I've been backtesting strategies for 6 years. Here's why yours is probably broken.

**I've been backtesting strategies for 6 years. Here's why yours is probably broken.**

Not trying to be a jerk, but I see the same mistakes over and over (because I made all of them first).

**Your data is lying by omission**

Backtesting the S&P 500? Cool. You know what's not in that data? All the companies that went bankrupt and got kicked out. You're only testing on the winners. It's like going to a gym and asking the jacked guys for fitness advice while ignoring everyone who quit - obviously the results look good.

Survivorship bias-free data costs money but it's worth it if you're serious.

**You kept tweaking until it worked, didn't you?**

Been there. "Hmm, 60% win rate. What if I add this filter? Oh now it's 70%. What about this indicator? 80%!"

That's not finding a strategy. That's overfitting to random noise. I've built strategies that had insane backtests and lost money week one of paper trading.

Rule of thumb I learned the hard way: if you can't explain why your edge exists in plain english without mentioning indicators, it's probably not real.

**The spread will eat you alive**

Your backtest assumes you bought at the exact close price. In reality: spreads, slippage, commissions, bad fills because you sneezed during market open.

I had a strategy showing 0.4% average profit per trade. Factored in realistic costs and suddenly it was like 0.05%. Not exactly worth the stress.

**Out-of-sample or it doesn't count**

Optimize on 2018-2023, then test on 2024-2025 data you never touched. If it falls apart, you just curve-fitted. This is non-negotiable.

Even better: walk-forward testing where you keep rolling the optimization window. It's tedious but it actually tells you if something works.

**What I do now:**

* Test across different market regimes (bull, bear, choppy sideways hell)
* Use recent data - markets evolve, strategies from 2015 don't always work in 2026
* Paper trade for at least a month before going live
* Keep a brutally honest journal comparing backtest expectations vs reality

A good backtest means your strategy *might* work. Not that it *will* work. I've learned to treat backtests as "this is worth paper trading" not "time to risk my savings."

What's the worst backtest-to-reality gap you've experienced? Mine was like 80% win rate → 45% when live. Still hurts to think about.