What you learn
- How the chart really moves: why direction is unpredictable and agitation is not, how noise is measured, what volume says and does not say. The mechanics of price, not recipes.
- The validation lab: the five gates, how to declare the rules before the numbers, how to spot look-ahead (when a test «knows» the future without you noticing), why the bot is not the backtest.
- The cost/stop law: the filter that tells you in advance on which instruments and timeframes a strategy can live, and on which it is dead before birth.
- How to read a macro release: regimes, tracks, why the surprise moves and then dies, what can and cannot be predicted.
- The mistakes I paid for: survivorship bias, rolling windows that look clever, signals that work only if you edit them afterwards. They cost less when told by someone who made them.
How it works
One-to-one online lessons, or small groups (four people at most). Each meeting has a task to do before and an exercise to do after, on your data and with your code if you have any. No slides to read: we work.
Who it is for
For people who know (or want to learn) some Python and want to stop trusting other people's backtests. For people who have understood that the problem is not finding a strategy, but knowing whether it is real.
Who it is not for
For people who want «the setup that works». I do not teach it because I have not found it, and the numbers say none of the famous ones passes the gates. I teach how to check, which is worth more.
The site's free guides are the course programme, in short form. If they are enough for you, do not buy anything.