Improved Market Data Acquisition Stability
Published on September 8, 2026
Market data can now be acquired even with minor OHLC inconsistencies
FactDecode checks the consistency of OHLC values—open, high, low, and close—in market data acquired from Twelve Data.
Previously, even very small inconsistencies in the source data could cause the entire data acquisition process to fail.
With this update, inconsistencies that fall within a defined tolerance are treated as minor. The dataset remains available for use, with a warning indicating that an inconsistency was detected.
This reduces cases where an entire dataset becomes unavailable because of small inconsistencies that have limited impact on actual analysis.
Acquired values are retained without correction
Even when an inconsistency is accepted as minor, FactDecode does not rewrite OHLC values or remove the affected row.
The values returned by the data provider are retained as received, while the presence of a minor inconsistency is clearly shown as a warning.
This allows users to continue using the data for analysis while preserving the original source values.
Clear inconsistencies will continue to stop data acquisition
Not all OHLC inconsistencies are accepted.
Data with clearly unnatural price relationships, or inconsistencies that exceed the defined tolerance, will continue to cause data acquisition to stop.
By allowing minor irregularities to remain available with a warning while continuing to reject clear data anomalies, FactDecode aims to maintain both acquisition stability and data integrity.
Impact on existing data and previous analyses
This update improves the validation process used when market data is newly acquired or refreshed.
It does not rewrite datasets that have already been stored or alter the values of previous analysis results.
For future acquisitions, when a minor OHLC inconsistency is detected, users can review the warning in Data Detail and continue using the dataset.
FactDecode will continue to distinguish between source values and data-quality notices rather than mechanically correcting external data, helping preserve a verifiable record of the data used for analysis.