Guide
Volume-spike alerts
An absolute volume threshold is wrong the moment conditions change. Comparing volume to its own recent average keeps the alert meaningful.
The problem with a fixed number
Suppose you set volume > 5000. In a quiet week that fires on every second candle. In a busy week it never fires at all, because baseline volume has moved above it. The number describes a market condition that has already changed.
Compare volume to itself
Instead, compare each candle's volume to the average of the candles before it:
volume > 2 x SMA(volume, 20)
This reads: the volume on the candle that just closed was more than twice the mean volume of the previous twenty candles. The threshold rescales itself as the market changes, so the alert keeps meaning the same thing.
Both sides are operands
In this builder, the right side of a condition is not restricted to a number. It can be a constant, a candle field, or a computed value like a moving average with a multiplier. Volume is not a special kind of alert — it is a field an operand can read, exactly likeclose is.
Picking a multiplier and a length
A multiplier around 2 catches genuine expansion without firing on normal variation. Push it to 3 or 4 if you only want outliers. The length sets what counts as "recent": SMA(volume, 20) on a 15m chart averages the last five hours.
The lookback limit
Averages here are capped at 24 periods (23 if the condition uses a crossover). That is a deliberate consequence of how the evaluator fetches data — it reads a fixed 25-candle window per check, which keeps the service inside its free-tier compute budget. A longer average would need more history than a single fetch can carry.