Data Normalization
Data normalization standardizes data from different sources into a consistent format. It ensures you can compare and combine data regardless of its origin.
Understanding the Concept
Every exchange has opinions. Some return timestamps as Unix milliseconds, others as ISO strings. Some put bid before ask, others reverse it. Some use camelCase, others use snake_case.
Data normalization irons out these differences. After normalization, a price from Binance looks exactly like a price from Coinbase looks exactly like a price from Kraken. Same fields, same formats, same units.
This matters because comparison requires consistency. You can't calculate arbitrage if one exchange reports BTC at 50000 and another at 50000.00 and another at "50,000 USD". Normalization fixes this.
Real-World Example
Raw data: Binance {"price": "50000.50"}, Coinbase {"price": 50000.5}, Kraken {"price": "50000.50000000"}. Normalized: all become {price: 50000.5, type: "number"}. Now you can compare them directly.
How PRISM Handles This
PRISM normalizes all data before delivery. Timestamps are always Unix milliseconds. Numbers are always numbers (not strings). Field names are consistent across all sources. You write one parser that works for every exchange.
View API Docs