Bring us your idea.
We run it.
How a request runs
- You write the rule
- In plain words: the condition, what counts as an entry, where it is wrong. If a rule cannot be written down, it cannot be tested — and that is usually the first useful finding.
- It gets pre-registered
- The rule is fixed before the run, not after seeing the result. This is the whole reason the answer means anything; a rule adjusted until it looks good is a description of the past, not a test.
- It runs on the archive
- Minute data with the bid/ask spread, not daily closes — so the cost of actually taking the trade is in the result rather than left out of it.
- You get the run back
- What happened, how many times it happened, and what would have had to be true for it to work. Not a verdict on you.
Why we built it this way
- Frequency, not prophecy
- The useful question is almost never “will this work” but “how often has this happened, and how often did it pay”. A number with the sample size next to it answers that. A conviction does not.
- No is a result
- Most ideas do not survive contact with the archive. You are paying for the answer, not for a yes — and finding out that an idea you were about to risk money on does not hold up is the cheapest it will ever be.
- Findings are published
- Completed runs go on the members' shelf, so the Lab compounds. One member's question becomes something everyone can read.
- It is not tied to gold forever
- The engine is ours. As the archive widens, the questions you can ask widen with it.
The terms
- Included
- One request per month, on every plan. It does not roll over. If the request cannot be
pre-registered we will come back and say why rather than guess at what you meant. - A second in the same month
- $100.
- What we will not do
- Tell you whether to take a trade, manage money, or run an idea we cannot state as a rule.
A backtest describes what happened in a fixed archive under a fixed rule. It is not a prediction, and nothing that came out of a run is a recommendation to trade.