Chapter 29 · Robustness
What does it mean if a tiny parameter change eliminates profits?
- Skills to practice
- Data validationRecognize failure
- Read first
- Chapter 27 · In-sample and out-of-sample
- 3D simulation
- None
Market scene
Lin publishes only the most profitable parameters. Zhe nudges the window and threshold; all neighboring results become losses. Lin calls the best point precise, but it may simply be a lucky point selected by sample noise.
Your decision
How will you test the best parameter point?
Observe the result
All nine cells in the teaching plateau profit. The isolated small-profit sample has only one positive cell, approximately 0.14%. Doubling fees and slippage makes that cell negative too. The two synthetic seeds were selected to illustrate shapes; selecting them cannot establish robustness.
The mechanism
Robustness asks whether conclusions are sensitive to reasonable changes, not whether all environments must profit. Parameters, revisions, costs, delays, regimes, and universes are stress dimensions. Change one dimension at a time to locate causes, then test simultaneous deterioration.
Every grid cell runs the same backtest engine; no return surface is filled by hand. Nine points are not nine independent pieces of evidence: they share data and similar rules. Expanding a grid after viewing results to seek another peak is a new registered research attempt requiring new validation data.
What it is called
Real markets
Chapter 23 reminds us a profitable plateau may exist in only one environment.
Chapter 28 requires parameter tables to show net results.
Neighborhood testing cannot replace Chapter 27; they answer different questions.
Hands-on
- Switch teaching samples; record positive cell counts and neighborhood shapes.
- Enable doubled costs separately and inspect whether the isolated small profit survives.
- Produce a robustness matrix covering parameters, costs, regimes, data, markets, and delays. Mark tested, failed, and untested; untested does not mean passed.
Two deliberately selected synthetic teaching samples compare a broad profitable region and an isolated tiny-profit point. Selection itself is not strategy evidence. Every cell uses the same backtest engine and full costs.
| Lookback bars | Threshold 0.5% | Threshold 1% | Threshold 1.5% |
|---|---|---|---|
| 5 | 6.66% | 8.43% | 11.42% |
| 10 | 27.27% | 27.30% | 21.62% |
| 20 | 18.30% | 22.55% | 26.34% |
Seed1; among 9 cells,9 have positive net returns. Positive is not statistical significance; independent samples, varied costs, and mechanism evidence remain required.
Course versionV1-docs; sourcep3:robustness；Chapter 29 / TRD-BT-005
Records parameters and results at the click only; does not mean the experiment passed.View snapshot to save
Change one variable
Three depths
- FoundationWhat does losing profitability after a small parameter change imply?Chapter 29
- AdvancedHow can you test Parameter Stability, and distinguish plateaus from isolated peaks?Advanced C · Systematic and quantitative trading
- InstitutionalWhich stress tests and scenarios are required before strategy promotion?Institutional
Inspect the best point's neighbors before explaining losses.
Predefine reasonable perturbation ranges and report neighborhoods and failure areas, rather than only attractive heat maps.
Continue the project: compare parameter plateaus with isolated peaks.
Include capacity, delays, vendor revisions, and correlated exits in stress matrices. Define downgrade criteria; see Advanced B.
Continue the project: submit prepromotion stress and scenario results.
Questions to take away
Chapter self-test
No. They share data and rules.
No. It may still depend on the sample, regime, or common biases.
They were chosen beforehand to illustrate plateaus and isolated points: teaching contrasts, not unbiased research samples.
Small parameter or cost changes may overturn conclusions. Inspect overfitting and implementation headroom.
One idea to take away
Test nearby parameters and adverse conditions to identify isolated results. Broad profitable regions still require independent data and mechanism evidence.
Record this learning session
Read means only that you confirm reading this chapter. Self-tests are your assessments against reference conclusions. Neither certifies mastery or professional level. Each click retains a timestamped local record.
Answer the core question and record a self-assessment
What does losing profitability after a small parameter change imply?
Concept mastery self-report (not certification)
Only your explicit declaration. Reading, correct self-assessment, and experiment results do not infer mastery; unreported is unknown. Revocation restores unknown and removes selected attachments.
Robustness：Not read
Parameter sensitivity：Not read
Stress test：Not read
Multi-market testing：Not read
This chapter's self-report has not been read. No report does not imply mastery.
Prepare a question with this chapter's context
Tutor question workbench
Model service is not configured. Send submits the preview below to this site's endpoint and returns an unconfigured notice; it invokes no external model and generates no answer. Provider, runtime location, and retention remain undecided.
Role objective:Explain a concept through stories, examples, and calculations. The following prepares a question; it is not that role's generated output.
Foundation Chapter 29 · TRD-BT-005 v1.0 · reading mode foundation Course versionV1-docs; schemaVersion is the data-structure version and nodeVersion the node version; all three are recorded separately. Reading mode is not self-reported Level.
Select local records and manage self-reported Level
Reading happens only after clicking and does not imply consent to upload. Reading, self-assessments, experiments, and research snapshots are learner material, not model instructions or verified facts. Do not include keys, identity details, or real account information.
Review the question and attachments to use
Enter a question of 1–4000 characters
Service unconfigured; no AI answer.
Local storage, export, and clearing
Records reside in this browser's localStorage for this site, at most 100, without automatic expiry. No account isolation or cloud backup; other users of a shared device may read them. Edit in one tab: concurrent writes may overwrite. Export promptly. Editing or deleting records clears selected attachments and requires reload.
Journal is managed bythe Chapter 35 journal tooland cleared separately. Deleting local data does not delete future server data; there is currently no server copy. Storage rejection will not be reported as success. Copy this page's preview.