About this site and the testing framework — why I built it
FX Backtest Diary records a home-grown framework for telling whether an automated trading (EA) idea is a real edge or just luck. Here's the thinking behind it and how it was built.
This site, “FX Backtest Diary,” is where I test automated FX trading (EA) ideas one by one and record the results. But it isn’t just a “did it make money or not” log. The real goal is to tell, as strictly as possible, whether an idea is a genuine edge or just happened to fit past data.
Why build a framework from scratch?
At first I figured an off-the-shelf backtester would be fine. But the more EAs I tested, the more I ran into these traps:
- Mistaking a “lucky fit” to historical data for a real edge (this is called overfitting)
- A strategy that looks fine on the daily chart but actually fails when you zoom into the intraday moves
- The data itself being broken, so the strategy only “won” thanks to those bad values
To close these traps one by one, the best approach was to build the testing machinery myself and understand every part of it. So I wrote a Python engine called btengine.
The guiding idea
The most important job of this framework isn’t really “finding winning strategies” — it’s being able to reject the ones that don’t work. It sounds dull, but it’s what protects your capital the most.
Think of it as a metal detector. When you search a beach (a huge pile of ideas) for real gold coins, without an accurate detector that ignores junk, you waste time and money digging up garbage. So first I carefully built a detector that doesn’t react to fake edges.
Concretely, it has these gates (each covered in its own article):
- Test on clean data (broken bars removed)
- Walk-forward / forward testing, using no future data at all
- Reconstruct intraday risk on the 1-minute (M1) chart
- Estimate the probability of passing with Monte Carlo
- Manage risk realistically with a portfolio engine that runs all symbols on one shared account
To be honest
There’s no flashy holy grail here. In fact, there are more “I tried it and it didn’t work” articles than success stories. But that’s exactly the value of this diary: honestly testing many ideas and keeping only what’s real — published as I go.
This site is a record of research and verification, not investment advice. Past results do not guarantee future performance.