MarketNebula scans every US-listed stock against filters you build yourself, then lets you test the idea against a year of real market history before a dollar is committed. Everything runs on paper money.
Most screeners quietly show you a subset. Ours reads a local copy of the market, so "scan all stocks" means all of them — not the first few hundred that fit inside a timeout.
Price, size, volume and sector. Performance across seven windows. Momentum, trend, volatility and range measures. Float, short interest, earnings surprise, analyst targets. Combine as many as you like.
“Scan all stocks” means all of them, not the first few hundred that fit inside a timeout. Every run reports how many names each filter removed, so an empty result tells you why it was empty instead of leaving you guessing.
The assistant turns a description into a concrete screen and explains its reasoning. It can only propose filters the engine can actually evaluate, so you never get a suggestion that quietly fails. Overrule any part of it.
Click any result for intraday through five-year charts, with extended hours drawn separately from the regular session, earnings markers, and a prior-close reference line so you can see at a glance whether today is genuinely green.
A strategy that looks good on a chart is not evidence. Every automated idea here has to survive testing against real history first, and most of them do not — which is the point.
Every automated idea is replayed against real market history with its decision frozen at the moment it would actually have been made. Nothing the strategy could not have known is allowed to reach it.
Results are measured after the real cost of getting in and out, at the moment of entry. Typical and average outcomes are reported separately, because a strategy can win most of its trades and still lose money.
Orders route to a paper account with real order handling, real fills and real position tracking — entirely simulated money. Nothing is proposed that could not actually be filled.
A scheduled pre-open pass over the entire market, surfacing a short, ranked shortlist before the session starts. The data layer is built; the ranking model is still being validated.
Automatic stop-loss and take-profit levels, and the intraday logic to act on them. Still being tested — so it is not here yet, and we would rather say so than ship a number we cannot stand behind.
Chart and candlestick patterns, each shown with its own measured hit rate rather than the folklore from a poster. If a pattern turns out to be a coin flip, the platform will say so on the pattern.
MarketNebula isn't open for public signup. Tell us who you are and what you'd use it for, and we'll take a look.
