This update changed how Gielinor Gains handles price history. Instead of repeatedly pulling full OSRS Wiki timeseries data, the site now builds its own store from the Wiki 5-minute feed.
Short version: same product experience, far fewer API calls, and a lighter footprint on shared community infrastructure.
The old problem
- Each precompute cycle was making roughly 1,400 to 2,500 Wiki timeseries calls.
- Those cycles were running every few minutes.
- That added up to nearly 1 million timeseries calls per day.
- Most refreshes only needed recent ticks, not a full historical pull for every item.
The new approach
The better path was the Wiki 5-minute feed. Gielinor Gains now consumes that feed and builds a custom timeseries store in Upstash KV.
That gives the app a data shape built for its own scoring, filtering, and charting needs without asking the Wiki for the same historical data over and over.
The result
- Wiki timeseries calls dropped from roughly 1 million per day to about 288 feed calls per day.
- That is a 99.97% reduction in that part of the data pipeline.
- The site should feel the same to users, and in some refresh paths it may feel snappier.
- Infrastructure cost and shared API pressure both move in a healthier direction.
The Vercel consumption chart tells the story clearly: the pipeline moved from expensive, high-volume API usage to a leaner system built for the way Gielinor Gains actually reads market data.
