I'm building an alerts system using market data APIs

Great use case. An alerts system for market data usually works best if you design it as a pipeline:

1. Ingest market data
2. Normalize and store it
3. Evaluate alert conditions
4. Deduplicate / throttle notifications
5. Deliver alerts via email, SMS, push, webhook, Slack, etc.

A few practical considerations:

1. Data source choice

Depending on what you’re tracking:

• Equities / ETFs: real-time or delayed quotes, OHLC bars, volume
• Crypto: often easier access, lower latency, more flexible APIs
• FX / commodities: verify licensing and freshness
• News/sentiment: useful for event-driven alerts

Look for APIs that provide:

• WebSockets for live updates
• REST for historical/backfill
• Reliable rate limits
• Corporate actions handling if you track stocks
• Clearly licensed redistribution rights

2. Alert types

Common alert conditions:

• Price above/below threshold
• % move over a time window
• Volume spike
• Moving average crossover
• 52-week high/low
• News keyword/entity match
• Volatility breakout

3. System design

A simple scalable architecture:

• Market data service pulls or subscribes to streams
• Event bus / queue (Kafka, Redis Streams, SQS, RabbitMQ)
• Rule engine evaluates alerts
• Notification service sends messages
• User preferences DB stores alert rules and channels

4. Important engineering issues

• Debouncing: avoid sending the same alert repeatedly
• Stateful rules: “crosses above” requires remembering prior state
• Backtesting: test alert logic on historical data
• Latency: if alerts are time-sensitive, prefer streaming over polling
• Rate limits: batch requests and cache aggressively
• Timezone / market hours: handle open/close and pre/post-market
• Data quality: missing ticks, outliers, stale quotes

5. Rule engine pattern

A clean approach is to represent alerts as:

• Trigger: condition on a symbol or set of symbols
• Window: instant, 1m, 5m, 1h, etc.
• Cooldown: minimum time between notifications
• Severity: informational, warning, critical
• Action: notify, log, webhook, trade, etc.

Example:

• “Alert me when AAPL crosses above $200, but only once per hour.”

6. Suggested MVP

If you want to ship quickly:

• Support price threshold alerts
• Use one market data API
• Store alert definitions in Postgres
• Run a worker that evaluates each tick/bar
• Send email + webhook notifications
• Add cooldowns and alert history

If you want, I can help you with any of these next:

• choose the right market data API
• design the database schema
• write alert-evaluation pseudocode
• build a WebSocket/polling architecture
• implement this in a specific stack like Python, Node.js, or Go