Six prompts below turn Claude or Codex into an honest strategy judge on your real TradingView chart — full history, per-year windows, out-of-sample splits. The seventh audits where your own trading bleeds. Copy freely; the prompts are the free half of the lab.
NEW HERE? · 2-MINUTE SETUP
Three commands, then any prompt works.
You need Node 18+, TradingView Desktop (a free account is fine), and Claude Code or Codex. The doctor checks everything itself.
PowerShell / terminal
npm i -g proea-studio
proea-studio install
proea-studio doctor
#1
THE KILL-GATE
Paste it. I dare your strategy to survive.
You get A verdict with every gate that failed — and the one honest next move.
paste into Claude Code / Codex
Build this idea as a Pine v6 strategy: [describe your idea in plain English]. Compile-check it. Then backtest it honestly on [OANDA:XAUUSD] [1h] — full history, per-year windows, and an in-sample/out-of-sample split — on my real TradingView Strategy Tester. Give me the verdict with every gate that failed, and the ONE mutation worth trying next. If nothing honest is worth trying, say so plainly. End with a summary card: VERDICT / trades / OOS PF / max DD / next move.
Free bridge chart control + Pine editing · With Studio full auto-backtest loop + kill-gate verdict
Prompt 1 in the setup video — one paste, ~78 seconds to verdict on a real chart.
#2
THE NEXT MOVE
Your strategy just died. Ask what's legal to try next.
You get The one mutation worth testing — or “a different family, not a tweak” said to your face.
paste into Claude Code / Codex
My strategy just failed the kill-gates. Here is the verdict card: [paste the verdict]. Tell me what is legal to try next under an honest research playbook. If the answer is a different strategy family — not a parameter tweak — say so plainly, scaffold that candidate, and run it through the same honest multi-window backtest. End with the new summary card: VERDICT / trades / OOS PF / max DD / next move.
Free bridge chart control + Pine editing · With Studio playbook of legal mutations + full re-test loop
Prompt 2 in the setup video — the loop: build, test honestly, kill, move on.
#3
THE SECOND OPINION
You already have a strategy. No encouragement — just the verdict.
You get An audit + honest multi-window backtest of YOUR existing Pine — no encouragement, just the verdict.
paste into Claude Code / Codex
Here is my existing Pine strategy: [paste the code or file path]. Audit it, compile-check it, then run it through the same honest multi-window backtest and kill-gates — full history, per-year windows, in-sample/out-of-sample split — on my real TradingView Strategy Tester. I want the verdict a fund would give me, not encouragement. End with a summary card: VERDICT / trades / OOS PF / max DD / next move.
Free bridge chart control + Pine editing · With Studio full auto-backtest loop + kill-gate verdict
#4
THE GURU TEST
That YouTube strategy? Test it faithfully. No strawmen, no mercy.
You get Whether the guru's rules survive real out-of-sample windows — and which single year did all the work.
paste into Claude Code / Codex
This strategy is from a video: [paste the rules exactly as taught]. Build it faithfully in Pine v6 — no improvements, no fixes — and test whether it survives real out-of-sample windows on my own TradingView Strategy Tester. Don't be polite. If it only works in one year, name the year and explain why that matters. End with a summary card: VERDICT / years survived / OOS PF / the one honest conclusion.
Free bridge chart control + Pine editing · With Studio faithful build + multi-window verdict
#5
THE GRAVEYARD
Five ideas walk in. A leaderboard walks out — most of it dead.
You get KILLED / WEAK / PASS on each idea, one line on why each died, one candidate worth a forward test.
paste into Claude Code / Codex
Take these five ideas: [list them in plain English]. Build, compile-check, and kill-gate test each one on [symbol] [timeframe] on my real TradingView Strategy Tester. Give me a leaderboard: KILLED / WEAK / PASS for every idea, one line on why each died, and which single candidate deserves a forward test. The leaderboard IS the summary card — end with it.
Free bridge chart control + Pine editing · With Studio batched auto-backtest loop + leaderboard verdicts
#6
THE LEAK FINDER
Find where YOUR money is actually leaking. No motivational talk.
You get “You lose the most in [session/day/behavior]; cutting it = this much net” — from your own history.
paste into Claude Code / Codex
Import my trade history from [file]. Find where I'm actually bleeding: by session, by day of week, by symbol, and by behavior after a loss. Then show me the ONE change with the biggest net improvement, with the exact numbers before and after. No motivational talk — just the leak and the patch. End with a summary card: THE LEAK / what it costs / THE PATCH / net improvement.
Needs Studio trade-history import + leak report (TradeLens engine)
#7
THE OVERFIT TRAP
Watch a beautiful backtest get manufactured — then caught.
You get The amazing full-history number, then what actually survives honest gates. The difference is the lesson.
paste into Claude Code / Codex
Take this strategy: [paste or describe it]. First, deliberately overfit it — tune parameters until the full-history backtest looks amazing on my TradingView Strategy Tester, and show me that number. THEN run the same strategy through the honest in-sample/out-of-sample kill-gates and show me what survives. Teach me the difference on my own chart. End with both numbers side by side: OVERFIT full-history vs HONEST out-of-sample.
Free bridge chart control + Pine editing · With Studio both runs automated + honest gate verdict
A PASS IS A HYPOTHESIS, NOT A PROMISE · WINDOWS WITH NO DATA ARE SKIPPED, NEVER FAKED · RESEARCH TOOLING, NOT FINANCIAL ADVICE · BACKTESTS ≠ FUTURE RESULTS