Restated history leaks backward
Demergers, discontinued operations and accounting changes can replace the version of prior results that investors originally received.
Historical research infrastructure
Use the history investors actually saw.
Altys stores Indian financial information with both the period it describes and the date it became knowable. Historical screens, factor tests and research can therefore use the contemporary version of the record rather than today’s restated hindsight.
Point-in-timeNo look-aheadRestatements preservedSource lineage
The operating problem
More information does not create continuous attention.
A clean ten-year table downloaded today is useful for understanding the current historical record. It is unsafe for testing what an investor could have known in the past.
Availability is confused with period end
A March quarter is not known in March. Using the period date as the knowledge date gives the model information before publication.
Corporate-action bases can split
Historical prices, shares, EPS and valuation multiples become inconsistent when adjustments are applied to only part of the calculation.
The Altys workflow
Research, decision and monitoring on one evidence trail.
Point-in-time integrity is a data contract that every layer must preserve.
01
Retain each vintage
Keep the originally published value and every subsequent revision rather than overwriting the old record.
02
Track availability
Store when each input became public and use the latest availability date when a metric depends on several inputs.
03
Respect reporting basis
Keep consolidated and standalone histories explicit and avoid mixing bases within a calculation window.
04
Pair price and fundamentals
Apply corporate actions consistently across price, share count, EPS and the valuation series built from them.
05
Fail honestly
Return unavailable when a historically valid input does not exist instead of filling absence with zero or a later value.
What the system adds
Infrastructure around the analyst, not a black box above them.
A backtest should not know the revised answer before the market did. Point-in-time data is what makes that promise testable.
Two-date modelEvery observation knows both the financial period and when it became usable in research.
Vintage preservationLater restatements remain available without erasing the record investors originally saw.
Calculated metricsDerived fields use explicit, code-owned formulas and carry the availability of their underlying inputs.
Audit trailA historical value can be traced to its source, basis, filing date and calculation method.
Questions, answered
What teams usually ask before a pilot.
What is point-in-time financial data?
Point-in-time financial data preserves the version of each value that was available on a given historical date. It distinguishes the period being described from the date the information became public.
Why is the latest historical series unsafe for backtesting?
The latest series may include restatements, corrections and business reclassifications published after the historical decision date. Using those values can introduce look-ahead bias.
Does point-in-time data mean using uncorrected errors forever?
No. It means preserving every vintage. Current analysis can use the corrected record, while historical analysis can retrieve the version that was knowable at the time.
Who needs point-in-time data?
Systematic funds, quant researchers, PMS and AIF teams, factor investors and anyone testing historical screens or reconstructing past investment decisions need point-in-time discipline.
Private preview
Research once. Let Altys keep watching what changes.
Bring a real company universe and a real monitoring problem. We will show how Altys can connect the data, research workflow and alerts around the way your team already invests.