On August 31, 2020, Apple executed a 4-for-1 stock split. On the preceding Friday, Apple closed at roughly $499. On Monday morning, it opened at roughly $125. An investor holding 100 shares now held 400 shares with identical total wealth.
However, if you fed raw 'Close' prices into a computerized trend-following strategy, the algorithm would register a catastrophic -75% one-day crash, trigger every emergency stop-loss, and corrupt five years of moving averages.
This is why financial datasets offer two distinct columns: Close and Adjusted Close. Choosing the wrong one is one of the most common ways beginner quant researchers silently ruin their backtests.
Close Price (Raw Close) is the actual, unadjusted cash price at which a security settled at market close on that specific historical day. It is essential for tax reporting, option strike matching, and verifying real transaction receipts. Adjusted Close backward-adjusts historical prices for corporate actions—primarily stock splits and cash dividends—to reflect true continuous total returns. Using unadjusted close prices in a backtest creates artificial price gaps that falsify indicators, trigger fake signals, and distort return calculations.
| Dimension | Raw Close Price (Unadjusted) | Adjusted Close Price (Split & Dividend Adjusted) |
|---|---|---|
| Definition | Actual traded settlement price on that calendar date | Hypothetical historical price reconstructed to preserve economic continuity |
| Stock Split Impact | Experiences sudden sharp drops (e.g. -50% on a 2-for-1 split) | Smoothly scales all prior history so no artificial cliff appears |
| Cash Dividend Impact | Ignores cash paid out, causing price to look like it dropped on ex-date | Reinvests dividend value into past series to show true total return |
| Technical Indicators | Completely corrupts long-term moving averages (50-day, 200-day) | Produces mathematically clean, uninterrupted moving average series |
| Best Use Case | Options strike pricing, broker statement reconciliation, execution audits | Strategy backtesting, factor ranking, long-term performance attribution |
The Silent Distortion of Cash Dividends
While stock splits create obvious cliff-like crashes that are easy to spot on a chart, cash dividends cause a far more insidious distortion.
On the ex-dividend date the stock trades without entitlement to the dividend, so market prices usually reflect that value adjustment, but the actual opening price is still determined by trading. Shareholders receive that cash dividend in their accounts, so total investor wealth is not diminished by the payout itself.
If you evaluate the stock using unadjusted Close, the stock appears to suffer an artificial loss every single quarter. Over a ten-year backtest, an unadjusted chart will make a high-dividend dividend-aristocrat look like a stagnant underperformer, when in reality its total return was compounding handsomely.
Never delete your unadjusted Close column! You still need raw Close for three critical tasks: 1. Option Trading: Options contracts trade at specific unadjusted dollar strikes ($150, $200). You cannot price historical option chains using adjusted stock prices. 2. Broker Fill Verification: Auditing old brokerage statements requires matching exact historical receipts. 3. Intraday Resistance Levels: Historical high-water mark psychological round numbers (like $1,000/share) were reached in raw dollar terms.
3 Practical Rules for Your Data Pipeline
To prevent corporate action distortions from poisoning your systematic models, follow these guidelines:
- 11. Use Adjusted Close for Technical Signals: Calculate moving averages, RSI, Bollinger Bands, and breakout thresholds exclusively on split-adjusted (or split-and-dividend adjusted) series.
- 22. Clarify Total Return vs. Price Return: Understand whether your data provider's 'Adjusted Close' includes dividends (Total Return) or only stock splits (Price Return). Using Total Return is ideal for multi-year investor performance, but intraday price breakouts must use split-only adjustments.
- 33. Match Trade Execution Units: When testing position sizing, ensure your share count calculation divides allocated dollar capital by the price series matching your execution universe.
Frequently Asked Questions
Does a stock split make existing shareholders wealthier?
No. A stock split simply divides the company's equity pie into more slices. If a company does a 2-for-1 split, you have twice as many shares, but each share is worth half as much. Total equity value remains unchanged.
Why do historical adjusted prices change whenever a new dividend is paid?
Because adjusted close formulas work backward from the present. Every time a new dividend is paid today, the adjustment multiplier updates, slightly shifting all historical adjusted close numbers across the entire multi-year series.
Can adjusted close prices ever become negative?
In rare cases involving massive cumulative cash dividends or distressed reverse splits over decades, standard backward-adjustment algorithms can mathematically generate negative adjusted prices. Specialized institutional databases cap adjustments to prevent negative prices.
