Skip to main content
Track Time stop
14 / 15
Search the archiveLibrary
Library

2018issue C097

Auditing rank-rotation tests for time-stop and slot rules

Editorial: audit a rank-rotation evaluation by writing the time-stop, daily fill rule, and capital-slot budget as part of the mechanical-trading-system, then keep leftover test-harness sizing separate from those design choices.

  • Editorial: write the time-stop, daily fill rule, and capital-slot budget into the mechanical-trading-system so a rank-rotation test can be audited as one procedure.
  • The archive evaluation used a technical-rating entry gate of 5 or better and a three-month time-stop as the only exit, with no stop-loss or extra exit rules.
  • New names could enter every day until capital or a 500-name ceiling blocked them, and reentry-after-time-stop was allowed if the name still passed the screen.
  • The archive described 1000 per name as unsuitable for live use and suggested at least 5000. Editorial: keep that harness size separate from the designed capital-slot budget.
Entries in this reading3 entries
Current trackTime stop24 readingsRank rotation86 readingsMechanical trading system322 readings

Editorial: a designer can audit a rank-rotation evaluation by writing the time-stop, the daily fill rule, and the capital-slot budget as part of the mechanical-trading-system, then keeping leftover test-harness sizing out of the design.

The notes below restate the historical workflow. TradersWeek comments on how to inspect that workflow are labelled as editorial.

What the archive evaluation covered

The technical-rating work was evaluated as a mechanical signal test, not as a complete trading system with a full exit stack.

The only exit used in that evaluation was a three-month time-stop. No stop-loss or extra exit rules were added.

The tested entry gate bought names with a technical-rating of 5 or better.

A weekly-rebalance rule was specified only for an analysts-rating portfolio test and was not applied to the technical-rating procedure.

Write fill, slot, and reentry rules into the procedure

After a three-month time-stop exit, a name could be bought again and held for another three months if it still met the entry screen. That reentry-after-time-stop rule is part of the procedure, not an afterthought.

The backtest accepted new entries every day until capital ran out or a 500-name ceiling was reached, then replaced each expired slot with a new candidate.

Editorial: daily fills, the name ceiling, and replacement of expired slots are rank-rotation constraints. Record them with the entry gate and the time-stop so the test can be replayed as one mechanical-trading-system.

Keep harness sizing separate from the slot budget

The test started with 300000 in equity and 1000 per name, implying 300 initial positions, with the name count allowed to rise if equity rose. That mix of starting equity, per-name size, and a maximum name count is the capital-slot budget that decided how many ranked signals became positions.

Those 1000-per-name settings were described as unsuitable for live use because commissions would be excessive, and a per-name size of at least 5000 was suggested to reduce that drag.

Editorial: treat the 1000-per-name figure as leftover test-harness sizing. Keep it separate from the capital-slot budget you write as a design choice.

Holding period as a design choice

A stiffness-coefficient was identified as a useful persistence concept from the rating method.

Holding-period choice was discussed as a design issue tied to drawdown.

Editorial: the time-stop is a precommitted holding-period bound. Choosing that bound belongs in the mechanical-trading-system, not in the leftover harness.

Educational research material, not investment advice. Historical source context does not establish present-day performance.
14 of 15 in the Time stop track
201946-47 pp.Next on Time stopConfirmation-first pyramiding with price and time stopsThe method opened only part of the intended position and added size after confirmation, and only while the trade still moved as planned.
All readings on this track · 15 readings
  1. 1989Preflight checklist, time-stops, and stand-aside rules
  2. 1991Parameter windows for averages, option overlays and time stops
  3. 1995Three setup gates and a session-count clock complete a long bottom
  4. 1995Evaluating index futures OHLC reversals with RSI and time stops
  5. 1995A weekly two-close breakout and an eight-week time stop
  6. 2000Constructing breakout entry rules and time stops
  7. 2004Constructing action-reaction swing targets and time stops
  8. 2005Constructing breakout entries with trailing and time stops
  9. 2006Evaluating a restrictive range entry with a timed exit
  10. 2008Evaluating point-and-figure column breaks with time stops
  11. 2011Intraday gold swing with a daily entry limit, a time-stop, and a fixed stop-loss
  12. 2012Last-hour entries with a book-level loss bound and a next-open time stop
  13. 2015Precommit stop-loss, trailing-stop, and time-stop rules before entry
  14. 2018Auditing rank-rotation tests for time-stop and slot rules
  15. 2019Confirmation-first pyramiding with price and time stops
By fieldSystem design203 readingsBy year2018172 readings
All 24 readings tagged Time stop
Also on Time stop5 readings
1995Three setup gates and a session-count clock complete a long bottom2006Evaluating a restrictive range entry with a timed exit1985A serial-dependence window from signed price transitions1989Preflight checklist, time-stops, and stand-aside rules1990Calendar windows for three-wave pullbacks