The counting discussion acknowledges perceptual and judgment errors. In the Galton passage, it emphasizes writing down the measurement question before recording and separating cases to avoid confusion.
Keep the measurement stable
An observation should mean the same thing from the first row to the last. If “panic,” “rebound” or “strong day” changes meaning midway, the summary statistic mixes different events. Record a definition and a consistent rule before collecting the data.
Worked example
A “large down day” could mean a fall of at least 2% from the previous close. If later rows use an intraday decline instead, the categories are no longer comparable.
Case connection
Performance depends on the sample definition. Fix who enters the denominator before measuring success.
SPIVA: include the missing funds
Source-grounded facts
SPIVA evaluates the whole starting opportunity set, including funds that disappear, to address survivorship bias.
Context
Fund-performance comparisons can quietly change when unsuccessful funds close or merge. SPIVA’s published methodology addresses that problem by retaining the starting opportunity set.
Outcome
Including disappeared funds reduces survivorship bias. It does not make every benchmark choice perfect; it makes the population being evaluated more explicit and harder to improve by hindsight.
- The comparison identifies the eligible fund universe at the beginning of the evaluation period.
- It accounts for funds that disappear rather than comparing only those still present at the end.
- It also uses relevant benchmarks and reports results over specified horizons.
Case analysis
A denominator is a research decision. Define which funds or observations qualify at the start, then keep that definition stable. If eligibility changes because a result is embarrassing or unavailable, the percentage answers a different question. Record both the counting rule and the treatment of missing cases so another reader can reconstruct your calculation.
Try it
Define one event using an observable threshold, a time interval and an explicit unit.