Ranked by EDGE REMAINING — genuine signal × how much of the move is still ahead × a leading desk firing before the crowd — not by how many desks agree (agreement is a lagging, consensus condition). The hero is what is EMERGING: a leading desk positioning into a quiet tape with room to run. The crowd-favourite that has already run is demoted on purpose. Context only, never a trade trigger.按「剩余空间」排序 —— 真实信号 × 行情尚未走完的幅度 × 领先信号台抢在人群之前发动 —— 而非看多少台一致（一致性是滞后的共识状态）。主推「萌芽」机会：领先信号台在安静的盘面中悄悄建仓、仍有上行空间。已经跑完的人群宠儿被刻意下调。仅供参考，绝非交易触发。
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间36%
-49% off high距高点-27pp RS 60d相对强度rolling over见顶回落
falteringcatalyst
GGGGGG
24opp机会
conv信念 17 · 1d台
early早期no entry无入场leads crowd by领先 +1
A leading desk leads, the tape is still catching up — early, ~79% edge remaining.
edge left剩余空间79%
NB$4.26NB
24opp机会
conv信念 43 · 1d台
distribution派发no entry无入场
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间34%
-64% off high距高点-36pp RS 60d相对强度rolling over见顶回落
faltering
MU$848.95MU
22opp机会
conv信念 60 · 1d台
distribution派发no entry无入场
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间31%
-30% off high距高点+83pp RS 60d相对强度rolling over见顶回落
falteringisolated
ASPI$3.88ASPI
20opp机会
conv信念 34 · 1d台
distribution派发no entry无入场
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间46%
-72% off high距高点-29pp RS 60d相对强度rolling over见顶回落
falteringcatalyst
USAR$15.65USAR
20opp机会
conv信念 34 · 1d台
distribution派发no entry无入场
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间46%
-60% off high距高点-37pp RS 60d相对强度rolling over见顶回落
falteringcatalyst
ALM$13.77ALM
20opp机会
conv信念 44 · 1d台
distribution派发no entry无入场
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间34%
-41% off high距高点-43pp RS 60d相对强度rolling over见顶回落
faltering
MSFT$393.82MSFT
20opp机会
conv信念 3 · 0d台
building酝酿no entry无入场
No strong cross-desk signal.
edge left剩余空间65%
-27% off high距高点-13pp RS 60d相对强度
catalystdiscovery
LEU$156.05LEU
19opp机会
conv信念 35 · 1d台
distribution派发no entry无入场
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间34%
-64% off high距高点-24pp RS 60d相对强度rolling over见顶回落
faltering
WDC$477.22WDC
19opp机会
conv信念 39 · 1d台
distribution派发no entry无入场
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间42%
-36% off high距高点+19pp RS 60d相对强度rolling over见顶回落
faltering+1 peers同伴: MU
UBER$72.46UBER
17opp机会
conv信念 24 · 1d台
distribution派发no entry无入场
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间62%
-28% off high距高点-12pp RS 60d相对强度
catalyst
ABBV$254.49ABBV
17opp机会
conv信念 35 · 1d台
distribution派发no entry无入场
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间42%
-2% off high距高点+19pp RS 60d相对强度
UUUU$11.49UUUU
17opp机会
conv信念 34 · 1d台
distribution派发no entry无入场
Desks lean bearish and price is rolling over — distribution; de-risk / avoid, not a long.
edge left剩余空间36%
-59% off high距高点-50pp RS 60d相对强度rolling over见顶回落
falteringcatalyst
Signal scorecard — is the edge real?信号记分卡 —— 优势是否真实？
the hub graded against what actually happened — context only, not a trade signal以实际走势检验本台 —— 仅供参考，非交易信号
5d天horizon周期measuring测量中
rank skill排名预测力+0.008
graded calls已评分判断9,961
emerging: beat SPY?萌芽：跑赢SPY？-0.7%no否
exhausted: lagged SPY?透支：跑输SPY？-0.4%yes是
10d天horizon周期measuring测量中
rank skill排名预测力-0.028
graded calls已评分判断8,792
emerging: beat SPY?萌芽：跑赢SPY？-1.9%no否
exhausted: lagged SPY?透支：跑输SPY？-1.8%yes是
21d天horizon周期measuring测量中
rank skill排名预测力-0.213
graded calls已评分判断3,169
emerging: beat SPY?萌芽：跑赢SPY？-8.7%no否
exhausted: lagged SPY?透支：跑输SPY？-3.1%yes是
Sector heat板块热度
where the desks cluster各台聚集之处
XLE Energy
mean均32.0names名1
XLV Health Care
mean均11.5names名22
XLK Technology
mean均10.3names名27
XLU Utilities
mean均10.0names名12
XLI Industrials
policy政策 ↑
mean均9.7names名16
XLC Communication
mean均6.2names名9
XLY Discretionary
mean均5.7names名12
XLRE Real Estate
mean均4.8names名3
XLB Materials
mean均4.8names名1
XLP Staples
mean均4.0names名9
XLF Financials
mean均3.7names名26
📊
Desk scorecard — is this desk beating the market?研究台记分卡 —— 本台是否跑赢大盘？every name this desk surfaces is graded against realized SPY-relative forward returns · context-only本台每个入选标的均以实际相对标普前瞻收益评分 · 仅作参考
Measuring测量中 —
matured observations accruing; no horizon has cleared the significance bar yet.到期观测持续累积；尚无周期通过显著性检验。126,685 live snapshots · 3,760 departures tracked126,685 条在册快照 · 已追踪 3,760 次离场
Horizon周期
n样本
IC · levelIC · 水平
IC · velocityIC · 速度
on-desk在台
departed离台
coverage覆盖
status状态
5d日
7353
+0.012
+0.100
-0.8%
-2.1%
7.0%
measuring测量中
10d日
5679
-0.096
+0.012
-2.2%
-4.3%
6.4%
measuring测量中
20d日
0
—
—
—
—
0.0%
measuring测量中
30d日
0
—
—
—
—
0.0%
measuring测量中
60d日
0
—
—
—
—
—
measuring测量中
90d日
0
—
—
—
—
—
measuring测量中
~ = provisional pooled IC (single cross-section); the daily Newey-West HAC IC that the "proven" gate trusts needs ≥6 grading dates. Velocity IC = the change in the desk's own score, the owner's second axis.~ = 暂定合并信息系数（单一横截面）；"已验证"门槛所依赖的日度 Newey-West HAC 信息系数需≥6个评分日。速度信息系数 = 本台评分自身的变化，即所有者的第二维度。
Engine-version A/B引擎版本 A/B— pre/post overhaul, where the ledger carries the stamp—— 改版前后，以账本已打戳的行为准
5d日hub-v3-trajectory: IC +0.150n=2079 · -1.7%
10d日hub-v3-trajectory: IC +0.107n=410 · -3.3%
Auto-findings自动结论— derived from the stats each run, to improve this desk's engine—— 每次运行从统计中导出，用于改进本台引擎
●5d: score VELOCITY IC (+0.100) exceeds LEVEL IC (+0.012) — the change in the desk's score ranks winners better than its level.5d: score VELOCITY IC (+0.100) exceeds LEVEL IC (+0.012) — the change in the desk's score ranks winners better than its level.
●5d: departed names (-2.1% avg) underperform on-desk names (-0.8%) — DROPS ARE CORRECT; the desk sheds names before they lag.5d: departed names (-2.1% avg) underperform on-desk names (-0.8%) — DROPS ARE CORRECT; the desk sheds names before they lag.
▲5d: 'exhausted' (-1.1%) outperforms 'emerging' (-1.7%) — possible residual inversion in the lifecycle staging.5d: 'exhausted' (-1.1%) outperforms 'emerging' (-1.7%) — possible residual inversion in the lifecycle staging.
●10d: score VELOCITY IC (+0.012) exceeds LEVEL IC (-0.096) — the change in the desk's score ranks winners better than its level.10d: score VELOCITY IC (+0.012) exceeds LEVEL IC (-0.096) — the change in the desk's score ranks winners better than its level.
●10d: departed names (-4.3% avg) underperform on-desk names (-2.2%) — DROPS ARE CORRECT; the desk sheds names before they lag.10d: departed names (-4.3% avg) underperform on-desk names (-2.2%) — DROPS ARE CORRECT; the desk sheds names before they lag.
▲10d: 'exhausted' (-3.4%) outperforms 'emerging' (-4.2%) — possible residual inversion in the lifecycle staging.10d: 'exhausted' (-3.4%) outperforms 'emerging' (-4.2%) — possible residual inversion in the lifecycle staging.
Notes & lessons笔记与经验— what we've learned, so the desk's engine keeps improving as data accrues—— 已学到的经验，使本台引擎随数据累积持续改进
2026-07-042026-07-04 inversion finding: the OLD hub engine was anti-predictive — pooled opportunity IC ~-0.09 @5d, and 'exhausted' names (+1.84% avg fwd) BEAT 'emerging' (-0.77%). The hub-v3-trajectory overhaul is the fix under test; this grader's engine_version A/B is the experiment that proves whether it worked.2026-07-04 inversion finding: the OLD hub engine was anti-predictive — pooled opportunity IC ~-0.09 @5d, and 'exhausted' names (+1.84% avg fwd) BEAT 'emerging' (-0.77%). The hub-v3-trajectory overhaul is the fix under test; this grader's engine_version A/B is the experiment that proves whether it worked.
2026-07-04PRICE LESSON: grade off data/yahoo/.parquet adjusted closes ONLY. The S&P-1500 breadth close cache (data/breadth/_closes_cache.parquet) is SPLIT-CORRUPTED — using it silently poisons every RS / forward-return number. A name missing from yahoo is a coverage skip here, never a wrong return.PRICE LESSON: grade off data/yahoo/.parquet adjusted closes ONLY. The S&P-1500 breadth close cache (data/breadth/_closes_cache.parquet) is SPLIT-CORRUPTED — using it silently poisons every RS / forward-return number. A name missing from yahoo is a coverage skip here, never a wrong return.
2026-07-04NOTE on the engine_version A/B: the historical hub ledger carries NO engine_version stamp (all ~63k rows pre-date the stamp), so the pre/post-overhaul split only begins accruing from 2026-07-04 forward. Until enough hub-v3-trajectory rows mature, the A/B reads one-sided; that is expected, not a bug.NOTE on the engine_version A/B: the historical hub ledger carries NO engine_version stamp (all ~63k rows pre-date the stamp), so the pre/post-overhaul split only begins accruing from 2026-07-04 forward. Until enough hub-v3-trajectory rows mature, the A/B reads one-sided; that is expected, not a bug.
A falsifiable scorecard of this desk's own surfaced names — its ranking score (level AND velocity) graded against realized SPY-relative forward returns at 5/10/20/30/60/90 trading days, split on-desk vs after-leaving. A signal is 'proven' only past n>=40 with a significant Newey-West t of the right sign; a significant negative IC is an 'alarm'. Context-only — nothing here sizes a position.