Using Restated Financial Data in Backtests Can Silently Inflate Strategy Returns

Financial statements exist in two forms: the original figures disclosed at filing and restated figures revised later through amendments or reclassifications. When building historical trading strategy backtests, using restated data against past dates introduces a form of look-ahead bias, since that revised information was not available to market participants at the time. A dataset from Tradevo Data illustrates the scale of the issue, flagging 18,772 restatements across over 314,000 point-in-time data rows. The correct approach, known as point-in-time backtesting, requires using only values whose original filing date falls on or before the backtest date. Choosing between as-reported and restated figures is not a statistical fix but a deliberate data design decision that can meaningfully affect the validity of a strategy's historical performance.