Tracking how records change over time is essential when you manage historical data. Predecessor and successor mapping links each version of a data item to its previous and next version, letting you reconstruct the full history, see exactly when a value changed, and maintain a clear audit trail. This approach directly addresses understanding changes over time in historical data, making it the best fit for the task. The other points describe related issues but not the mechanism for tracking history. Sourcing historical data due to purge speaks to data availability rather than how changes are tracked; divergent price levels across instruments highlight data quality or normalization concerns; and omitting weekends and holidays affects statistical analysis and continuity, not the core method for managing historical records.

Price weighting means a stock’s influence on the index depends on its share price. The Dow Jones Industrial Average uses a price-weighted scheme, so higher-priced stocks move the index more than lower-priced ones, regardless of how large the company is. The S&P 500, by contrast, is weighted by market capitalization, so bigger firms by value have a greater say in its level. The Dow also consists of 30 stocks, not hundreds, and is not equally weighted. So describing the Dow as price weighted best captures how its movements are driven relative to the S&P 500.

Backtesting measures how a strategy would have performed by applying its rules to past market data within a modeling framework. By running the strategy on historical prices, you can estimate returns, risk, drawdowns, and other performance metrics, while keeping the rules fixed. It’s about learning from how the strategy would have behaved with real, historical data, not about predicting future prices from current conditions, not about monitoring live performance in real time, and not about generating synthetic data for stress testing. So the essence is evaluating a strategy’s hypothetical performance using historical data inside a model.

Time-series and calendar-aware data organization is what drives how financial market data should be stored. In this context, data exists as sequential records—prices, quotes, trades, and other market signals—each tied to a timestamp and an instrument. The storage design needs to support efficient time-based access, alignment across multiple assets, and correct handling of the market’s calendar (for example, five trading days in a week with weekends and holidays when markets are closed). This means using structures and indexing that make time-range queries fast, enabling consistent resampling and aggregation (such as converting tick data to minute or daily bars), and accommodating different granularities. It also requires accounting for gaps due to holidays or irregular trading hours, so the storage model can represent missing data cleanly and preserve the integrity of time-based analyses. Other options describe how data might be processed or used—such as collecting unstructured text data, running backtests, or calculating risk metrics—but they’re about analytics and tooling rather than the storage design itself.