Canonical Schema

When every dataset uses different column names - date vs ts_event vs timestamp, asset vs ticker vs symbol - every downstream notebook needs special-case handling.

The Problem

Provider A delivers date, ticker, close; Provider B uses ts_event, symbol, price; Provider C uses timestamp, asset, adj_close. Without a canonical schema, every downstream notebook needs provider-specific renames.

The Pattern

WRONG

import polars as pl

# Different column names per dataset - downstream code breaks constantly
etfs = pl.read_parquet("etfs.parquet")        # has: date, ticker, close
futures = pl.read_parquet("futures.parquet")   # has: ts_event, product, settle
crypto = pl.read_parquet("crypto.parquet")     # has: timestamp, symbol, close