The Build Lab
One complete algorithm build, shown in full — the idea, the build conversation, a real bug, an honest out-of-sample result — then the same skeleton, empty, with fields to fill in for your own idea.
- A real opening-range strategy built end-to-end on a real stack
- The bug chapter — a timezone error that made the backtest lie, caught and fixed
- The honest numbers: in-sample 2.41 vs out-of-sample 0.86, and what that teaches
- Your turn: the same skeleton as a worksheet that saves privately on your device
DAY
Not a theory of building.
One real build — then yours.
The Complete Build
Idea → rule → code → first look, on a real stack: Databento data, Python research, NinjaTrader 8 with a CrossTrade bridge.
The Actual Conversation
The real back-and-forth with Claude — including the messages that caught problems, not just the ones that worked.
The Bug Chapter
A genuine timezone bug that made the backtest beautiful and wrong — how it was spotted, and the lesson that generalizes.
The Honest Result
In-sample 2.41 became out-of-sample 0.86. Most products would hide that. Here it's the most important chapter.
Your Turn
The same skeleton, empty — pattern, precise rule, data assumptions, IS/OOS split, devil's-advocate pass — saved privately as you type.
Two Editions
The interactive HTML lab plus a clean PDF edition — the worksheet prints with blank fields for handwriting.
Watch one build honestly.
Then run your own.
The real conversation
The actual build, step by step — what was asked, what came back, and what got questioned.
The backtest that lied
A timezone error made every trade enter an hour early. The chapter that teaches you to be suspicious.
The worksheet
The same skeleton, empty — your idea, your rule, your split — saved privately on your device as you type.
Actual screens from the lab · your worksheet answers stay private on your own device.
Everything you might be wondering.
The honest answer — and the whole point: it survived its out-of-sample test with a weaker result than in-sample (0.86 vs 2.41), which is exactly what a real, marginal first build looks like. This isn't a strategy to copy; it's the method, demonstrated truthfully.
No. The build conversation is shown in full, in plain English, and your worksheet asks for decisions, not code. If you can describe a pattern, you can run your idea through it.
The Starter Kit teaches the concepts; the Build Lab shows one complete, honest application of them. Either order works — together they're the course and the case study.
One download with both editions: the interactive HTML lab (the worksheet saves privately on your device, works offline) and a clean PDF edition — print it and fill the worksheet in by hand.
You're covered by our 14-day money-back guarantee. Email us within 14 days and we'll refund you in full — no awkward questions.
Take the whole lab, risk-free.
Read the whole build, run your own idea through the worksheet. If seeing one honest build doesn't change how you approach yours, email us within 14 days for a full, friendly refund.
See one done honestly.
Then build yours.
The fastest way to learn the method is watching it applied — bug, humbling result and all. Instant access, yours forever, 14-day guarantee.