Insights

How to Handle Corporate Actions in Backtests

Alphanume Team · April 20, 2026

Splits, dividends, and adjustments done right — the corporate-actions handling that backtests routinely get wrong.

Corporate actions — stock splits, dividends, spinoffs, mergers, share repurchases — change the relationship between historical prices and current prices. Handling them correctly is a non-trivial data-engineering problem and a frequent source of subtle backtest errors. Most price-history datasets implement adjustment, but the details matter and naive use produces silent errors.

The two main types of adjustment

Two common kinds of price adjustment:

Split adjustments. For a 2-for-1 split, all pre-split prices are divided by 2 to make the price series continuous. This adjusts for the mechanical change in share count without distorting return computation.
Dividend adjustments. For each dividend, pre-dividend prices are adjusted down by the dividend amount (or, more commonly, by a factor that produces a continuous total-return series). The result is a total-return price series.

Both types of adjustment exist as separate series in well-structured data:

"Unadjusted" or "as-traded" prices: Actual prices as they traded on each date.
"Split-adjusted" prices: Continuous through splits only.
"Total-return" or "fully-adjusted" prices: Continuous through splits and dividends.

Which adjustment to use

For most return computation: total-return prices.

For computing dollar volume or absolute price levels at historical dates: unadjusted prices.

For computing total return excluding dividends: split-adjusted (without dividend adjustment).

The mix matters. A common error: use total-return prices to compute "average closing price was $25" — but the $25 in 2018 may correspond to an actual traded price of $40 if there have been splits and dividends since.

Spinoffs

Spinoffs are the most complex common corporate action:

The parent company distributes shares of a subsidiary to its shareholders.
On the spinoff date, the parent's stock price drops by approximately the value of the spinoff per parent share.
The shareholder now owns both parent and spinoff shares.

Treating the parent's price drop as a "loss" without accounting for the spinoff value is the standard error. Total-return adjustment should treat the spinoff as a distribution and adjust accordingly.

For event studies that include securities going through spinoffs in the event window, the proper handling is to either:

Compute the spinoff-inclusive total return for the period.
Exclude the spinoff event from the sample (cleaner but reduces sample size).

Mergers and acquisitions

For cash acquisitions:

Final price is the deal price.
The security delists at deal close.
Backtest holdings receive cash on the delisting date.

For stock acquisitions:

The acquired security's terminal value is the value of received shares.
The acquirer's security continues with the merger reflected in its share count.
Backtest holdings convert to the acquirer's shares at the exchange ratio.

For mixed cash-and-stock acquisitions: handle proportionally.

Stock dividends and splits

Adjacent to splits:

Stock dividends. 5% stock dividend = 1.05-for-1 split. Adjust as a split.
Reverse splits. Often associated with distress; 1-for-10 reverse split divides outstanding shares by 10 and multiplies price by 10. Adjust as a split (in the opposite direction).
Stock splits with cash components. Some splits include partial cash payments; handle each component separately.

The point-in-time problem

Adjustment factors change as new corporate actions occur. A price downloaded today is adjusted for all corporate actions through today. A price downloaded a year ago was adjusted only for actions through that date.

For point-in-time backtests:

Store the as-of-then adjustment factors.
Or store unadjusted prices and apply adjustments as of the historical date being simulated.

Naive use of currently-adjusted data in historical backtests introduces look-ahead through the adjustment process.

Volume adjustment

Volume must also be adjusted for splits:

A 2-for-1 split doubles the share count; volume in the post-split series is effectively half the pre-split volume per dollar traded.
For computing historical ADV in dollars, multiplying split-adjusted prices by split-adjusted volume gives correct dollar-volume.
For computing share-volume (e.g., short interest as % of shares outstanding), care is needed across split boundaries.

Common errors

Mixing unadjusted and adjusted prices. Computing a return as (adjusted close / unadjusted close − 1).
Failing to adjust for dividends. Underestimating total returns for dividend-paying names.
Using current adjusted prices in historical backtests. Introduces look-ahead.
Mishandling spinoffs. Treating the parent price drop as a loss.
Forgetting reverse splits. Producing seemingly-extreme returns that are actually mechanical adjustments.

Validation

To verify corporate-action handling:

Compute long-horizon total returns and compare to published benchmark returns.
Spot-check known-event names (recent splits, large dividends, mergers).
Verify that pre-split prices in the adjusted series make sense relative to share-count history.