For AI agents: the complete documentation index is at llms.txt. Every page is also available as markdown by appending .md to its URL, or by sending an Accept: text/markdown request header.

Find local minimum and maximum

Find the minimum and maximum values within a time window around each row to detect local peaks, troughs, or price ranges.

Problem​

You want to find the local minimum and maximum bid price within a time range of each row - for example, the min/max within 1 second before and after each data point.

Solution 1: Window function (past only)​

If you only need to look at past data, use a window function with RANGE:

Local min/max from preceding 1 secondDemo this query
SELECT timestamp, bid_price,
  min(bid_price) OVER (ORDER BY timestamp RANGE 1 second PRECEDING) AS min_price,
  max(bid_price) OVER (ORDER BY timestamp RANGE 1 second PRECEDING) AS max_price
FROM core_price
WHERE timestamp IN '$now - 1m..$now' AND symbol = 'EURUSD';

This returns the minimum and maximum bid price from the 1 second preceding each row.

Solution 2: WINDOW JOIN (past and future)​

If you need to look at both past and future data, use a WINDOW JOIN. QuestDB window functions don't support FOLLOWING, but WINDOW JOIN allows bidirectional lookback:

Local min/max from 1 second before and afterDemo this query
SELECT p.timestamp, p.bid_price,
  min(pp.bid_price) AS min_price,
  max(pp.bid_price) AS max_price
FROM core_price p
WINDOW JOIN core_price pp ON symbol
  RANGE BETWEEN 1 second PRECEDING AND 1 second FOLLOWING
WHERE p.timestamp IN '$now - 1m..$now' AND p.symbol = 'EURUSD';

This returns the minimum and maximum bid price from 1 second before to 1 second after each row.

When to use each approach​

ApproachUse When
Window functionYou only need to look at past data
WINDOW JOINYou need to look at both past and future data
Related Documentation