Every investor who has used a stock-tracking app knows the failure mode: notifications arrive constantly, most of them are noise, and after a few weeks you either mute the app or start ignoring it. AxisFolio’s alerting system was built to avoid that outcome by asking a different question before every notification goes out — not just "did something happen," but "does this actually change the risk profile of a position you hold."
Every corporate action starts as raw data
Dividends, stock splits, buybacks, and earnings dates all arrive as structured events with no inherent urgency attached. The naive approach is to forward every one of them as a push notification. The more useful approach is to score each event against the position it affects — size, timing relative to other portfolio activity, and historical volatility around similar events for that ticker.
An example
A routine quarterly dividend on a stable, low-volatility holding scores low — it’s informational, and it is bundled quietly rather than pushed as an interruption. An earnings date landing during an already elevated-volatility window for a concentrated position scores high, because the combination of timing and exposure genuinely changes the near-term risk picture.
Ranked, not just filtered
Filtering alone would mean silently dropping "low-importance" events — which erodes trust the first time something small turns out to matter. Instead, everything is retained but ranked, so the feed always shows what matters most first, with lower-priority items still visible below rather than discarded.
- High-risk events surface first, with the specific reason for the ranking shown alongside them.
- Routine events are still logged and visible, just not pushed as interruptions.
- The scoring model updates as volatility and portfolio composition change — a ranking is never static.
Design principle
An alert system earns trust by being right about what to skip, not just right about what to flag.
Why this is a signal-engine problem, not a notifications problem
It would be easy to treat this as a UI feature — batching, snoozing, notification preferences. Those help, but they don’t solve the underlying issue, which is that raw event data has no opinion about what matters to a specific portfolio. That opinion has to come from the same risk-aware layer that scores every trading signal, which is why alerting in AxisFolio is built on top of the signal engine rather than as a separate system next to it.