Buy ten prop accounts and run your strategy on all of them. This page tells you how many pass, how many blow, what you clear, and the name of the rule that kills the rest.
Bring a backtest, a script, or just an idea. You get the count and the one thing to change. Everything is computed in this browser. Nothing leaves your device unless you press Submit.
Pick a shape, set the numbers, and it runs on 18.5 years of NQ embedded in this page — no upload, no server, no code. Then it flows straight into the prop simulator below, exactly as a dropped CSV would. Every variant you try is counted, and the noise floor for that many tries is shown beside your result.
These stack on top of whatever shape you picked. The event dates are the
verified list from news_calendar.csv — 605 rows of
FOMC, CPI and NFP days.
Paste a Pine Script or NinjaScript strategy. We do not run it and we do not translate it — nothing here executes your code, and there is no interpreter in this page to execute it with. We read the parts we recognise, map them onto the builder, and show you that reading line by line beside your own source, with everything we could not read quoted back verbatim. If we could not read enough of it we say so and refuse to test it, which is the right answer rather than an error. Your script text is included in a submission if you choose to send one — including one we refused.
All of it. Parsing, statistics, the backtester, the simulations, the firm ranking, the charts. The page loads nothing from anywhere — no fonts, no scripts, no images, no telemetry — and runs offline. The only request it can ever make is the one you trigger by pressing Submit. There is no AI model involved at any point; every number comes from deterministic arithmetic you can re-derive.
4,614 NQ RTH sessions, 2008 to 2026, at daily, 60-, 30- and 15-minute
resolution, are inside this file — gzipped, base64'd, and unpacked by the browser's own
decompressor. Nothing is fetched. They are generated by
proptester/build_data.py, which round-trips every decoded bar against the
1-minute source before it writes, and the page's self-test re-checks that the coarse bars are
exactly their fine constituents aggregated. There is no eval() and no code
upload in the builder: strategies are templates and parameters, which is the only way a
tool like this can be safe to hand to strangers.
Only if you press Save: your saved setups (parameters and, so you can resume, the trade text you loaded), your custom firm presets, and your remembered column mappings live in this browser's localStorage. Delete removes them.
Your parsed trade sequence, the rules you tested against, and every computed number — enough to reproduce the analysis independently. Plus whatever name, contact and description you choose to type. Nothing leaves your device unless you press Submit. Copy and download hand the file to you and send nothing; a human reads what arrives, and we do not publish or redistribute it.
Maths ported from rotation_econ.py / ROTATION_ECONOMICS.md
(account engine, trailing conventions, per-trade risk cap, fee clock, E[$/fee], correlated
rotation), regime_lens.py / BOOK_TIERS.md (tripwire statuses,
DECAY_FLOOR 0.80), CURRENT_REGIME_REAUDIT.md §2 (the noise floor),
EDGEFUL_ORB_TEST.md §4 (expected-maximum multiplicity),
SMALL_SL_SWEEP.md (what a tighter stop costs in winners) and the daily-profit-cap
payout mechanic from ZENOMTRADER_COURSE.md §2.4. Every ported function names its
source in the code.