4Research
We trade on our own research. We don't publish what we trade or how: a strategy tends to stop working once everyone knows about it, and posting positions would turn a research site into a tip sheet. What we do publish is what we learn, as dated notes with the data, the method and the limits, including the results that didn't hold up.
4.1How we work
We built this archive because we needed a record of the options and equity markets we could trust, with every method written down. Our research runs on the same data and follows four rules.
Write it down first
We decide what we're testing, and how we'll judge it, before we look at the results. Otherwise it's too easy to pick a test that agrees with you.
Test it forward
A backtest is only a starting point. An idea has to work on live data it has never seen before we say anything about it.
Publish what fails
If a result doesn't hold up, we publish that too. Leaving out the misses makes any body of research look better than it really is.
Never publish a position
No holdings, no trade ideas, no recommendations, not even after the fact. Our notes describe the market, never our trades.
4.2Findings
Newest first. The numbers in each note are fixed as of its date. When new evidence comes in, it goes in a new note.
- Four tries at beating the market by picking smaller companies We built four rule-based ways of picking US small and mid-sized stocks and tested each on years of history, failed companies included. The best one picked better stocks than the pool it chose from, but by too little to rule out luck, and none of them beat simply owning the S&P 500.
- Moon phases, eclipses and planets against 55 years of stock returns We ran 168 tests of celestial and calendar effects and corrected for running so many. Three passed the strictest correction: the Monday effect, which is real, and two long astronomical cycles, which are artefacts of a short sample.