Great backtest. How many did you try?
QuantSuite counts every version you test and shows what is left once luck is taken out.
A local research terminal for Windows. Trading is paper-only in v1: simulated fills, no live order path in the build.
Backtest Facts
Survives up to N=8. The bar is 95% confidence it beats luck. Raw SR 1.80 | PSR(0) 99.9% | DSR(N=10) 93.7%.
Same backtest. Three different answers.
The only thing that changes is how many versions were tried first. Most backtests never mention it. QuantSuite keeps the count for you.
Tried 1 version
Nothing to discount. It clears the bar.
Tried 10 versions
Luck accounts for 0.91 of the 1.80. Just under the bar.
Tried 100 versions
Luck accounts for 1.47 of the 1.80. Not much is left.
We put a label on ourselves, too.
Every model our weekly retrain produces has to clear the same deflated bar before it replaces the current one. Most do not, and the count is below, read from the file your install ships with.
Fifty-two candidates looked good enough to train and were not promoted. The most recent, 2026-08-26, is the whole argument in one row: a raw PSR of 0.82 — a number that looks healthy on its own — deflates to 0.30 once its N=4 trials are counted, with DSR 0.44 over 116,554 out-of-sample observations. It beat the incumbent champion on the same evaluation window and was refused anyway, because neither the absolute floor nor the trial-corrected bar cleared. We ship no proven edge on any bot, and say so in the product rather than in a footnote. A gate that passes everything is decoration; this is what one that works costs — and these are the rows in the registry your copy installs with, so you can open Neural Analysis and check this page against your own machine.
Every result in QuantSuite carries its label.
Real captures of the terminal on our machine, each cropped to one panel and otherwise unedited. The full, uncropped set is on the tour.
Every number shows its work.
Backtests report the raw Sharpe with PSR, DSR and the N behind them, like SR 0.14 | PSR 100% | DSR 29% (N=470). A figure that cannot be computed soundly is held back, with the reason printed beside it.
It tells you when it does not know.
Our own models have not cleared the deflated bar, so the confidence gauge stays empty. Your results get the same treatment.
And it is a whole trading terminal.
Charts, screeners, options, flow, filings, portfolio risk and paper-trading bots, each panel stamped with its data source and time basis.
What is inside.
One local terminal, served by your machine to your browser.
Markets
Dashboard, screener, charts with replay, breadth and alerts. Every panel names its source and price basis.
Options and flow
Options lab, GEX with its assumptions shown, 0DTE, flow and dark-pool prints with their reporting lag labelled.
Research
Backtesting in a Pine subset or restricted Python. Every run counts as a trial, including the ones you threw away.
Risk and the AI analyst
VaR, expected shortfall and a kill switch on every position. A local AI analyst writes the desk note. Its seat in the bot ensemble is earned against resolved outcomes and is 0.00 today.
Counted from the terminal’s own navigation: 69 screens · 8 modules. The analyst’s vote is graded like everything else: Its ensemble vote carries only the weight it has earned against resolved outcomes — measured 0.00 on 2026-09-05, so it currently changes nothing.
Against retail charting and rented terminals.
Five rows from the full matrix — including the ones where the other tools win, which is on the compare page in the same table.
| Capability | QuantSuite | TradingView | QuantVue | Aladdin |
|---|---|---|---|---|
| Price (annual) | $299.99 / yr | ~$300 / yr (Premium tier, approx.) | ~$1,200 / yr (approx., unverified) | Enterprise contract — typically 6–7 figures, quote-based |
| Deflated performance stats (PSR / DSR with trial count N) | ✓ on every metric | — | unverified | partial (risk analytics; DSR not publicly documented) |
| On-device AI analyst (nothing sent to a cloud) | ✓ local model, your GPU/CPU | — | unverified | cloud-hosted |
| Honest backtesting (Pine + Python sandbox, trial accounting) | ✓ | Pine only, no deflation | unverified | — (portfolio risk focus, not retail strategy backtests) |
| Runs offline / data stays on your machine | ✓ | cloud | cloud (unverified) | cloud / hosted |
A license key, not an account.
Signed key issued instantly. Verified fully offline. Cancel anytime. No tiers, no feature locks, no per-seat math — details on the pricing page.
- All 69 screens, one terminal, one build
- On-device AI analyst
- Bring-your-own broker + data keys
- Offline HMAC-signed license
- Windows today, macOS/Linux coming
- The identical build — all 69 screens
- On-device AI analyst
- Bring-your-own broker + data keys
- Offline license, valid 365 days
- Windows today, macOS/Linux coming
Twelve months at $50 is $600. The annual plan is $299.99 — same build, no feature gap.