Drawbacks of different biases (Not Behavioral)

Just for a quick revision. Have a couple of quick questions if anyone can help:

  1. Survivorship Bias : Returns are overstated (Only the returns of the historical surviving funds are shown) - Popular with Hedge Funds

  2. Stale Price Bias : Correlation (in absolute value) is understated whereas measured standard deviation can be lower/higher - Popular with Hedge Funds

  3. Backfill Bias : Returns are overstated (Past non-performing funds added on Manager’s discretion) - Popular with Hedge Funds

  4. Appraisal/Smoothed Data : Volatility and correlation (in absolute value) with rest of the portfolio is understated - Popular with Real Estate

  5. High Frequency Data : Correlation (in absolute value) with rest of the portfolio is understated.

  6. Vintage Year Effect Bias : Comparing deals closed in the same year for greater consistency can lead to bias if overweight is given on specific years with positive performance - Popular with Private Equity Funds

  7. Data Mining Bias : Repeatedly drilling information until an analyst finds some statistical significance just by chance - Popular Analyst Bias

My questions are:

  1. In which type of investment vehicles are the probabilities of High frequency data more common?

  2. I am a bit confused between Time Period Bias and Regime Change. Can someone explain the difference between the two? Can both of them take place simultaneously?

High frequency data , suffers from asynchronism , or lack of sync --> lower correlation. For example, daily stock movement will have lower correlation than monthly stock movement.

Regime change results from monetary policy changes ot fiscal policy changes. Time period bias is the data is chosen to reflect a specific time. time period may not reflect the regime change.

Thanks!

Regime changes results not only from monetary policy changes and fiscal policy changes than for any kind of political and other regime changes for long data series. E.g. Most today capitalistic Eastern European countries were socialistic 30 years ago, so using long series data for those countries result in non-stationary and may not be valid for forecasting.