In January 2025, I published a mean reversion strategy with a free backtester.

11 months later, here are the out of sample results:

- +30.4% return (vs +24.4% Nasdaq)
- -10.2% max drawdown (vs -22.8% Nasdaq)
- 72% win rate
- 83 trades

Here's the exact system: The idea is simple:

Buy stocks that drop sharply - but only if they're still in an uptrend.

Short-term panic creates opportunity.
Long-term trend provides the safety net.

This is mean reversion 101.

Leveraged ETFs can be a great source of alpha — if you treat them as mean-reverting wrappers 📉📈 rather than long-term investments.

Here’s a simple $TQQQ / $QQQ swing strategy that beat buy & hold with only 24% average capital use 👇 Leveraged ETFs (like $TQQQ, 3× $QQQ) don’t perfectly track their target.

⚡ Daily compounding
⚡ Volatility drag
⚡ Rebalancing frictions

Result: “Excess decay” appears when TQQQ underperforms its ideal 3× of QQQ over short horizons. That excess tends to mean revert.

1/6 The Power of "Boring": How Low Volatility Supercharges Mean Reversion

Mean reversion is a popular concept, but it doesn't work the same on all stocks.

I backtested the exact same system on two different baskets of stocks. The difference in performance is stunning. Here's the one factor that changes everything. 🧵👇 2/6
The system is the same - a long-only, mean-reversion strategy on S&P 500 stocks:

🟩 Green Line = Prioritizes LOW-volatility stocks
🟥 Red Line = Prioritizes HIGH-volatility stocks

The chart speaks for itself. Low volatility is smoother, safer, and vastly more profitable.

Getting more and more tradable systems designed by LLMs.
This is what I’ve learned actually works for me.
👇 Step-by-step breakdown of the process that turns AI ideas into real systems I can paper trade. 1/
I start in ChatGPT (o3 model).
I use a custom bio where I describe myself as an experienced trader—this primes it into “expert mode.”
Then I share a picture from a real trading session and describe the context.

In this case: intraday breakout from a volatility “squeeze.”
I explain the type of data I have + what kind of realistic performance I expect (after slippage & fees).
Then I ask GPT to brainstorm and rank strategies by:
- difficulty to implement
- likelihood of working live

1/5 Let's talk about something crucial but often overlooked in deep research, especially in systematic trading: Note-Taking. 📝
When you're diving into years of complex research, building strategies, and testing hypotheses, how you organize your knowledge is paramount. Without structure, you'll inevitably find yourself going in circles, re-discovering old insights, or losing valuable connections.
It's a huge, hidden productivity drain. You need more than just scattered files; you need a second brain. 2/5 My personal game-changer?
The Zettelkasten Method, implemented using Obsidian (@obsdmd). 🧠 Obsidian is free and works directly with plain Markdown (.md) files stored locally on your computer. This means:
✅ You truly OWN your data.
✅ No internet required – full offline access.
✅ Notes are simple text files, readable by anything, forever.
This isn't just storage; it's about connecting notes into a powerful knowledge network – a 'second brain'. Optional sync keeps it available across devices if needed.

(1/7) Fixed Size vs. Volatility Targeting in Momentum Rotation: A Deep Dive

A common question arose regarding my NDX rotational strategy: Why use 20% volatility targeting when fixed sizing shows higher net profit in backtests (like the one shown)? A valid point at first glance! The fixed-size backtest ($373k profit) clearly outperforms vol targeting ($218k profit) on an absolute basis without compounding. (2/7) The Allure & Danger of Raw Profit
It's tempting to chase the highest P&L figure. Decades ago, many focused solely on this. However, experience teaches a harsh lesson: absolute performance isn't the only metric that matters, and often, it's not the most important. Survival and consistency are key.