Current Gate
Gate Rules
RSI-Net EBITDA
thr 235.970 higher · cov 49% · dist 80
KEEP68★★★★★
FCF Efficiency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
ROIC-EBITDA QC
thr 163.660 higher · cov 44% · dist 41
REVIEW20★★★☆☆
Avg EPS Growth
thr 32.770 higher · cov 71% · dist 55
REVIEW52★★★☆☆
FCF Quality
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
11 / 127 winners caught at ≥4/5
9% retained
Pass-Set Quality (precision)
Avg Return
+51.65%
return_pct
Signal Independence
A
overlap grade
Low overlap — 5 distinct signal families: RSI-Net EBITDA, FCF Efficiency, ROIC-EBITDA QC, Avg EPS Growth, FCF Quality. No direct input redundancy.
Top Picks Under This Gate
Quick Swaps — 1 rule changed
+7.8 vs current
Gate Rules
RSI-Net EBITDA
thr 235.970 higher · cov 49% · dist 80
KEEP68★★★★★
FCF Efficiency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
ROIC/Ebitda Spread
thr 19.870 higher · cov 49% · dist 60
CANDIDATE91★★★★☆
Avg EPS Growth
thr 32.770 higher · cov 71% · dist 55
REVIEW52★★★☆☆
FCF Quality
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
11 / 127 winners caught at ≥4/5
9% retained
Pass-Set Quality (precision)
Avg Return
+49.55%
return_pct
Signal Independence
?
overlap grade
Phase B auto-generated. Check overlap_pairs at runtime.
Upside / Downside
Restrictiveness
Similar to current — single rule change
Liberal Risk
Low — only one signal family updated
Pros / Cons
- Minimal change: 1 rule swap. ROIC/Ebitda Spread has sil=60 among non-gate L2 candidates.
- Auto-generated — validate composite score + overlap before adopting.
Top Picks Under This Gate
+6.8 vs current
Gate Rules
RSI-Net EBITDA
thr 235.970 higher · cov 49% · dist 80
KEEP68★★★★★
FCF Efficiency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
ROIC L1
thr -16.780 lower · cov 0% · dist 69
ADD★★★★☆
Avg EPS Growth
thr 32.770 higher · cov 71% · dist 55
REVIEW52★★★☆☆
FCF Quality
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
16 / 127 winners caught at ≥4/5
13% retained
Pass-Set Quality (precision)
Avg Return
+42.88%
return_pct
Signal Independence
?
overlap grade
Phase B auto-generated. Check overlap_pairs at runtime.
Upside / Downside
Restrictiveness
Similar to current — single rule change
Liberal Risk
Low — only one signal family updated
Pros / Cons
- Minimal change: 1 rule swap. ROIC has sil=69 among non-gate L2 candidates.
- Auto-generated — validate composite score + overlap before adopting.
Top Picks Under This Gate
+6.0 vs current
Gate Rules
RSI-Net EBITDA
thr 235.970 higher · cov 49% · dist 80
KEEP68★★★★★
FCF Efficiency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
ROE L1
thr -26.390 lower · cov 0% · dist 61
CANDIDATE★★★★☆
Avg EPS Growth
thr 32.770 higher · cov 71% · dist 55
REVIEW52★★★☆☆
FCF Quality
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
14 / 127 winners caught at ≥4/5
11% retained
Pass-Set Quality (precision)
Avg Return
+42.84%
return_pct
Signal Independence
?
overlap grade
Phase B auto-generated. Check overlap_pairs at runtime.
Upside / Downside
Restrictiveness
Similar to current — single rule change
Liberal Risk
Low — only one signal family updated
Pros / Cons
- Minimal change: 1 rule swap. ROE has sil=61 among non-gate L2 candidates.
- Auto-generated — validate composite score + overlap before adopting.
Top Picks Under This Gate
+5.7 vs current
Gate Rules
RSI-Net EBITDA
thr 235.970 higher · cov 49% · dist 80
KEEP68★★★★★
FCF Efficiency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
P/C L1
thr 3.270 lower · cov 0% · dist 60
CANDIDATE38★★★★☆
Avg EPS Growth
thr 32.770 higher · cov 71% · dist 55
REVIEW52★★★☆☆
FCF Quality
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
13 / 127 winners caught at ≥4/5
10% retained
Pass-Set Quality (precision)
Avg Return
+41.35%
return_pct
Signal Independence
?
overlap grade
Phase B auto-generated. Check overlap_pairs at runtime.
Upside / Downside
Restrictiveness
Similar to current — single rule change
Liberal Risk
Low — only one signal family updated
Pros / Cons
- Minimal change: 1 rule swap. P/C has sil=60 among non-gate L2 candidates.
- Auto-generated — validate composite score + overlap before adopting.
Top Picks Under This Gate
+5.7 vs current
Gate Rules
RSI-Net EBITDA
thr 235.970 higher · cov 49% · dist 80
KEEP68★★★★★
FCF Efficiency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
ROA L1
thr -12.840 lower · cov 0% · dist 52
CANDIDATE20★★★☆☆
Avg EPS Growth
thr 32.770 higher · cov 71% · dist 55
REVIEW52★★★☆☆
FCF Quality
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
14 / 127 winners caught at ≥4/5
11% retained
Pass-Set Quality (precision)
Avg Return
+42.84%
return_pct
Signal Independence
?
overlap grade
Phase B auto-generated. Check overlap_pairs at runtime.
Upside / Downside
Restrictiveness
Similar to current — single rule change
Liberal Risk
Low — only one signal family updated
Pros / Cons
- Minimal change: 1 rule swap. ROA has sil=52 among non-gate L2 candidates.
- Auto-generated — validate composite score + overlap before adopting.
Top Picks Under This Gate
+5.5 vs current
Gate Rules
RSI-Net EBITDA
thr 235.970 higher · cov 49% · dist 80
KEEP68★★★★★
FCF Efficiency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
Rev Growth Efficiency
thr 855.600 higher · cov 47% · dist 66
CANDIDATE38★★★★☆
Avg EPS Growth
thr 32.770 higher · cov 71% · dist 55
REVIEW52★★★☆☆
FCF Quality
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
11 / 127 winners caught at ≥4/5
9% retained
Pass-Set Quality (precision)
Avg Return
+55.34%
return_pct
Signal Independence
?
overlap grade
Phase B auto-generated. Check overlap_pairs at runtime.
Upside / Downside
Restrictiveness
Similar to current — single rule change
Liberal Risk
Low — only one signal family updated
Pros / Cons
- Minimal change: 1 rule swap. Rev Growth Efficiency has sil=66 among non-gate L2 candidates.
- Auto-generated — validate composite score + overlap before adopting.
Top Picks Under This Gate
+3.5 vs current
Gate Rules
RSI-Net EBITDA
thr 235.970 higher · cov 49% · dist 80
KEEP68★★★★★
FCF Efficiency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
ROIC-EBITDA QC
thr 163.660 higher · cov 44% · dist 41
REVIEW20★★★☆☆
Quality Value Score
thr 0.060 lower · cov 22% · dist 78
ADD78★★★★☆
FCF Quality
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
11 / 127 winners caught at ≥4/5
9% retained
Pass-Set Quality (precision)
Avg Return
+47.14%
return_pct
Signal Independence
?
overlap grade
Phase B auto-generated. Check overlap_pairs at runtime.
Upside / Downside
Restrictiveness
Similar to current — single rule change
Liberal Risk
Low — only one signal family updated
Pros / Cons
- Minimal change: 1 rule swap. Quality Value Score has sil=78 among non-gate L2 candidates.
- Auto-generated — validate composite score + overlap before adopting.
Top Picks Under This Gate
+2.9 vs current
Gate Rules
RSI-Net EBITDA
thr 235.970 higher · cov 49% · dist 80
KEEP68★★★★★
FCF Efficiency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
ROIC-EBITDA QC
thr 163.660 higher · cov 44% · dist 41
REVIEW20★★★☆☆
Quality Value Score (Test Version)
thr 1.090 lower · cov 48% · dist 58
CANDIDATE50★★★☆☆
FCF Quality
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
12 / 127 winners caught at ≥4/5
9% retained
Pass-Set Quality (precision)
Avg Return
+43.35%
return_pct
Signal Independence
?
overlap grade
Phase B auto-generated. Check overlap_pairs at runtime.
Upside / Downside
Restrictiveness
Similar to current — single rule change
Liberal Risk
Low — only one signal family updated
Pros / Cons
- Minimal change: 1 rule swap. Quality Value Score (Test Version) has sil=58 among non-gate L2 candidates.
- Auto-generated — validate composite score + overlap before adopting.
Top Picks Under This Gate
-11.9 vs current
Gate Rules
RSI-Net EBITDA
thr 235.970 higher · cov 49% · dist 80
KEEP68★★★★★
FCF Efficiency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
Quality Value Score
thr 0.060 lower · cov 22% · dist 78
ADD78★★★★☆
Avg EPS Growth
thr 32.770 higher · cov 71% · dist 55
REVIEW52★★★☆☆
FCF Quality
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
20 / 127 winners caught at ≥4/5
16% retained
Pass-Set Quality (precision)
Avg Return
+33.41%
return_pct
Signal Independence
A
overlap grade
Overlap-safe selection — Quality Value Score has no HIGH conflicts with any kept rule in Current Gate.
Upside / Downside
Restrictiveness
Incremental — one swap beyond Current Gate
Liberal Risk
Low — single additional metric change; preserves rest of base gate
Pros / Cons
- Builds on Current Gate — one additional refinement (minimal-change optimization path)
- Quality Value Score (sil=78, ADD) replaces ROIC-EBITDA QC (REVIEW) — overlap-safe
- No HIGH overlap detected — all 5 metrics from distinct signal families
Top Picks Under This Gate
Broader Alternatives — 2+ rules changed
-13.9 vs current
317 tickers pass all rules — gate has no discriminating power at these thresholds.
Gate Rules
Quality Value Score
thr 0.060 lower · cov 22% · dist 78
ADD78★★★★☆
FCF Efficency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
Weighted FCF Efficiency
thr 1.800 lower · cov 64% · dist 71
CANDIDATE58★★★★☆
P/S v P/C Multiple
thr 13.480 lower · cov 41% · dist 52
CANDIDATE38★★★☆☆
ROIC-FCF Quality Gate
thr 0.130 lower · cov 40% · dist 33
SKIP58★★☆☆☆
Control Group Validation
Ctrl Retention (recall)
42 / 127 winners caught at ≥4/5
33% retained
Pass-Set Quality (precision)
Avg Return
+19.44%
return_pct
Signal Independence
B
overlap grade
No HIGH overlaps. FCF family fills 2 of 5 slots — gate has 4 effective signal dimensions, not 5. One DIR_LOCK on ROIC-FCF Gate.
Upside / Downside
Restrictiveness
Likely more restrictive — two high-signal FCF gates vs one
Liberal Risk
Medium — FCF signals correlated, gate may let through FCF-fakers simultaneously
Pros / Cons
- Eliminates Watson 5 direction controversy entirely (21 consecutive opposing)
- FCF Efficency (A, MPS 94.47) is the top-ranked metric by MPS — strongest available signal
- All rules have confirmed lower-wins directions (except ROIC-FCF Gate flag)
- Highest average MPS of FCF-focused alternatives
- 3 of 5 rules are FCF-family metrics — gate loses diversity of signal type
- FCF Efficency and Wtd FCF Efficiency share thematic overlap (both FCF efficiency ratios)
- ROIC-FCF Gate direction lock conflict persists
- Coverage gap: FCF Efficency only 47.1% of tickers have valid data
- If FCF signals are collectively wrong in a cycle, gate has little non-FCF backstop
Top Picks Under This Gate
-15.2 vs current
459 tickers pass all rules — gate has no discriminating power at these thresholds.
Gate Rules
Quality Value Score
thr 0.060 lower · cov 22% · dist 78
ADD78★★★★☆
FCF Efficency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
FCF-EBITDA Convergence Score
thr 2.140 lower · cov 33% · dist 47
CANDIDATE52★★★☆☆
P/S v P/C Multiple
thr 13.480 lower · cov 41% · dist 52
CANDIDATE38★★★☆☆
FCF Quality Score
thr 0.320 lower · cov 33% · dist 70
KEEP52★★★★☆
Control Group Validation
Ctrl Retention (recall)
31 / 127 winners caught at ≥4/5
24% retained
Pass-Set Quality (precision)
Avg Return
+19.97%
return_pct
Signal Independence
B
overlap grade
FCF cluster exists (FCF Eff + FCF-EBITDA Conv + FCF Quality Score) but no direct formula overlap. P/S v P/C and QVS remain independent signals.
Upside / Downside
Restrictiveness
Similar to current — FCF Quality Score adds a real FCF quality filter
Liberal Risk
Low-Medium — FCF Quality Score threshold may be strict
Pros / Cons
- Removes weakest gate metric by SIL score (Watson-ROIC CI, 37/100)
- FCF Quality Score is BF #1 combo partner — strongest empirical backing
- FCF Quality Score direction confirmed lower (ctrl validated, 53 sessions)
- No HIGH overlap with remaining gate metrics
- P/S v P/C retained — provides independent valuation signal
- Gate becomes more FCF-concentrated (FCF Eff + FCF-EBITDA Conv + FCF Quality Score)
- FCF Quality Score SIL score (55) lower than FCF-EBITDA Conv (63) — mid-tier confidence
- P/S v P/C still SWAP_OUT candidate (sil=24) — future swap needed
Top Picks Under This Gate
-16.3 vs current
Gate Rules
Quality Value Score
thr 0.060 lower · cov 22% · dist 78
ADD78★★★★☆
P/S v P/C Multiple
thr 13.480 lower · cov 41% · dist 52
CANDIDATE38★★★☆☆
Weighted FCF Efficiency
thr 1.800 lower · cov 64% · dist 71
CANDIDATE58★★★★☆
Avg. EPS Growth
thr 32.770 higher · cov 71% · dist 55
REVIEW52★★★☆☆
CARC
thr 0.360 lower · cov 72% · dist 44
CANDIDATE20★★★☆☆
Control Group Validation
Ctrl Retention (recall)
38 / 127 winners caught at ≥4/5
30% retained
Pass-Set Quality (precision)
Avg Return
+20.14%
return_pct
Signal Independence
A-
overlap grade
5 distinct signal families — best family diversity of all alternatives. No HIGH overlaps. Minor direction lock on CARC. Each rule covers a genuinely different dimension.
Upside / Downside
Restrictiveness
Less restrictive — lower-grade metrics have wider thresholds
Liberal Risk
Medium-High — CARC direction uncertainty + EPS Growth gaming risk
Pros / Cons
- Highest overall ticker coverage — all 5 rules ≥65% (vs Watson 5 at 42.6%, PEG at 30.7%)
- CARC has 76.3% coverage — second only to P/S v P/C (83.7%)
- More tickers get scored, reducing blind spots from N/A data
- Avg EPS Growth (71.0% coverage) now upgraded to B-grade (MPS 66.72) — adds meaningful growth dimension
- CARC dropped to D-grade (MPS 32.30) — significant weakening since last assessment
- CARC direction lock conflict (locked=lower, SM CSV raw=higher) — flagged INVESTIGATE in SIL
- Avg EPS Growth only D ranked by return score — directional confidence mixed
- No pure FCF signal — gate may miss quality FCF compounders
- Growth (EPS Growth) + Quality (QVS) could be gamed by one-time earnings spikes
Top Picks Under This Gate
-17.6 vs current
Gate Rules
Weighted FCF Efficiency
thr 1.800 lower · cov 64% · dist 71
CANDIDATE58★★★★☆
SM Watson 5
thr 51.120 lower · cov 42% · dist 42
SKIP52★★★☆☆
Quality Value Score
thr 0.060 lower · cov 22% · dist 78
ADD78★★★★☆
P/S v P/C Multiple
thr 13.480 lower · cov 41% · dist 52
CANDIDATE38★★★☆☆
ROIC-FCF Quality Gate
thr 0.130 lower · cov 40% · dist 33
SKIP58★★☆☆☆
Control Group Validation
Ctrl Retention (recall)
47 / 127 winners caught at ≥4/5
37% retained
Pass-Set Quality (precision)
Avg Return
+20.73%
return_pct
Signal Independence
C
overlap grade
HIGH overlap persists — Wtd FCF Eff (col42) is a direct input inside Watson 5 (col53). Two rules double-counting the same FCF signal. Direction flags on 2 of 5 rules.
Upside / Downside
Restrictiveness
Similar to current — Watson 5 + QVS combination is strict
Liberal Risk
Low — Watson 5 threshold (51.94) acts as broad filter
Pros / Cons
- Surgical change — only 1 rule swapped, all other gate knowledge preserved
- ROIC-FCF Quality Gate (A, MPS 81) replaces PEG Adj Hybrid (F, MPS 23)
- Average gate MPS jumps from 68.6 → 74.3
- SM Watson 5 direction AT_RISK (21 consecutive opposing) remains unresolved
- HIGH overlap between Wtd FCF Eff and Watson 5 persists (col42 inside Watson 5)
- ROIC-FCF Gate has a direction lock conflict (locked=lower, SM CSV raw=higher)
- Effective signal independence still ~3-4 of 5 rules due to Watson 5 absorption
Top Picks Under This Gate
-12.5 vs current
338 tickers pass all rules — gate has no discriminating power at these thresholds.
Gate Rules
Quality Value Score
thr 0.060 lower · cov 22% · dist 78
ADD78★★★★☆
FCF Efficency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
FCF-EBITDA Convergence Score
thr 2.140 lower · cov 33% · dist 47
CANDIDATE52★★★☆☆
P/S v P/C Multiple
thr 13.480 lower · cov 41% · dist 52
CANDIDATE38★★★☆☆
ROIC-FCF Quality Gate
thr 0.130 lower · cov 40% · dist 33
SKIP58★★☆☆☆
Control Group Validation
Ctrl Retention (recall)
52 / 127 winners caught at ≥4/5
41% retained
Pass-Set Quality (precision)
Avg Return
+20.32%
return_pct
Signal Independence
B+
overlap grade
No HIGH overlaps — Wtd FCF Eff removed, eliminating the main conflict. FCF family still fills 2 of 5 slots. LOW overlap between col31 and col54 (share EV/EBITDA as indirect input only). One DIR_LOCK.
Upside / Downside
Restrictiveness
More restrictive — two high-threshold A-grade FCF rules act as tighter joint filter
Liberal Risk
Low — but if both FCF signals simultaneously fail in a market cycle, nothing else catches it
Pros / Cons
- All 5 rules are A-grade (MPS ≥ 79) — highest possible average grade combination
- Average gate MPS: 84.9 (vs 68.6 current)
- FCF-EBITDA CS and FCF Efficency are SIL's top two ranked metrics by confidence
- No B or F grade rules — eliminates two weakest current gate signals
- Still FCF-heavy: FCF Efficency + FCF-EBITDA CS both trace to FCF family
- FCF Efficency (47.1%) and FCF-EBITDA CS (70.3%) combined reduce effective ticker universe
- ROIC-FCF Gate direction lock conflict still present
- SIL flags FCF-EBITDA CS as "HIGH input overlap" — may not add independent signal beyond current gate
- Removing Wtd FCF Eff loses a B-grade rule that has performed well in the gate
Top Picks Under This Gate
-13.2 vs current
Gate Rules
Quality Value Score
thr 0.060 lower · cov 22% · dist 78
ADD78★★★★☆
FCF Efficency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
FCF-EBITDA Convergence Score
thr 2.140 lower · cov 33% · dist 47
CANDIDATE52★★★☆☆
P/S v P/C Multiple
thr 13.480 lower · cov 41% · dist 52
CANDIDATE38★★★☆☆
Watson-ROIC Confidence Index
thr 9.540 lower · cov 35% · dist 7
SKIP38★☆☆☆☆
Control Group Validation
Ctrl Retention (recall)
31 / 127 winners caught at ≥4/5
24% retained
Pass-Set Quality (precision)
Avg Return
+24.28%
return_pct
Signal Independence
A
overlap grade
No HIGH overlaps, no direction conflicts. FCF family fills 2 slots but col31 and col54 are distinct formulas with LOW overlap (share EV/EBITDA indirectly only). 5 meaningful signal dimensions.
Upside / Downside
Restrictiveness
Moderate to strict — FCF Efficency and FCF-EBITDA CS jointly set a high FCF quality bar
Liberal Risk
Low-Medium — Watson-ROIC CI provides non-FCF quality backstop
Pros / Cons
- No HIGH overlaps between any rule pair — most independent gate structure of all alternatives
- 4 A-grade metrics (MPS 75+), 1 C-grade — strong average MPS 84.1
- All 5 rules have direction-confirmed status (no AT_RISK, no locked conflicts)
- Watson-ROIC CI adds a ROIC×Watson quality signal not present in any other gate combination
- FCF Efficency (#1 by MPS at 94.47) finally enters the gate — currently excluded despite being top signal
- FCF-EBITDA CS (#3 by MPS at 91.0) provides the FCF-to-EBITDA quality bridge signal
- Eliminates both current gate problems: Watson 5 direction conflict + PEG F-grade
- 5 distinct signal families: Quality Composite + FCF Efficiency + FCF-EBITDA + Valuation Ratio + ROIC-Watson
- Watson-ROIC CI is C-grade (MPS 56.04) — weakest rule in the gate
- FCF Efficency only 47.1% ticker coverage — some tickers scored on 4/5 rules
- FCF-EBITDA CS flagged as "HIGH input overlap" in SIL due to formula chain — METRIC_OVERLAP_MAP needs v8 review
- Watson-ROIC CI has 48.6% coverage — two rules below 65% threshold
- Removing Wtd FCF Eff loses a rule that has accumulated gate history (v6-v8)
Top Picks Under This Gate
Experimental
+1.4 vs current
Gate Rules
RSI L1
thr 57.560 higher · cov 0% · dist 37
SKIP28★★☆☆☆
P/C L1
thr 3.270 lower · cov 0% · dist 60
CANDIDATE38★★★★☆
Quality Value Score
thr 0.060 lower · cov 22% · dist 78
ADD78★★★★☆
FCF Efficency
thr 0.070 lower · cov 47% · dist 96
KEEP80★★★★★
FCF-EBITDA Convergence Score
thr 2.140 lower · cov 33% · dist 47
CANDIDATE52★★★☆☆
Control Group Validation
Ctrl Retention (recall)
26 / 127 winners caught at ≥4/5
20% retained
Pass-Set Quality (precision)
Avg Return
+43.30%
return_pct
Signal Independence
B+
overlap grade
No HIGH overlaps. L1 rules (RSI, P/C) cover momentum + valuation dimensions absent from L2 rules. LOW overlap between col31 and col54 (indirect EV/EBITDA input).
Upside / Downside
Restrictiveness
Likely moderate — RSI momentum filter may be loose at universe-percentile threshold
Liberal Risk
Medium — RSI threshold needs validation; FCF quality rules provide backstop
Pros / Cons
- RSI provides momentum anchor absent from all current L2 gates
- P/C (L1) is simpler and more direct than P/S v P/C Multiple (L2) — same dimension
- FCF Efficency (#1 MPS at 95.96) + FCF-EBITDA CS (#3 at 95.38) as quality anchors
- QVS (col27, A-grade) adds profitability-quality filter
- Tests whether L1 momentum signal adds lift on top of L2 quality signals
- RSI threshold is experimental — universe percentile may not match optimal momentum zone
- P/C L1 threshold experimental — L2 P/S v P/C Multiple has validated threshold
- FCF Efficency 47.1% coverage — some tickers scored on 4/5 rules only
- FCF-EBITDA CS flagged HIGH input overlap in SIL — METRIC_OVERLAP_MAP v8 review needed
- Two experimental thresholds (L1) mixed with three validated thresholds (L2) — hard to diagnose if wrong
Top Picks Under This Gate
-28.1 vs current
Gate Rules
RSI L1
thr 57.560 higher · cov 0% · dist 37
SKIP28★★☆☆☆
P/C L1
thr 3.270 lower · cov 0% · dist 60
CANDIDATE38★★★★☆
P/FCF L1
thr 29.100 higher · cov 0% · dist 47
CANDIDATE20★★★☆☆
EV/EBITDA L1
thr 20.310 higher · cov 0% · dist 52
CANDIDATE★★★☆☆
P/S L1
thr 5.650 higher · cov 0% · dist 52
CANDIDATE18★★★☆☆
Control Group Validation
Ctrl Retention (recall)
4 / 127 winners caught at ≥4/5
3% retained
Pass-Set Quality (precision)
Avg Return
+33.19%
return_pct
Signal Independence
A
overlap grade
Pure raw ratios — zero formula overlap possible. 5 distinct market dimensions: Momentum, Price/Cash, Price/FCF, Enterprise Value, Price/Sales.
Upside / Downside
Restrictiveness
Unknown — universe-percentile thresholds untested at gate scale
Liberal Risk
Unknown — validate pass-all count before considering deployment
Pros / Cons
- RSI ctrl_d=0.42 (n=493) — strongest single signal in control lab across all 43 metrics
- P/C ctrl return: lower=5.06 vs higher=2.60 — strong separation (n=394)
- P/FCF ctrl_d=0.21 (n=222) — price-to-cash-flow separation confirmed
- 5 fully independent signal families — zero formula overlap possible (raw ratios)
- No direction locks or AT_RISK flags — all from control lab with n≥200
- EXPERIMENTAL — thresholds are universe percentiles, not cohort-validated breakpoints
- MPS grades withheld — Holdings cohort (N≈23) too small; full universe scoring pending
- RSI is momentum — may be partly tautological (high RSI = already performing)
- EV/EBITDA ctrl_d=-0.02 (weak) — control lab signal thin at n=212
- P/S ctrl_d=-0.04 (weak) — included for coverage diversity, not signal strength
- No L2 compound quality check — gate may pass cheap junk without quality filter
Top Picks Under This Gate