The problem with "just download the history"
A 20-year daily price series from a modern vendor looks clean on screen. It is clean because the vendor has already made a large number of adjustment decisions on your behalf. When you splice two sources together, or when you build the history yourself from raw announcements, those decisions are no longer made for you — and if they are made inconsistently across assets, your cross-asset research will be quietly wrong.
The three decisions that matter most for Australian data are split and capital-return adjustment, dividend treatment, and surviving universe.
Splits and capital returns
ASX-listed companies often return capital via schemes that are not simple share splits. A 1-for-1 consolidation, a return of capital, or a special dividend all change the price series in different ways. If you apply a uniform "adjust for splits" rule across all of these, your adjusted history will not match what an investor actually experienced. We recommend documenting, for each corporate action, the exact convention used, rather than relying on a vendor back-out.
Dividends
For equities, the choice between price return and total return should be explicit and consistent across the universe. A common error is to mix price-return history for older names with total-return history for newer ones, which flatters the newer names and biases any comparison against older listings.
Surviving universe
If your historical universe contains only names that still exist today, your back-tests will inherit survivorship bias. For Australian research this matters particularly for small-cap and sector-specific studies, where delisted names can materially change long-run conclusions. Building a point-in-time universe — including delisted and merged names — is the single most effective step you can take.