Setting thresholds that mean something
A threshold should correspond to a decision you have already made. "Notify me if this drops twelve percent" is useful if twelve percent changes what you do; otherwise it is noise you will learn to ignore.
Percentage moves relative to recent volatility are generally more informative than fixed prices, because a three percent move means different things for different assets.
Alert fatigue is the real failure
Too many alerts produce the same outcome as no alerts: they are dismissed without reading. A small number of alerts tied to real decisions outperforms a dashboard of notifications.
If an alert fires and you do nothing, either the threshold is wrong or the alert should not exist.
What alerts genuinely cover
Price thresholds, unusual volume relative to an asset's own median, approach to a level with a known touch history, and conditions on assets you hold. These are all checkable continuously and cheaply.
Delivery matters: an alert that arrives through a channel you do not check is not an alert.
Conditions should also be able to expire. An alert set during one market regime and left running for six months fires on a threshold that no longer means anything, and by the time it arrives you have usually forgotten why you set it in the first place.
The limitation stated plainly
An alert never fills an order. If the price moves while you are asleep, you receive a notification about something that already happened and may have reversed before you read it.
FBT Swap can monitor conditions and notify you. It cannot execute an unattended trade, because every swap requires your wallet signature — alerts never fill unattended.