Understanding Corporate Actions & Historical Data

What a corporate action is

A corporate action is something a company does that affects its own shares — how many exist, what each one is worth, what they pay out, or whether the company continues to exist in its current form.

Five are worth knowing before you read a backtest, because each one affects historical price data:

One more term does all the work in this article. The ex-date is the cut-off: from that date, the share trades without the entitlement attached to the action. Someone buying on or after the ex-date does not receive the bonus shares or the demerged company. It is the moment the change becomes real in the price.

Why this matters to a price chart

Here is the problem in one example. It is illustrative, with round numbers chosen to make the arithmetic visible.

A company announces a 1:1 bonus — one free share for every share held.

A 1:1 bonus issue, before and after

The day before

• 100 shares held
• ₹200 a share

=

₹20,000

position value

The day after

• 200 shares held
• ₹100 a share

=

₹20,000

position value

Nobody gained anything and nobody lost anything. The same value is simply divided into twice as many pieces.

But look at what the raw price history now shows: ₹200, then ₹100. A strategy reading that data sees a 50% overnight crash. Every rule you might have written reacts to it — a moving average bends, a rate-of-change filter registers a collapse, a stop loss triggers on a fall that never happened.

This is the whole issue. The price genuinely changed; the shareholder's wealth did not. Historical data that records the first without accounting for the second is not just imprecise, it is actively misleading.

NiveshBuddy's history runs from 2021, on end-of-day prices, and it is adjusted for corporate actions — which is the answer to this problem for everything that looks backwards. What changes is how a simulation, which moves forward through today's market, deals with an event as it happens. The rest of this article takes the two in turn.

How the simple cases are handled

For bonus issues and stock splits, NiveshBuddy adjusts automatically. Both the historical price data and the position quantities are corrected so the record stays continuous — in a backtest and in a simulation alike.

In the example above, the historical prices before the bonus are restated on the new basis, so the series reads as a smooth ₹100 rather than a fall from ₹200. The holding of 100 shares becomes 200. The position's value is unchanged at every point, and the chart shows what actually happened to the investor's money rather than what happened to the label on one share.

In a simulation you can see it happen. The holding's quantity and average price are restated, and the row in Active Portfolio carries a Restated pill — its value and profit or loss are unchanged. If the ratio leaves a fraction of a share (a 3:2 split of 101 shares, say), that fraction is paid out in cash, and the trade log records it as "Closed: split or bonus fraction paid in cash".

Why these two events can be handled this way: the arithmetic is exact. A 1:1 bonus or a 2:1 split has one unambiguous conversion, known in advance, with no judgement involved. Adjusting is simply applying it.

Dividends need no special treatment at all, for a related reason. The historical price data a strategy is evaluated against is already adjusted for them, so a dividend never shows up as an artificial fall on its ex-date. There is no dividend rule in the builder, nothing to configure, and no dividend event to hunt for in a trade log — the effect is simply present in the data the strategy reads.

Mergers and demergers: a backtest and a simulation differ

Mergers and demergers do not have that property. After a merger, a shareholder may hold shares in a different company, at a ratio set by the terms of the deal, in an entity with a different business and a different price history. After a demerger, they hold two things where they held one, and the new listing may have no trading history at all.

In a backtest, the data has already done the work. A backtest runs once, over a history that is already adjusted for corporate actions — mergers and demergers included. The event is settled in the data before the backtest reads it, so the backtest does not sell anything because of it.

In a simulation, there is no adjusted history to lean on yet. A simulation moves forward through the live market, meeting each event as it is announced, with terms that are not yet reflected in any series. Continuing the holding through the event would mean guessing at a conversion and a valuation. So NiveshBuddy does not attempt it: the affected holding is sold in the simulation before the ex-date.

What a simulation does with a holding facing a merger or demerger

1. The event is scheduledAn ex-date is set for the merger or demerger
2. The holding is soldBefore the ex-date, before the structure changes
3. Proceeds become cashThe sale proceeds sit in Cash Balance
4. Redeployed at the next rebalanceCash re-enters on the strategy's ordinary schedule

This is a deliberate modelling choice with a stated cost, not an oversight. The platform's aim is a record that avoids assumptions it cannot make reliably — and stepping aside is the honest version of that. It does mean the simulation does not follow whatever the merged or demerged entity did next; NiveshBuddy is not reconstructing that outcome inside the simulation. It also means a simulation and a backtest of the same strategy can differ around such an event — one more reason the two are not expected to match exactly.

Reading this in your own results

Once you know the mechanics, the effects become recognisable rather than puzzling.

None of this is something you configure, in a backtest or a simulation. There is no setting to change, and no decision to make while building a strategy. It is worth understanding for one reason: so that when you see it in your results, you read it correctly instead of concluding the strategy misbehaved.

Common misconceptions

• "A price drop on a split means the stock fell." It means the share was divided. Your position's value is unchanged, and the adjusted history is what reflects that honestly.
• "Bonus shares are a gain." They are the same value spread across more shares. A bonus is worth having for reasons of liquidity and accessibility, but the moment it happens it adds nothing to what you own.
• "A simulation should just keep holding through a merger." Continuing would require assuming a conversion and a valuation that cannot be modelled reliably while the event is still unfolding. Selling first states the uncertainty instead of hiding it inside a result.
• "A merger sale in my simulation means the strategy made a decision." The strategy's rules did not fire. The event forced the sale, and the cash returns to the strategy's normal process at the next rebalance.
• "My backtest must have sold the stock at its merger too." It did not. A backtest reads history that is already adjusted for the event.
• "A dividend will show up as a drop in my backtest." It will not. The price data is already adjusted for dividends, so there is no artificial fall to explain and nothing for you to account for.
• "Corporate actions are an edge case I can ignore." Dividends, bonus issues and splits are routine in Indian equities. A backtest over several years across a broad universe will meet many of them.

Key takeaways