Corporate Actions: a beginner's guide
This page explains, from zero, what splits, bonuses, rights issues and dividends are, why they make
stock charts "jump", and how to review them on the Corporate Actions page
(/corporate-actions). No finance background is assumed.
A corporate action can make a share's price drop overnight even though nobody lost money. We fix old prices with a factor so charts and backtests don't see a fake crash. Your job on the review page is to check each factor on the chart and click Verify (the factor is right), Ignore (it's not a real corporate action), or leave it for later.
1. Why prices "jump" when nothing happened
Imagine a pizza cut into 4 slices. If you cut every slice in half you now have 8 slices, but you still have exactly one pizza. Each slice is simply worth half as much.
Shares work the same way. A company worth ₹1,000 crore split into 1 crore shares has a share price of ₹10,000. If the company splits every share into 5, there are 5 crore shares, and each is worth ₹2,000. The company and your holding are worth exactly the same, but the price on the chart drops from ₹10,000 to ₹2,000 overnight.
Raw closing price (no adjustment)
10,000 |■■■■■■■■■■■■■
| | <- looks like an 80% crash!
2,000 | ■■■■■■■■■■
+-------------+-----------> time
ex-date
A backtest reading these raw prices would think the stock crashed 80% and sell, or report a huge loss that never happened. That's why we adjust the old prices.
2. The factor: how we fix old prices
For each corporate action we store one number, the factor. It is a multiplier applied to every price before the ex-date. Prices from the ex-date onward are never touched.
| Split example | Before ex-date (raw) | × factor 0.2 | After ex-date |
|---|---|---|---|
| Close | 10,000 | 2,000 | 2,000 |
Adjusted closing price
2,000 |■■■■■■■■■■■■■■■■■■■■■■■ <- one continuous line, no fake crash
+-------------+-----------> time
ex-date
Important facts:
- We never change the stored prices. The database and Parquet files keep the raw prices
forever. The factor is applied only when a chart or backtest reads the data
(
eq_store.get_eod(adjust='price')). A mistake is therefore always reversible: change the row's status and the next read is correct again. - Only rows with status AUTO or VERIFIED are applied. REVIEW, IGNORED and SKIP rows are never applied.
- When there are several actions, their factors multiply. A stock that had a 1:2 split and later a 1:1 bonus has its oldest prices multiplied by 0.5 × 0.5 = 0.25.
- For splits, bonuses and detected jumps, volume is divided by the same factor, because the number of shares changed.
3. The action types
Split (sub-division)
What: each share is cut into smaller shares. NSE describes it as a face-value change, e.g. "Face value split from ₹10 to ₹2".
Factor = new face value ÷ old face value = 2 ÷ 10 = 0.2
Example: close ₹1,000 the day before, opens around ₹200 on the ex-date. Adjusted, the old ₹1,000 becomes ₹200. Continuous.
A consolidation (reverse split, e.g. ₹1 → ₹10) works the same way, with a factor above 1 (10 ÷ 1 = 10).
Bonus
What: the company gives existing holders free extra shares. "Bonus 1:2" means 1 new share for every 2 held.
Factor = held ÷ (new + held) = 2 ÷ (1 + 2) = 0.667
Example: close ₹300 the day before. After a 1:2 bonus you hold 3 shares instead of 2, so each should trade near ₹200 (300 × 0.667). A 1:1 bonus gives a factor of 0.5.
Bonuses in debentures or preference shares don't change the number of equity shares, so they are marked SKIP automatically.
Rights issue
What: holders may buy new shares at a discount ("issue price"). "Rights 1:4 @ ₹50" means 1 new share for every 4 held, at ₹50 each.
Because new shares are sold below the market price, the price drops a little. We use the standard TERP (theoretical ex-rights price):
TERP = (held × close before + new × issue price) ÷ (new + held)
factor = TERP ÷ close before
Example: close ₹100, rights 1:4 @ ₹50 → TERP = (4×100 + 1×50) ÷ 5 = 90 → factor 0.9.
If the issue price is missing we can't compute the factor, and the row stays REVIEW.
Dividend
What: cash paid to holders, e.g. ₹5 per share. The price usually drops by about the dividend on the ex-date.
Factor = (close before − dividend) ÷ close before = (200 − 5) ÷ 200 = 0.975
Dividends are small and happen every year, so they are only applied in total return mode
(adjust='total'). Normal charts (adjust='price') ignore them. A nil dividend is marked SKIP.
Symbol change
What: the company renames its ticker (INFOSYSTCH → INFY in 2011). Prices don't change. The row
tells the system to join the old symbol's history in front of the new one
(follow_renames=True), so charts show one continuous history.
What to do: check that both names are the same company. Verify if yes.
Delist
What: the stock stopped trading on NSE (merger, buy-back, liquidation, or moved to BSE only). The row only marks where the history ends. It never changes prices.
The review page shows a small grey DELISTED tag on any stock with no NSE trade in the last 45 days. Changing a dead stock's factor only affects that stock's old history. It still matters for backtests of the years when it traded (see survivorship bias), so verify big ones; small ones can be ignored.
Other (demerger / scheme)
What: a company splits off a part of its business into a new listed company (demerger), or another scheme NSE doesn't describe with numbers. The price drops by the value of the part that left, but NSE's text rarely gives a ratio.
What to do: look at the chart. The factor is roughly ex-date open ÷ previous close. Type it into the Factor box, check the chart is continuous, then Save + Verify. If you can't tell, leave it in REVIEW.
Adjust (detected price jump)
What: NSE's own list is incomplete, especially before 2005. The system therefore scans every stock for overnight jumps of about 25% or more that no listed action explains, and creates an ADJUST row with source DETECTED.
- [high] tier: the jump matches a common split or bonus ratio (2:1, 1:1, 5:1...), there was no
trading gap, and the next days confirm it. These are almost always real and were verified in
bulk (
accept-high). - [low] tier: the jump doesn't match a clean ratio. It could be a real action with an odd ratio, a demerger, or just a genuine big market move (results, fraud news...). Needs a human.
The note shows the evidence, e.g.
[low] no common factor: ro 1.828 rc 1.862 next 1.853, prev row 2026-08-14 (3d):
| Part | Meaning |
|---|---|
ro | ex-date open ÷ previous close |
rc | ex-date close ÷ previous close |
next | the next day's close ÷ previous close |
prev row … (Nd) | the previous day with a price, and how many days back it is |
If ro, rc and next are all close to the same number, the price really moved to a new level
and stayed there, which points to a corporate action. If they disagree, it was more likely
volatility.
4. The statuses
| Status | Who sets it | Applied to prices? | Meaning |
|---|---|---|---|
| AUTO | the system | ✅ yes | NSE's numbers were complete and the price gap on the chart matches. |
| REVIEW | the system | ❌ no | Something doesn't add up. A human must decide. |
| VERIFIED | you | ✅ yes | You checked the chart and the factor is right. |
| IGNORED | you | ❌ no | Not a real price-changing event. |
| SKIP | the system | ❌ no | Nothing to adjust (nil dividend, bonus in debentures...). |
Your VERIFIED and IGNORED decisions are permanent. The nightly sync never overwrites them. ↺ Back to review undoes your decision.
5. How to review a row, step by step
- On the Corporate Actions page, pick a filter (for example click the
ADJUST: N to reviewchip). - Click a row. The right panel shows the candle chart around the ex-date. The dashed line is the ex-date.
- Switch between Raw (no adjustment) and With this factor:
- Raw shows the jump as it is in the data.
- With this factor shows what charts will look like if you verify.
- Decide:
What "good" looks like (Verify): in Raw, the candles jump at the dashed line. With the factor, the candles before the line move up or down so the whole chart is one smooth series.
What "bad" looks like (fix or Ignore): with the factor there is still a step at the dashed line, or a new step appears that wasn't in Raw.
Keyboard shortcuts (after clicking a row): ← / → previous / next row, V verify and move on, I ignore and move on. Clicking outside the text boxes first makes the keys work.
6. The badges
gap Nd
The previous NSE price is N days before the ex-date. The "jump" compares today's price with a price from long ago. That usually means the stock was suspended, or it traded only on BSE in between (on 2026-04-20 many old stocks returned to NSE trading after 20+ years). The chart has no bars before the ex-date, so Raw and With this factor look identical and the chart can't confirm anything.
What to do: Ignore. The factor is meaningless. A developer can ignore all of these at once:
python -m nseeod.corp_actions ignore-gaps # dry run: shows the list
python -m nseeod.corp_actions ignore-gaps --apply # gap of 365 days or more
python -m nseeod.corp_actions ignore-gaps --min-days 90 --apply
Amber = more than 90 days, red = 365 days or more.
DELISTED
The stock no longer trades on NSE (last trade more than 45 days ago). See Delist.
7. FAQ
Does the ISIN stay the same when a company changes its name?
Usually yes. A pure name or symbol change keeps the ISIN, because the security is the same. A new
ISIN is issued when the security itself changes: a split or consolidation (face value change), a
merger, partly-paid shares becoming fully paid, or a capital reduction. So ISIN helps link renames
(scr_master.isin is updated every day), but it isn't a single key for a stock's whole history.
Why keep delisted stocks at all? To avoid survivorship bias. A backtest that only uses today's stocks leaves out every company that failed, so past returns look better than they really were. See Survivorship Bias.
I verified something by mistake. Is the data damaged? No. Prices are stored raw. Click ↺ Back to review (or Ignore) and the next chart read is correct again.
Why don't dividends show on normal charts?
Normal charts use adjust='price', which applies only splits, bonuses, rights and detected jumps.
Dividends are included only in total-return mode.
8. Glossary
| Term | Meaning |
|---|---|
| Ex-date | The first trading day the share trades without the benefit (bonus, dividend...). The price adjusts on this day. |
| Record date | The date the company checks who its shareholders are. Usually 0-1 trading days after the ex-date in India (T+1). |
| Face value (FV) | The nominal value printed on a share (₹1, ₹2, ₹10). Splits change it. It is not the market price. |
| Cum price | The last close before the ex-date. |
| Factor | The multiplier applied to every price before the ex-date. |
| TERP | Theoretical ex-rights price, the expected price after a rights issue. |
| Raw / adjusted | Raw = exactly as traded. Adjusted = old prices multiplied by the factors. |
For developers
- Code:
backend/nseeod/corp_actions.py(parse, reconcile, detector,accept-high,ignore-gaps), APIbackend/app/api/corp_actions.py, pagefrontend/src/pages/CorporateActions.jsx. - Reading adjusted prices:
eq_store.get_eod(symbol, adjust='price' | 'total', follow_renames=True). - Full technical reference: Corporate actions package docs.