Back-Adjusted Price Series

If a stock trades at $100 the day before a 2-for-1 split and $50 the day after, the raw price series shows a 50% drop that never actually happened to any shareholder's wealth — each holder now just has twice as many shares worth half as much. A backtest computing returns off raw, unadjusted prices would record a catastrophic one-day loss that no investor experienced. Back-adjusted price series exist to fix exactly this: they rescale historical prices so a return calculated across a corporate action reflects what an investor who held the position actually earned.

The idea

Splits, stock dividends, spin-offs, and cash dividends all change a raw price without changing an investor's underlying wealth in the way a simple price difference would suggest. Back-adjustment works backward from the present: on the date of a corporate action, all prior historical prices are multiplied by an adjustment factor (for example, 0.5 for a 2-for-1 split, or a slightly-less-than-one factor for a cash dividend reflecting the dividend as a fraction of price) so that the percentage return calculated from adjusted prices across the action date matches the investor's actual return, including reinvested dividends where applicable.

There's a real tradeoff hidden here: back-adjusted prices are correct for calculating returns, but they no longer represent prices that ever actually traded, which matters if a strategy's logic depends on the literal historical price level (for instance, a stop order set at an exact dollar level, or an options strategy referencing historical strike distances). For those uses, the raw, unadjusted price series is still needed alongside the adjusted one.

Worked example

A stock closes at $100 on Friday and undergoes a 2-for-1 split over the weekend, opening Monday at $50. The raw price series shows Friday $100, Monday $50 — a −50% return. The back-adjusted series instead multiplies every price before Monday by 0.5, so Friday's adjusted price becomes $50, matching Monday's $50 and correctly showing a roughly flat return, consistent with the fact that a holder's total position value didn't materially change over the split.

What this means in practice

Every serious backtest of individual equities needs to run off a back-adjusted series, and needs to know whether that series adjusts for dividends as well as splits, since "price return" and "total return" series diverge substantially over long horizons for dividend-paying stocks. Mixing raw and adjusted prices within the same backtest — a common bug when merging data from different vendors — silently manufactures huge fake gains or losses around every corporate action date.