Idea →testable system.Honestly.

Three steps. No coding.

• 1 · Drop a fileOpen Claude (or ChatGPT / Cursor / Gemini) and give it START_HERE_AI.md. That one file teaches the AI the whole kit.
• 2 · Say your idea“Buy pullbacks to the 20-EMA in an uptrend.” Vague is fine — it interviews you and fills the gaps you'd skip.
• 3 · Get a strategyIt writes a real, compiling TradingView strategy you can backtest today — entries, exits, stops and risk, all in.

The whole job — copy, paste, done.

# Open Claude / ChatGPT / Cursor / Gemini, then:
⤵  drop in:  START_HERE_AI.md

"Read START_HERE_AI.md and pine-strategy-foundation/SKILL.md, then
 elicit a blueprint for a trend-pullback idea on NAS100 1h, and
 build a Pine v6 strategy from it I can backtest."

→  it asks a few questions, then writes a strategy you can test.

Every line of code traces back to your idea

Open the source and each block carries a plain comment tying it to your plan — your market becomes the costs, your trend filter becomes the gate, your entry becomes the trigger. And the header says it out loud: untested — a compiling strategy is not a profitable one.

What it writes first: a plan you can read

# NAS100 Trend Pullback — the plan, in plain language
name:    "NAS100 Trend Pullback"
idea:    "Buy pullbacks to the 20-EMA in a confirmed uptrend"
allowed: 1D close > 200-EMA, higher-highs / higher-lows intact
entry:   price pulls back to the 1h 20-EMA, then a bull bar closes
stop:    below the pullback swing low
target:  1.5R, then the prior swing high
risk:    0.5% per trade

# the honest part, written straight into the file:
status:  untested
verdict: "NOT a profitable system until validation proves otherwise."

From a simple dip-buy to what a hedge fund trades

Five of the deepest blueprints in the box — each a real economic idea with a strict built-in fail-test, not a vibe:

Proof the depth is real: 33 years, three crashes

The vol-targeting blueprint above, built into Pine and tested on SPY across 33 years (1993–2026) — through the 2000, 2008 and 2020 crashes:

1 — The Regime Router: a clean KILL

Our deepest build — a router that switches capital between trend, mean-reversion and breakout legs by market regime. We shipped it with toggles so you run its own fail-test live, on your own chart.

The test asks one thing: does the fancy ensemble beat its best single leg? Live on EURUSD the answer is 0.717 vs 0.673 — both under 1.0. No edge. A clean KILL — and we screenshotted the loss instead of burying it.

2 — SMC Order Block: it survives, and we still say “not yet”

The order block every retail trader has drawn by hand — here defined as a mechanical rule a computer can actually test, no hindsight boxes.

Flip one toggle and the same machinery buys any dip instead of the order block. The order block won — PF 1.414 vs 0.96 — so the box is doing real work, not decoration. And we still call it inconclusive (only 30 trades). Winning the test and saying “not yet” is the entire discipline.

3 — ICT Silver Bullet: you can't even test it

The internet sells the 10–11 ET “Silver Bullet” as a daily money-printer. Built strictly and mechanically — no hindsight — here is what honest looks like:

Run honestly, the strict Silver Bullet qualified once in 7.5 weeks — a roughly once-every-two-months event. You can't validate one trade, and that is the finding: every Silver Bullet win-rate you've ever seen is hindsight. A vendor who shows you a one-trade backtest and says “don't trust this” just earned more credibility than every Silver Bullet course combined.

Kill it · can't-confirm-it · can't-even-test-it

Three systems, three honest failure modes. That is what “we sell the test, not the dream” actually looks like — and it's why the next thing a serious trader reaches for is the same kit, not a refund.

So why pay, when a free chatbot can sketch a script?

Because a blank prompt gives you a toy. Foundation refuses the three things that make a free script dangerous with real money:

What you can ask it to do

Drop in START_HERE_AI.md and ask, in plain English:

• “Turn my idea into a plan”It walks you through a guided build, recommends a sensible default at each step, and forces the honest fields you'd skip.
• “Interview the system in my head”Trade by feel? It draws the system out of you and writes it down — including the “when is it wrong?” you never said out loud.
• “Build it into TradingView Pine”It turns your plan into a faithful, compiling, non-repainting strategy you can backtest — every block traced to your plan.
• “Adapt a worked example”Start from one of 18 ready blueprints and bend it to your instrument and idea.

What's in the box

• START_HERE_AI.mdThe one file you drop into your AI. It teaches the AI the whole kit — what to read, the rules it can't break, a ready first prompt.
• The skill + the flowsThe guided build, the interview, and the idea → Pine bridge your AI follows — plus a checker that hard-fails any “guaranteed / win-rate” language.
• 18 worked blueprints + 6 built strategiesFrom beginner trend-pullback to institutional stat-arb — every one shipped honestly marked “untested.”
• Plain-English guidesQuickstart, copy-paste prompts and extension ideas — written for humans, not just the AI.

The moment you pay

• Instant accessPay by card, Google Pay or PromptPay and the kit is yours right away — one ZIP, link emailed too.
• Yours forever + free updatesNot a rental. No expiry, no seat check; every future version is free — email support@pulltrade.app.
• 7-day refund — before you downloadChanged your mind and haven't downloaded? Full refund within 7 days, no questions. Once you download, the window closes — files can't be returned.