dev, I can't stress this enough. Learning to programmatically access and manage data via cloud APIs is the real skill. Analyzing a static CSV is step one; building a system that pulls new data automatically is the whole game.

[SuperQuant @0xSuperQuant]

My goal is to learn in public, stay accountable, and connect with others on the same path.

Follow along!

[SuperQuant @0xSuperQuant]

Hi, I'm 0xSuperQuant.

I've spent 20 years as a software engineer building complex, large-scale systems.

Now, I'm publicly documenting my entire journey to becoming a retail systematic trader. 🧵👇

[SuperQuant @0xSuperQuant]

What to expect from my account:

1. Daily learnings in Python, Pandas, Probability, Stats, & ML.

2. My research into quant strategies & building my first backtests.

3. Building my blog https://t.co/aExT9OwXZ3 in public.

4. All from a 20-yr software dev's perspective.

[SuperQuant @0xSuperQuant]

@exec_sum This is it. The "art" of discretionary M&A and IPO modeling is officially being codified into a training dataset.

They're paying $150/hr to systematize the entire traditional finance playbook.

The bankers are the training data for their own replacement.

[SuperQuant @0xSuperQuant]

@TheFractalyst @DarwinexZero That last line is the entire mental shift. "Don't try to build models, try to break it."

This is the engineer's approach vs. the gambler's.

You're not looking for a perfect green curve; you're looking for a robust edge that survives all your attempts to kill it. Great thread.

[SuperQuant @0xSuperQuant]

@shivst3r 100%. The "builder" creates a scalable asset.

A systematic quant is building an asset (the system). A discretionary trader is working a high-stress job.

[SuperQuant @0xSuperQuant]

@GoshawkTrades Because they are masters at the single most important skill in trading: finding a faint, true signal in a universe of noise.

Astronomy is just the ultimate signal-processing problem. Markets are no different.

[SuperQuant @0xSuperQuant]

@noalphadecay That last part is the secret, isn't it? The R&D process is a game of intellectual curiosity, which is the 'fun' part. Live trading is a game of emotional discipline.

Many of us are builders and researchers at heart.

[SuperQuant @0xSuperQuant]

@imtommitchell Spot on list for a solid foundation. For those of us in quantitative finance, I'd add statsmodels (for time series analysis/regressions) and Plotly (for interactive charting) as essentials.

[SuperQuant @0xSuperQuant]

@ezekiel_aleke Awesome share! Data cleaning is 80% of the job, so it's always great to see clear, simple explanations. Was there a specific part of this that you found most helpful?

[SuperQuant @0xSuperQuant]

@MoonDevOnYT Very interesting! What are you using the Kalman filter for specifically? Signal generation, smoothing price data, or something else?

[SuperQuant @0xSuperQuant]

@Globalflows 100% This is the core truth. It’s not the asset, it’s the execution.

That last part is key: surround yourself with high-integrity, high-drive people. You can't win alone, and you definitely can't win with the wrong team.

[SuperQuant @0xSuperQuant]

@puregamma A devastating, system-level failure. Sympathies to the trader.

This is a brutal lesson in counterparty risk. The failure wasn't the strategy, but the exchange's collateral pricing model.

A quant's model must always account for the platform's own risk model failing.

[SuperQuant @0xSuperQuant]

@spicyofc This is the foundation. For systematic trading, risk management is the system.

It's not just a stop-loss. It's the position sizing, correlation, and max drawdown rules all defined before a trade is ever placed. Great guide.

[SuperQuant @0xSuperQuant]

@dotnethero @brodriguesco Great question. They're complementary tools.

- cuDF: Unbeatable for GPU pipelines when data fits in VRAM.
- Polars: The CPU king for general ETL, backtesting, or any dataset > VRAM.

For serious quant work, you'll end up using both.

[SuperQuant @0xSuperQuant]

@brodriguesco Haha, 'micro-econometrics flashbacks' is the ultimate compliment for a critical review.

Since you're stress-testing the models, what are your favorite AI agents out there right now, ranked?