Back to course

Embargoes and the Final Holdout

Purging removes training observations whose labels mechanically overlap the test window. What it cannot see is dependence that outlives the label: serial correlation in features and returns that connects a test label to training rows sitting just beyond the purged zone.

The embargo

Volatility clusters, order flow persists, features are built on rolling windows, so the training row that begins the day after a test block ends is still statistically tethered to it, purge or no purge. The embargo closes this by excluding a further buffer of training data after each test block: where the purge removes provable overlap, the embargo absorbs correlation that is real but not visible in the label windows.

embargo: also drop train i    if    ti∈(test end,  test end+e]\text{embargo: also drop train } i \;\; \text{if} \;\; t_i \in (\text{test end},\; \text{test end} + e]

The rest of this lesson is for subscribers

Unlock every lesson in Research Validation and Backtesting, and every other premium course.

Subscribe to continue

Test your knowledge

Questions are only available to subscribers.
Continue
NextHow Many Trials Stand Behind a Result

Keep reading Research Validation and Backtesting

21 lessons in this course, and every other premium course, on one subscription.

Subscribe now
  • Every lesson in every course, with the worked examples and interactive simulators
  • Graded questions on every lesson, with explanations for the wrong answers as well as the right one
  • The trainers, timed assessments and brainteaser library that go with them