Semtum

The lab

Every idea goes through the same checks before anything is believed, and almost none survive them. This page is rebuilt from the research record at every update.

Tested
946+7 in the last 10 rounds
Rejected
932+6 in the last 10 rounds
Waiting on forward data
13+1 in the last 10 rounds
Passed
1+0 in the last 10 rounds

Strategies tested, round by round

05001000Round 3Round 489

Touch or hover the chart to read any round.

strategies tested so fara round that produced a candidatethe round that produced the pass

A round is one question put to the data. Some test one rule, some sweep dozens of variants, which is why the line climbs in steps.

Where they ended up

932 rejected13 waiting on forward data1 passed

Families of ideas already answered

25 families of strategy, each tested enough times that another variant is unlikely to change the answer.

21 rejected2 not testable1 mixed1 does not work

The newest rounds

  1. Round 489FAIL: switching between SPY and T-bills on the Chicago Fed's adjusted financial-conditions index loses 5% a year to SPY after CXO's data endsNO VERDICT
  2. Round 487Survivor-only screen: buying firms after an activist cooperation agreement loses 2.1% over three months, and a fifth of them cannot be pricedSCREEN
  3. Round 486FAIL, wrong sign: buying a Korean stock after a sharp rise and an "undetermined" answer to the exchange's inquiry loses 5.4% in 10 sessionsNO VERDICT
  4. Round 485FAIL: Korean firms beat their market by 3.2% in the 60 sessions after a first Value-up disclosure, but the weeks are too few and too bunched to tell it from luckNO VERDICT
  5. Round 484Spin-off parents from the announcement, by EDGAR's Form 10 index: 38 dated announcements — stopped at its gate, no NSTOPPED
  6. Round 483Spin-off parents from the announcement: 34 announcements survive the hand check — stopped at its gate, no NSTOPPED
  7. Round 482Following hedge funds' futures positioning into commodity producers' stocks clears every declared test, on small miners and a strong 2020CANDIDATE
  8. Round 481FAIL: holding a country's ETF over the two weeks before its national election does no better than any two weeksNO VERDICT
  9. Round 480Survivor-only screen: shorting small firms after a new-debt 8-K (Item 2.03) loses 1.8% an event, and half the events cannot be pricedSCREEN
  10. Round 479FAIL: shorting a country's ETF the session after its team is knocked out of a tournament loses 0.34% a tradeNO VERDICT

Every round, with the reason for its verdict, is in the searchable ledger that comes with the course. See what is in it