Data and analytics
Data and analytics tests data quality, reference data, time-series handling, and the calculations used to turn raw observations into decision inputs. Expect questions on data lineage, point-in-time integrity, and performance measurement.
Data foundations
- Reference and master data identify instruments, counterparties, calendars, and prices; errors propagate into risk, settlement, and reporting.
- Point-in-time data prevents look-ahead bias. Survivorship bias arises when failed or delisted entities are excluded from historical samples.
- Data lineage maps source, transformation, and consumption so a reported number can be reproduced and audited.
Analytics
Returns can be simple (P1/P0 − 1) or logarithmic (ln(P1/P0)); VWAP weights traded prices by volume. Validate models with out-of-sample data and monitor drift.
Worked example
Trades: 100 shares at 10.00, 100 at 10.10, and 300 at 10.20. Total value = 1,000 + 1,010 + 3,060 = 5,070; total volume = 500; VWAP = 5,070/500 = 10.14. The simple average of the three prices is 10.10, which underweights the large 10.20 trade. Using the simple average would misstate the execution benchmark.
Traps
Mixing trade date and settlement date in position data; using unadjusted prices across a split; ignoring stale or duplicate ticks; confusing correlation with causation; reporting an average without its weighting scheme.