Bias

Question 1

BackTesting

Answer 1

The general method for seeing how well a strategy or model would have done after the fact.

Question 2

Bias In Trading

Answer 2

Several common biases can arise when using historical data for trading strategy development. They usually have biased results upwards.

Question 3

Survivorship Bias

Answer 3

Backtests are conducted on datasets that exclude data for companies/funds etc. that no longer exist (e.g. due to bankruptcy).
Such backtests are likely to have upwardly biased results, since the surviving entities will tend to have performed better than those that no longer exist.

Question 4

Data-Snooping Bias

Answer 4

Data is used more than once to develop a model (by trial and error).
Unsurprisingly, models that appear to do well historically can be found by exhaustive/extensive search.
Cross-validation (discussed earlier) can help to avoid data-snooping bias.

Question 5

Time-Period Bias

Answer 5

A test design is based on a time period that may make the results time-period specific.

Question 6

Look Ahead Bias

Answer 6

Using data you do not have yet.