Price, trend and momentum
Moving averages and crossings, RSI, MACD, ADX, ATR bands, Bollinger and Keltner squeezes, breakouts from a range you define.
A screener thatkeeps working afteryou close the tab
Write the conditions once — in indicators or in plain language — and they run on a schedule across every pair you follow. What fired is kept, so you can see what the screen actually caught rather than what you remember it catching.
What makes a screen worth trusting
It runs on a schedule, not when you remember
A screen is a set of conditions plus a cadence: every five minutes, at the hourly close, once a day. It runs on the server against closed candles, so the value it tests is the value that will still be there tomorrow — a condition evaluated on a forming bar is testing a number that has not happened yet, and it is the single commonest way a screen appears to work and does not.
Every firing is kept
When a rule fires it is written down: which symbol, which condition, which run, at what price. That history is the only honest way to judge a screen — a rule that fires forty times a day is noise no matter how good the one you remember was, and you cannot see that without the record.
A degraded run says so
If market data stalls and a symbol cannot be evaluated, the run is marked degraded rather than reported as a clean pass with fewer results. A screen that quietly returns nothing looks identical to a market with nothing in it, and those two need to look different.
What it finds can go straight into research
A hit is a starting point, not a signal. From the alert you can ask the assistant what changed, pull the instrument’s history, run the setup through a backtest, and — if you get that far — draft an order that still has to pass your risk limits like every other order.
What you can screen on
Indicators computed on closed candles, the same series the backtester reads — so a condition means the same thing in the screen and in the test.
Volatility and regime
Realised volatility, squeeze conditions and a trend/range regime label with a warm-up long enough that the label stops flipping.
Derivatives
Funding, open interest and the long/short ratio, for the pairs where perpetuals are where the crowd actually is.
Cross-symbol ranks
Rank the whole universe by a measure and take the top of it, rather than testing each symbol against a fixed threshold that suits one market and not the next.
Delivery where you will see it
A firing reaches the console and, if you want it, Telegram. It never reaches an exchange: an alert is information, and turning one into an order is a separate decision with the risk gate in front of it.
Questions about screening
How many symbols can it watch?
The whole universe you have enabled per venue, on the cadence you set. The practical limit is data, not screens: a symbol with too little history cannot be evaluated for a condition that needs a long look-back, and it is reported as such rather than silently skipped.
Can the AI write the conditions for me?
Yes — describe what you are looking for and it drafts the rule, which you then read and save. It is drafting a filter, not a strategy, and it does not enable anything on your behalf.
Will it place a trade when a screen fires?
Not on its own. An alert can start a conversation, a research job or a strategy you have already promoted, and any order that comes out of that still passes the same risk checks as one you typed yourself.
What if a screen stops finding anything?
Look at the run history first. Runs that were degraded tell you the data stalled; clean runs with no hits tell you the market changed. Those get treated as the same thing in most tools, and they should not be.
Run a screen tonight, on paper
Paper accounts get the whole screener — the same schedules, the same alert history — against live market data.