OL
Internal Research
Methodology Lab
Research Health First
Overview
Is the dataset healthy enough to trust, and is methodology selectivity adding value over structurally executable opportunities?
Unique Entry Events
42
market opportunities
Structural
42
executable events
Selective
17
methodology selected
Manual
11
human-selected reference
Snapshot Complete
96%
entry-time evidence
20S Mature
29
13 still immature
Dataset Health
must be trustworthy before performance
Frozen evidence coverage
88%
Compact evidence
9%
Refetched evidence
3%
Version completeness
100%
Identity / orphan errors
0
Selection Lift
selected vs not-selected structural events
20-session expectancy+0.34R
2R reach-rate lift+11.8 pts
Stop-rate difference−7.1 pts
Median MAE difference−0.18R
Illustrative only. Same entry event is never counted twice as two independent market opportunities.
Priority Experiments
pre-registered
52-week low floor: 25% vs 30%Counterfactual cohort: 25–<30% above low
Provisional
D2 minimum contraction swing2.0 / 2.5 / 3.0 / 4.0
Insufficient
D3 shrink tolerance1.00 / 1.05 / 1.10 / 1.15
Provisional
RS-line gatePASS vs otherwise-valid FAIL
Meaningful
Entry distance0–3 / 3–5 / 5–8 / >8%
Provisional
Research Signals
review, not conclusions
RS-line FAIL cohort has weaker 20-session expectancy
Observed difference meets project sample floor but remains univariate and regime-clustered.
25–30% above-low cohort does not yet justify raising the floor
Effect is small and the challenger population is still immature.
Entry Events + Experiments
Trades
Research-oriented trade view. Source, frozen decision state, fixed-horizon behavior, and realized outcome remain distinct.
| Ticker | Source | Setup | Q | T | Regime Used | Pivot Dist | Stop | 20S | Realized | Evidence |
|---|---|---|---|---|---|---|---|---|---|---|
| NVDA | Selective | VCP | A | A | B | +1.2% | 5.8% | +1.7R | Open | Frozen |
| PLTR | Structural | Flat Base | B | C | C | +0.8% | 7.4% | −0.9R | −1.0R | Frozen |
| CRNX | Manual | Power Play | C | B | B | +2.4% | 8.9% | +0.6R | +0.4R | Compact |
Parameter Validation
Experiments
Current rules against pre-registered challengers. Fixed-horizon outcomes lead; small samples remain explicitly labeled.
52-Week Low Floor
Current 25% vs 30% challenger
<25%n 24
25–<30%n 18
≥30%n 67
Current evidenceProvisional
D2 Minimum Swing
Replay from raw swing candidates
2.0%n 31
2.5% currentn 27
3.0%n 22
4.0%n 13
D3 Shrink Tolerance
Replay same structures, alternate tolerance
1.00n 12
1.05n 21
1.10 currentn 27
1.15n 33
Experiment Matrix
illustrative metrics| Experiment | Current | Challenger | Added / Removed | 20S Exp. | Stop Rate | 2R Rate | Sample | Evidence |
|---|---|---|---|---|---|---|---|---|
| 52w low floor | 25% | 30% | −18 events | +0.05R | −1.9 pts | +1.2 pts | Provisional | mixed |
| D2 | 2.5% | 3.0% | −5 structures | +0.12R | −4.0 pts | +3.1 pts | Provisional | frozen |
| D3 | 1.10 | 1.05 | −6 structures | +0.18R | −3.2 pts | +4.4 pts | Provisional | frozen |
| Breakout RVOL | 1.50× | 1.40× | +7 events | −0.09R | +5.1 pts | −2.7 pts | Insufficient | frozen |
False-Negative Lab
Candidates
Rejected and near-threshold names that matter for methodology validation. Tracking survives future universe dropout.
| Ticker | Observed | Setup | Rejected By | Exact Value | Current | Counterfactual | 20S Result | Maturity | Evidence |
|---|---|---|---|---|---|---|---|---|---|
| XYZ | Aug 10 | VCP | RS Line | FAIL | PASS required | Ignore RS-line | +2.1R | 20S | Frozen |
| ABC | Aug 12 | VCP | 52w Low | 27.4% | 25% | 30% | −0.8R | 20S | Compact |
| DEF | Aug 16 | Flat Base | EPS | 17.8% | ≥20% | ≥15% | — | Immature | Frozen |
| GHI | Aug 18 | VCP | Stop Ceiling | 10.8% | ≤10% | ≤12% | — | Immature | Frozen |
The <25% above-52w-low rejection cohort originates at weekly qualification and should retain that cadence label in real data.
Integrity Before Conclusions
Data Audit
Snapshot completeness, identity, versioning, evidence provenance, decision consistency, and forward-tracking health.
Complete Snapshots
96%
404 / 421
Missing Versions
0
all observations versioned
Duplicate Events
0
entry_event_id integrity
Orphan Trades
0
observation reference present
Regime Mismatches
14
used vs final
Compact Only
38
not full replay packets
Refetched
11
provider drift possible
Forward Truncated
2
explicit reason required
Decision Consistency
must remain zero
Selective without trade_plan_allowed0
Pilot auto-logged0
Structural with hard-gate fail0
Observation missing parameter_set_id0
Trade missing scanner version0
Evidence Gaps
investigate before analysis
Raw swing candidates absent7
Full RS-line series absent4
Earnings unknown3
Fundamentals partial9
Identity unresolved0