1988issue C121-2
Reverse the rule before you trust the system
This archive article treats system work as a critique of one complete decision procedure. It specifies the reverse-signal, the abstention cases, and the loss-control path first, then asks whether serial-dependence in the return-sequence changes that same procedure.
- A recurring habit is to keep testing a procedure that loses while never treating the inverted signal as an equally serious test case.
- Loss-first-coverage places protective and abstention rules ahead of any search for winning trades, because protection is treated as theoretically required and later supports acting with confidence.
- Single-idea-focus, a foreign-strategy-trial, and persistence after a brief informal trial are required to finish one procedure, including its inverse.
- Related work still finds serial-dependence in returns treated as stationary and random, while examining the return-sequence improves volatility measurement only slightly.
The missing reverse-signal
A recurring development habit is to keep testing a procedure that loses while never treating the inverted signal as an equally serious test case. Decision research is described as misplaced when it tracks only the direction prices appear to be going and ignores the locations they fail to reach.
The reverse-signal is the inverted counterpart of an entry, exit, or abstention rule, tested as a separate case of the same procedure rather than ignored because the original rule lost.
Protection before the search for winning trades
Protective coverage is placed ahead of winning because it is presented as theoretically required and because completed protection is what later supports acting with confidence. Loss-first-coverage is that development order: protective and abstention rules are specified before any search for winning trades.
Finishing one procedure
Openness is defined as constructing a strategy foreign to the researcher’s current habits rather than remaining inside a familiar approach. A foreign-strategy-trial is that deliberate attempt to build and critique a decision process outside the existing habit set.
Focus means remaining with one primary idea long enough to specify its components, including the inverse of the expected rule, instead of leaving at the first alternative. Single-idea-focus is that stay with one primary rule set.
A newly sketched procedure is not treated as finished after a brief informal trial. Discipline and persistence are required after the first pass.
Compare, then follow the specified process
Integration asks why different approaches prove rewarding or unrewarding and what those differences add to an understanding of collective market behavior. An integration-check is that comparison.
After the specification work is done, the procedure is to be followed consistently without day-to-day modifications, using prepared knowledge to act instead of withdrawing from difficult decisions.
Serial-dependence as a later check
Related work from the same period treats stock-market returns as stationary and random while still finding some serial-dependence. Serial-dependence is ordered dependence among successive returns that can remain even when the series is otherwise treated as stationary and random.
The return-sequence is the order in which returns arrive, used as an input to volatility measurement rather than only the set of return magnitudes. Examining that order was said to improve volatility measurement only slightly, and that result is used as a reason to concentrate further research on price dependence.
All readings on this track · 15 readings
- 1987Evaluating weekly price cycles with autocorrelation cutoffs
- 1988Serial correlation is not an independence test
- 1988Event-triggered chi-square tests for jittered serial links
- 1988Reverse the rule before you trust the system
- 1990Chi-square evaluation of lagged breadth correlations
- 1990Two-gate out-of-sample tests for trading rules
- 1990Sign-cell correlation, chi-square checks, and lag shifts
- 1998Test serial dependence before you optimize a mechanical system
- 2003How a one-rank rotation is constructed for an offset autocorrelation check
- 2004Constructing first-order residual autocorrelation tests
- 2004A two-stage Durbin-Watson residual screen
- 2011Constructing trend labels from increment correlation atoms
- 2012Overnight versus daytime return evaluation
- 2012Constructing a multi-horizon folded-correlation reversal filter
- 2016Tuning indicators to measured market cycles