Tuesday 11 March 2025
A new approach to cleaning option pricing data has been proposed, one that aims to identify and remove errors that can skew the accuracy of financial models. The technique, developed by a researcher at North-West University in South Africa, is specifically designed for use with option pricing datasets recorded on a single date.
Option pricing data is notoriously difficult to work with, as it’s often plagued by typos, incorrect values, and other errors. These mistakes can have serious consequences, leading to inaccurate predictions and poor investment decisions. To combat this problem, the researcher has developed a three-step process for identifying and removing problematic options from the dataset.
The first step involves checking each option price against theoretical bounds to ensure it doesn’t lead to arbitrage opportunities. In other words, if an option is priced too high or low compared to its underlying asset, it’s likely an error and should be removed. This step helps to eliminate obvious mistakes that could easily be exploited by traders.
The second step involves fitting a second-degree polynomial to the option prices grouped by their time to maturity. The residuals from this fit are then used to identify outliers, which are likely errors in the data. A confidence interval is calculated based on these residuals, and any options outside of this interval are flagged as potential errors. This step helps to catch more subtle mistakes that might not be immediately apparent.
The final step involves removing duplicated options, which can occur when multiple traders record the same option price. To handle this, the researcher uses the recorded open interest (the number of outstanding contracts) to determine which option is most likely to be accurate.
After applying these steps to six datasets of option prices, the resulting cleaned data showed a significant reduction in errors. The S&P 500 dataset, for example, was reduced from 576 options to 430 after cleaning, while the PowerShares dataset was reduced from 480 put options to 281. These results suggest that the new approach can be effective in improving the accuracy of option pricing models.
The implications of this work are significant, as accurate option pricing data is crucial for making informed investment decisions. By providing a reliable method for cleaning and validating option pricing data, researchers and traders can have greater confidence in their models and make more informed decisions. The researcher’s approach also has potential applications beyond finance, such as in other fields where data quality is critical.
Overall, this new technique offers a promising solution to the problem of errors in option pricing data.
Cite this article: “Cleaning Option Pricing Data: A New Approach to Improving Accuracy”, The Science Archive, 2025.
Option Pricing, Data Cleaning, Error Removal, Financial Modeling, Arbitrage Opportunities, Polynomial Fitting, Confidence Intervals, Open Interest, S&P 500, Powershares
Reference: Jaco Visagie, “A statistical technique for cleaning option price data” (2025).
