ABOUT VERDICT
Verdict exists because its founder — a futures and forex trader — got tired of AI chart tools that sound confident and see nothing. Ask a vision model about your chart and it will describe fair value gaps that filled hours ago, find equal lows that were never there, and place a break of structure wherever it feels right. Then it will do it differently five minutes later.
The fix wasn't a better prompt. It was refusing to let the AI guess at anything a computer can measure. That decision is the entire product, and every principle below follows from it.
MEASURE, DON'T GUESS
If a claim can be computed, it must be computed. The AI's job is judgment on verified facts — never inventing the facts themselves.
EVERY MISS BECOMES A TEST
A chart the engine misreads isn't an apology, it's a regression case. The suite only grows; the engine only gets harder to break.
SAY WHEN IT'S A GUESS
When extraction can't verify a chart, the card says ESTIMATE in plain sight. Confidence is labeled, never implied.
NO PROMISED WIN RATES
Verdict grades structure for study. Any calibration claim will come from published, measured outcomes — or it won't be made at all.
COLD ON PURPOSE
No emojis, no hype, no guru voice. A grade with criteria and observations you can check yourself. The restraint is the respect.
The rest of the story is on the roadmap — built in the open, measured in public.