The Strategy That Lost Money for 2 Years (Then Made 19% Annually)
- Fabio Capela
- Systematic trading , Trading psychology , Portfolio management , Risk management , Personal journey , Algorithmic trading , Market regimes , Strategy development
- August 23, 2025
Table of Contents
Most trading success stories skip the embarrassing part. Here’s what actually happened.
Year 1: The Confident Idiot Phase
Started in 2016 with $50k and a “foolproof” mean reversion strategy. The backtest showed 67% annual returns with a Sharpe of 2.8. I was already calculating yacht prices.
Three months later: down 18%.
The strategy was buying stocks that gapped down at the open. Except my backtest was using the pre-gap price to enter positions. I was literally buying at prices that never existed. Classic look-ahead bias, but I was too arrogant to see it.
Year 2: The Overengineering Phase
Fixed the obvious bugs, added 47 indicators, 12 risk filters, and machine learning. Because if simple doesn’t work, complex must be better, right?
End of year 2: up 3.2%.
Beat inflation. Lost to literally everything else. My wife’s “just buy SPY” portfolio destroyed me. The worst part? I couldn’t explain why my strategy wasn’t working. Too many moving parts, no idea which ones were broken.
Year 3: The Breakthrough
Threw everything away. Started with one simple observation: momentum works, but it’s unstable. Some months it prints money, others it bleeds.
The insight: don’t predict when momentum will work. Detect when it’s working.
Built a regime detection system:
- Momentum regime: ride trends aggressively
- Mean reversion regime: fade extremes
- Choppy regime: reduce position sizes
- Crisis regime: go to cash
No predictions. Just reactions to current market conditions.
Years 4-8: The Boring Success Phase
Annual returns: 23%, 31%, -4%, 42%, 17%, 21%
Average: 19% CAGR with 12% max drawdown.
The strategy is embarrassingly simple now:
- Measure market regime using 5 indicators
- Apply the appropriate sub-strategy
- Size positions based on regime confidence
- Rebalance weekly, not daily
No AI. No complex math. Just systematic rules applied consistently.
What Actually Matters
After 8 years and probably $100k in losses from mistakes, here’s what actually matters:
Transaction costs matter more than alpha. A strategy making 50 bps per trade dies at 45 bps costs. Most of my early strategies were just feeding brokers.
Simple beats complex. My current strategy has 20 lines of core logic. My failed ML strategy had 2,000. Guess which one I actually understand?
Regime awareness beats prediction. Stop trying to predict where markets will go. Start recognizing where they are.
Live trading beats backtesting. Paper trade for 3 months minimum. Your backtest is lying to you, guaranteed.
The Uncomfortable Truth
Most systematic traders quit during years 1-2. They lose money, can’t figure out why, and assume it doesn’t work.
It does work. But the learning curve is expensive. Budget for it. Expect to lose money initially. Start with capital you can afford to lose entirely.
The traders making consistent money aren’t smarter. They just survived long enough to debug their assumptions.
Your Move
Stop optimizing your backtest. Start trading with small money. You’ll learn more from one month of live trading than one year of backtesting.
The market is the only teacher that matters. And tuition is expensive.
But it’s worth it when your systematic strategy finally clicks and starts compounding at 19% annually while you sleep.
Ready to shortcut the learning curve? I’ve packaged everything into TheSimplePortfolio - the strategy that survived 8 years of real markets.
Currently managing systematic strategies at TheSimplePortfolio. 19% CAGR since 2016. Still making mistakes, but smaller ones.