All performance math deliberately reads the adjusted close series. That includes Screener trailing returns, category comparisons, and Allocation models' Backtest curves.
An adjusted close folds dividends and splits back into the price path, so a return computed from it is a total return: what you'd have earned with distributions reinvested. Raw closes would make an income fund look like it "loses" its yield every distribution date and would break across every split.
Consequences worth knowing
- A fund's chart level here can differ from the price your broker quotes. The shape and the returns are what's comparable, not the raw level.
- Dividends are already inside every performance number. Adding income on top of these returns would double-count it.
- When a data source misses a split, the adjusted series breaks visibly (a cliff). The manual-split mechanism exists to repair exactly that.
- A distribution that goes ex between two price updates lowers the provider's adjusted closes for every earlier day. Update prices re-reads the last 30 days it already stored to catch that and re-scales the earlier adjusted closes to match, so the distribution date reads as an ordinary day. The 30 days also catch a distribution the provider only records days or weeks after it happened. If a fund still shows a one-day drop on a distribution date that its category peers don't share, Rebuild history re-downloads the provider's whole corrected series.
- A split between two price updates is caught the same way. Once the provider has reported the split, the update re-scales both the raw and the adjusted closes it stored before it. Your account's value before the split then stays at the price you actually traded at, and the adjusted series shows no cliff. A split you entered yourself is never applied this way. It stays a read-time correction.
- Some providers record a distribution one trading day away from the day the fund's price actually dropped. Read as stored, that shows a jump one day and a fall the next (for one growth fund, +106% then −52%), which inflates its volatility and drawdown. When the adjustment is 5% or more, the app reads that one day at the adjustment it should have carried, so the two days show what really happened. The downloaded prices themselves are left exactly as the provider sent them, and the price a holding is valued at is never changed.
If a number here disagrees with a fact sheet, check the window first (fact sheets love month-end windows), then the price source. Total-return math itself is the industry standard; it is the input series that varies.