Alerts fail in one specific way
Not by missing the move. By firing so often that you stop reading them. Every trader who has abandoned an alert system did so after the fortieth notification that did not matter, and the fix is almost never a better trigger — it is fewer of them.
So the useful question is not "what can I get alerted on" but "which alerts have actually earned their place". Those are different questions and only one of them has an answer you can check.
What actually triggers an alert
TradePilot has two layers, and they are worth separating because they fail differently.
Price alerts are the simple layer: a ticker goes above a number, goes below a number, or moves by a percentage. Three conditions, no interpretation. There is also an automatic layer that flags a rapid rise against a baseline price, and a take profit signal on a position you hold — both of which log the baseline, the current price and the percentage change that fired them, so you can see the arithmetic rather than trust the label.
Custom alerts are the composable layer: a named alert on a ticker with a list of conditions, each a field, an operator and a value. The list is the point. One condition is a tripwire; several together is a thesis.
Why multi-condition alerts are quieter and better
"NVDA above $190" fires on noise — any intraday wobble past the line, repeatedly, including at 3am. "NVDA above $190 and volume above average and sentiment positive" describes a situation rather than a number, and situations are rarer than numbers.
A single-condition alert on a volatile ticker can fire a dozen times in a session and tell you nothing each time. Adding a second condition usually cuts that to near zero, and the one that survives is worth opening the app for.
The number most alert systems hide
Every custom alert in TradePilot carries a trigger count and the timestamp it last fired. That sounds like bookkeeping. It is the most useful field on the object.
An alert that has fired 200 times is not a signal, it is a subscription to noise, and you now have evidence rather than a feeling. An alert that has never fired in three months is either badly calibrated or describes something that does not happen — equally worth knowing. Without those two fields the only way to judge an alert is to remember how annoying it has been, and nobody remembers accurately.
Sort your alerts by trigger count occasionally. The top of that list is where your notification fatigue is coming from, and it is usually two or three alerts doing all the damage.
Where notifications actually arrive
Two channels, and being precise matters because it changes what you can rely on.
Desktop notifications through the browser, which need permission granted and the browser running. Mobile push via ntfy, which you enable in Settings and point at a topic — by default a public ntfy server, or your own if you run one.
Neither is a guaranteed delivery channel. A phone in do-not-disturb, a closed browser or a flat battery all silently swallow an alert, and there is no email fallback in the alerting path. Treat alerts as a convenience that often reaches you, not as a safety net that always does. If a position genuinely must not be missed, the broker's own order types are the mechanism for that, not a notification.
A setup that stays readable
Start with fewer alerts than feels right. Three or four across positions you are actively managing beats twenty across a watchlist, because twenty becomes zero the moment you start dismissing them without reading.
Prefer a condition that describes a decision you would actually make. "Below my stop" is a decision. "Crossed a round number" is a coincidence. If a triggered alert would not change what you do that day, it should not exist.
Then review the trigger counts after a fortnight and delete the loudest one. Repeat until the notifications you get are ones you open.
Set up and prune your alerts on the alerts page, or create a free account to start.
Nothing here is financial advice. Alerts are a notification convenience, not a guaranteed delivery mechanism or a risk-management tool.
Tier alerts by what you would actually do
If you keep more than a handful, sort them by the response they deserve rather than by ticker. Three tiers is enough.
Interrupt me: a price through a level you have already decided to act on, or a high-conviction recommendation on something you hold. Conviction is scored 0-10, so this is the 8-and-above end — not a percentage, whatever other tools have trained you to expect.
Tell me today: a move worth knowing about that does not change anything before this evening. These belong on a channel you check deliberately rather than one that buzzes.
Do not alert at all: anything you would read and take no action on. That is most of what people configure in their first week, and deleting those is what makes the remaining alerts readable.