Today's price next to last year's financials: reading the time lag in financial data

Open a stock page and the current price sits right above ROE and the debt ratio. Being on the same screen, they look like numbers describing the company at the same instant. But the two run on different clocks.

The price changes by the second during trading hours; a trade that just went through shows up in the number right away. ROE and the debt ratio, by contrast, come from financial statements that the company compiles and releases once a quarter. The debt ratio on screen today does not reflect today's liabilities but the liabilities as of the last closing date, when the books were closed.

The gap is months, not days. And it is not a problem of stale data but a property of financial statements as a source. A company's financial condition is not the kind of information that is disclosed in real time.

First, the accounting period itself is a span. For a period ending in late March, the numbers bundle the activity from January through March into one result. They are a summary of the whole span, not a snapshot of the moment it ended, so any change during the period is already blended into the average.

Second, there is a preparation period between the closing and the disclosure. Closing the books, getting audited and filing the report take time. In Korea, quarterly and half-year reports are due within 45 days after the period ends and the annual report within 90 days. US-listed companies typically take around 40 days for a quarterly report and 60 to 90 days for an annual report, depending on company size. That is a month and a half at the shortest and three months at the longest.

Third, once disclosed, the data takes another few days to a few weeks to be organized by a data provider. The work involves fitting hundreds of companies' differing formats into the same line items.

Fourth, there is the collection cycle. Jupick refreshes its data at set times, but running a refresh does not make a report appear that does not exist yet. A refresh only fetches what is already public.

Add the four together, and the period the financial metrics on screen describe usually lies from two or three months to about half a year in the past.

There is one more problem, less known than the lag: a number, once published, can change later.

Financial statements are revised through correction filings. When an accounting method changes, items are reclassified during an audit or an error is found, numbers already released are rewritten. This is even more common in economic statistics, to the point that the Federal Reserve Bank of St. Louis runs a separate archive that keeps a full snapshot of each release. Last year's growth rate looked up today can differ from last year's growth rate looked up last year.

This distinction carries more weight than it seems. When judging whether a decision made at some past point was right, retracing it with the final revised figures on record now amounts to scoring it with information that did not exist at the time. People who work with data call this the point-in-time consistency problem.

Jupick does not keep past snapshots of its metrics. It does keep a separate collection of recent grade changes for stocks whose grade changed, but that is only a record of which letter the grade changed from and to. What ROE or the debt ratio was at that time is not kept.

So the metrics on this site are always “the value on record now,” and what was known at a particular past moment cannot be reconstructed. Jupick states this limit rather than hiding it.

Data carries at least two kinds of dates: the period the data describes, and the time Jupick's pipeline fetched that data.

Merging the two into one always produces a falsehood. Show only the update time, and last year's fourth-quarter numbers fetched early this morning look like today's numbers. Show only the accounting period, and there is no way to tell whether the value is still current or has sat unchanged for months.

So Jupick attaches the fiscal reference date to financial metrics and records the time a document was updated separately. The label goes only as far as the precision the data allows. For Korean stocks, the disclosure source gives the fiscal year but not the closing month, so the label stops at the fiscal year. Most companies close in December, but some close their books in March or June, so writing in a month would be an unverified claim.

For overseas stocks that come with quarterly data, Jupick states only the end point, as a year and a month. There is also a reason it does not write “based on a given quarter's results.” That wording would mean three months of figures added up, while the actual value strings together the latest twelve months, so it would not match the facts.

Even within one stock page, some items carry a reference date and some do not. For overseas stocks, some stability and dividend items cannot be received from the data source as they are and are filled in by another method, and for values filled that way there is no basis for saying when they are from. So Jupick puts no reference date on those items at all. Leaving it blank seemed better than writing a date that is only half right.

First, metrics jump suddenly at quarter boundaries. On the day a new report is reflected, revenue growth or operating margin can move sharply overnight. Nothing happened to the company; the reference span simply shifted by one quarter. The grade sometimes changes along with it.

Second, update timing differs by stock. Some companies close their year in a month other than December, and Korean and overseas stocks fall under different disclosure systems altogether. When two stocks are compared side by side, it does happen that one rests on data from three months ago and the other on data from six months ago.

Third, sudden events are not in the numbers yet. Even when something reshapes the financial structure entirely, such as a large acquisition, a rights offering or the divestiture of a business unit, the metrics keep describing the company as it was before the event until a report that reflects it comes out. The news arrives first and the numbers arrive later.

So treating the financial metrics alone as the full picture of where a company stands today is a mistake. A metric summarizes one past span, and what happened after that span is not in the number. The reference date on screen is the clue to how large that gap is.

A reference date does not mean every item for that date was finalized in the same way. An audited annual report and an unaudited quarterly report go through different levels of verification. The same label can sit on data of different weight.

An old reference date is not a bad sign in itself. Companies with a late closing month update later by structure. Filtering out stocks with old dates would end up filtering out every company with a particular closing month.

And nothing in this article is about when to trade anything. Advice on what to do ahead of an earnings release is outside what Jupick does. Jupick does not forecast future earnings or prices, and goes only as far as stating honestly when the already public data is from.

This content is for information only and is not investment advice. Investment decisions and responsibility for them rest with the user.