House Thesis Scoreboard & Portfolio Kill-Signal Guard
Every Tradestie House Thesis is published with pre-committed qualitative falsification signals and benchmarked against split-adjusted sector ETFs. Winning, lagging, retired, and invalidated calls stay permanently on this public ledger.
Check Any 3–5 Tickers Against Active Thesis Kill-Signals
Instantly test whether your stocks sit inside an Active Bullish Sleeve, an Avoid / Headwind Sleeve, or a thesis under active Monitoring / Invalidated status.
Tradestie Portfolio Risk Sentinel — Personalized Kill-Signal Guard
Automated cross-reference of your Connected Brokerage holdings, Saved Portfolios, and Watchlists against active Avoid sleeves and Thesis Governance signals.
Automated Avoid-Sleeve & Kill-Signal Warnings Pro Guard
Flags any position across your Connected Brokerage, Saved Portfolios, or Watchlists the moment it enters an active Avoid sleeve or when a thesis falsification check shifts from Intact to Monitoring or Invalidated.
Conviction Alignment & 1-Click Alpha Audit Pro Guard
Maps aligned holdings to intact House Thesis conviction sleeves with split-adjusted benchmark alpha tracking and direct 1-click portfolio risk review in Alpha Assistant.
Split-Adjusted House Thesis Live Scoreboard
Click any thesis row to inspect its conceptual summary, falsification signals, and constituent sleeve attribution.
| House Thesis & Category | Inception | Bench | Basket % | Bench % | Excess (pp) | Kill-Signal Status |
|---|---|---|---|---|---|---|
|
Late-Cycle Rotation: Overweight Healthcare, Financials & Energy While Tech Corrects
Macro Sector Rotation
Bullish
· Sample:
LLY,
REGN
+3 Pro
|
2026-10-15 | SPY | +7.80% | -2.26% | +10.06pp | 0/3 Fired — Intact |
Thesis Summary & Sleeve CompositionAs technology and semiconductor capital flows decelerate, institutional leadership is rotating toward cash-generative franchises in Healthcare (XLV), Financials (XLF), and Integrated Energy (XOM, XLE) exhibiting superior relative strength and institutional accumulation breadth against the broad benchmark.
Avoid / Headwind Sleeve:
SMH
+1 Pro
(Avoid Basket: -10.29%)
Pre-Committed Falsification & Kill-Signals
|
||||||
|
Hyperscalers ranked: who captures the surplus when intelligence gets cheap
Hyperscale Cloud
Framework
· Sample:
GOOGL,
AMZN
+1 Pro
|
2026-08-31 | QQQ | +0.64% | -5.14% | +5.78pp | 0/5 Fired — Intact |
Thesis Summary & Sleeve CompositionWhen the cost of intelligence falls tenfold, the freed value goes to one of three places: consumer surplus (if competition forces pass-through), application margin (if distribution is moated), or infrastructure rent (if volume grows faster than price falls). Each hyperscaler is a bundle of businesses sitting at different spots on that map, and the market prices them as one undifferentiated 'AI blue...
Avoid / Headwind Sleeve:
MSFT
(Avoid Basket: 3.27%)
Pre-Committed Falsification & Kill-Signals
|
||||||
|
The Custom ASIC + AMD Host-CPU Alliance: Why Google TPUv8 & OpenAI Jalapeno Displace Single-Vendor NVDA Racks
Custom Silicon & Compute
Long / Short Pair
· Sample:
AMD,
AVGO
+3 Pro
|
2026-11-05 | SOXX | +17.21% | +13.19% | +4.02pp | 0/3 Fired — Intact |
Thesis Summary & Sleeve CompositionHyperscaler diversification away from single-vendor Nvidia (NVDA) racks has reached a structural inflection: Alphabet (GOOGL) confirmed AMD Zen 5 EPYC "Turin" as the host CPU for its next-generation TPUv8 (TPU 8tx / 8i Max) systems, while OpenAI confirmed EPYC Turin as the host node for its custom Jalapeno ASIC (co-developed with AVGO).
Avoid / Headwind Sleeve:
NVDA
+2 Pro
(Avoid Basket: 2.49%)
Pre-Committed Falsification & Kill-Signals
|
||||||
|
Sector Peak Rollover: Asymmetric 2x Put Spread + 1x OTM Call on Overbought Stocks
Asymmetric Hedging
Bearish
· Sample:
MCD,
PEP
+3 Pro
|
2026-11-02 | SPY | +1.65% | +0.30% | +1.35pp | 0/2 Fired — Intact |
Thesis Summary & Sleeve CompositionShorting extended equities outright or with unhedged puts frequently fails during late-stage blow-off squeezes. Pairing defined-risk bear put spreads with upside call convexity inside peak-rollover sectors where institutional inflows are decelerating captures downside air-pockets while hedging late-cycle squeeze risk. Pre-Committed Falsification & Kill-Signals
|
||||||
|
Cheap tokens: agents eat labor budgets, seat-based SaaS is the short
Enterprise SaaS & Agents
Framework
· Sample:
AVGO
|
2026-08-30 | QQQ | -6.30% | -5.14% | -1.16pp | 0/1 Fired — Intact |
Thesis Summary & Sleeve CompositionWhen intelligence becomes a cheap input commodity: (1) agents become the default architecture (100-10,000x tokens per task — only viable cheap, and the reason cheap tokens INCREASE compute demand); (2) the marginal cost of making software collapses — moats shift entirely to proprietary data, distribution, network effects, compliance; code stops being an asset; (3) seat-based pricing dies twice —...
AVGO
-6.3%
Avoid / Headwind Sleeve:
CRM
+3 Pro
(Avoid Basket: -3.10%)
Pre-Committed Falsification & Kill-Signals
|
||||||
|
Own What Algorithms Can't Deflate (token-cost stack)
AI Economics & Compute
Framework
· Sample:
ASML,
TSM
+2 Pro
|
2026-08-30 | SOXX | -14.76% | -12.36% | -2.40pp | 0/3 Fired — Intact |
Thesis Summary & Sleeve CompositionToken price = margin + utilization waste + algorithmic efficiency + chip cost + power. The historic ~10x/year decline in cost per fixed-capability token came mostly from the compressible layers (lab/NVDA margins, batching, quantization/MoE/distillation) — NOT from hardware. As those layers exhaust, the price of intelligence converges toward the incompressible physical layers: leading-edge fab... Pre-Committed Falsification & Kill-Signals
|
||||||
|
Chinese LLM parity: model layer commoditizes, silicon independence is the real question
Global AI & Silicon
Framework
· Sample:
NVDA,
AVGO
+2 Pro
|
2026-08-31 | QQQ | -10.53% | -5.14% | -5.39pp | 0/1 Fired — Intact |
Thesis Summary & Sleeve CompositionThree kinds of parity with opposite implications: (A) benchmark parity — already true (DeepSeek/Qwen/Kimi); (B) cost-adjusted open-weights parity — effectively arrived with Kimi K3-class releases; (C) parity ON DOMESTIC SILICON — still false: Chinese frontier models train on NVIDIA. China has model parity without silicon independence; the binding constraint is HBM + advanced lithography, not talent. Pre-Committed Falsification & Kill-Signals
|
||||||
|
3D stacking / hybrid bonding: the packaging supercycle picks
Advanced Packaging
Bullish
· Sample:
BESI,
SNPS
+3 Pro
|
2026-07-22 | SOXX | -24.95% | -19.37% | -5.58pp | 0/1 Fired — Intact |
Thesis Summary & Sleeve CompositionCFET transistors and the memory wall both funnel into the same manufacturing step: hybrid bonding and 3D stacking. US-listed picks: BESI (hybrid bonding tools), SNPS (3D design EDA), UCTT (subsystems); Asia-listed: ASMPT, Disco. Established Jul 2026. Reinforced by the chokepoint-durability framework: packaging is the only physical chokepoint that the software efficiency wave adds demand to, because it... Pre-Committed Falsification & Kill-Signals
|
||||||
|
The Invariant Stack: Own the Layers Every Future Must Pass Through
AI Infrastructure & Power
Framework
· Sample:
ETN,
POWL
+9 Pro
|
2026-08-30 | SPY | -17.38% | -1.65% | -15.73pp | 0/4 Fired — Intact |
Thesis Summary & Sleeve CompositionYou don't have to predict which future wins. AI datacenters, humanoid robotics, agent swarms, even orbital compute — every candidate for the next big thing consumes the same base layers: energy, power delivery, compute, memory, data. Demand at those layers is horizon-proof. That is the first filter: invariant demand. Pre-Committed Falsification & Kill-Signals
|
||||||
|
Momentum rotation: refiners + precious metals + biotech displace AI infrastructure
Macro Sector Rotation
RETIRED
Closed 2026-11-10
|
2026-08-27 | SPY | -6.21% | -2.75% | -3.46pp | 3/5 Triggered — Retired |
Post-Mortem & Historical AttributionClosure Governance Note: Falsification triggers reached: Late-August precious metals + refiner momentum rotation completed its cycle; capital rotated into large-cap Healthcare (XLV) and Financials (XLF) in October/November. Institutional momentum leadership rotated heavily toward downstream energy refiners (crack-spread margin beneficiaries), precious-metals streamers, and commercial-stage biotech, while prior-cycle AI infrastructure leaders consolidated intermediate drawdowns rather than exhibiting primary trend leadership. Falsification Ledger at Closure
|
||||||
|
AI-Assisted Healthcare: own the profitable operators, not the broken pure plays
Healthcare & Biotech
RETIRED
Closed 2026-11-10
|
2026-08-13 | XBI | -12.05% | -1.15% | -10.90pp | 1/5 Triggered — Retired |
Post-Mortem & Historical AttributionClosure Governance Note: Falsification triggers reached: Operator basket materially underperformed its biotech benchmark as pure-play biotech M&A outpaced healthcare IT/diagnostics multiples. Institutional flow and fundamental evidence favors profitable healthcare operators, diagnostics, and specialty biopharma franchises over cash-burning AI drug-discovery pure plays whose technical breadth and commercial unit economics remain structurally challenged. Falsification Ledger at Closure
|
||||||
|
Datacenter Optics: Margin Migrates to Lasers and Test
Optical Interconnects
RETIRED
Closed 2026-11-10
|
2026-08-30 | SOXX | -25.13% | -13.95% | -11.18pp | Cycle Completed — Retired |
Post-Mortem & Historical AttributionClosure Governance Note: Governance relative-drawdown stop: 800G/1.6T laser & optical test basket sold off amid broader AI hardware digestion; thesis paused until pluggable-to-CPO margin data stabilizes. AI clusters are in the middle of an optical interconnect transition — 800G modules today, 1.6T ramping through 2027, and co-packaged optics (CPO) arriving behind it. The market expresses this through transceiver module makers and DSP chip vendors. We think that's the wrong end of the chain: whichever architecture wins, the durable margin pools one layer down, in lasers and in test. Falsification Ledger at Closure
|
||||||
|
The Memory Wall Is Widening: KV-Cache Tiering Makes AI Memory a Three-Tier Story (HBM -> DRAM -> NVMe)
Thematic Research
RETIRED
Closed 2026-11-10
|
2026-08-30 | SOXX | -25.44% | -13.95% | -11.49pp | Cycle Completed — Retired |
Post-Mortem & Historical AttributionClosure Governance Note: Falsification signal triggered: DRAM/HBM spot pricing momentum softened and memory equipment/makers lagged as inference software quantization reduced near-term bandwidth intensity. Frontier inference is memory-bound, and the binding constraint is now spilling past HBM. The dominant AI workload is shifting from chat to agents — long, multi-turn sessions with a median input of roughly 142,000 tokens — and serving them economically depends on keeping the model's working memory (the KV cache) rather than recomputing it. Measured production traffic shows 91-94% of that cache gets... Falsification Ledger at Closure
|
||||||
|
Boring Bill of Materials: inference scale-out shifts value to unpriced volume parts
AI Infrastructure & Power
RETIRED
Closed 2026-11-10
|
2026-08-09 | SPY | -32.34% | -2.99% | -29.35pp | Cycle Completed — Retired |
Post-Mortem & Historical AttributionClosure Governance Note: Governance relative-drawdown stop: Power/cooling/rack component names experienced severe Q4 multiple compression after Q3 capex pull-forward. Cross-referencing supply-chain dependencies against valuation and basing discipline highlights unpriced volume enablers — mature specialty foundries, power generation, cooling, and electrical contractors — rather than consensus-priced GPU, HBM, and EUV lithography headlines. Falsification Ledger at Closure
|
||||||
|
Physical AI: robots pay the same chokepoints — own actuation, skip the edge-silicon story
Robotics & Physical AI
RETIRED
Closed 2026-11-10
|
2026-08-30 | SPY | -33.41% | -2.23% | -31.18pp | Cycle Completed — Retired |
Post-Mortem & Historical AttributionClosure Governance Note: Governance relative-drawdown stop: Early-stage robotics/actuation hardware corrected sharply during Q4 industrial de-stocking; waiting for 2027 commercial deployment inflection. The load-bearing question in physical AI is not which robot wins — it is where the robot's brain runs. A robot stacks four inference tiers: reflex and safety loops that must run on the machine (that's physics, and it's a microcontroller story, not an AI story); perception that stays local because streaming ten cameras to the cloud costs more than it saves; a roughly once-per-second planning loop — the... Falsification Ledger at Closure
|
||||||
|
GPU fungibility beats dedicated inference ASICs: speed tiers become software
AI Infrastructure & Power
RETIRED
Closed 2026-11-05
|
2026-08-31 | QQQ | -2.31% | -5.14% | +2.83pp | Cycle Completed — Retired |
Post-Mortem & Historical AttributionClosure Governance Note: Superseded by Nvidia Governance Shock & Merchant-to-ASIC Rotation thesis (Nov 2026). Persistent-megakernel compilers (TileRT, MIT-licensed, tile-ai) now hit Cerebras/Groq-class interactivity on stock NVIDIA nodes: 494-600 tok/s/user on 700B-1T MoE models (GLM-5.1, DeepSeek-V3.2) on 8x B200, 3.6x faster than conventional FP8 serving, and Xiaomi claims 1000+ TPS on a 1T model. Production at Xiaomi (MiMo UltraSpeed) and Z.ai (GLM HighSpeed) - both carved a premium speed tier out of GPU... Falsification Ledger at Closure
|
||||||
|
Small-cap biotech washout: own the basket the patent cliff must bid for — never the name
Biotech & Life Sciences
INVALIDATED
Closed 2026-09-27
|
2026-08-31 | SPY | +1.37% | -1.65% | +3.02pp | 1/4 Fired — Invalidated |
Post-Mortem & Historical AttributionClosure Governance Note: Falsification signal triggered: 10-year real yield re-accelerated durably beyond prior cycle highs. Four years of attrition have washed out small-cap drug developers — delistings, reverse splits, programs shut down — leaving the surviving class trading at a fraction of peak while a buyer with a calendar problem watches: big pharma faces roughly $180-250 billion of annual branded revenue losing exclusivity through 2030-33, a hole internal R&D started today cannot fill in time. With mega-mergers... Falsification Ledger at Closure
|
||||||
|
AI Chokepoint Durability Ranking: Packaging > Litho > Power > HBM
Semiconductors & WFE
INVALIDATED
Closed 2026-09-13
|
2026-07-22 | SOXX | -22.64% | -19.37% | -3.27pp | 1/4 Fired — Invalidated |
Post-Mortem & Historical AttributionClosure Governance Note: Falsification signal triggered: Hyperscaler behind-the-meter power bypass and Neocloud AICP GPU-rental floor pause. Ranks the four physical AI infrastructure chokepoints — advanced packaging, EUV lithography, grid power equipment, and HBM memory — by structural durability against algorithmic efficiency gains (quantization, KV-cache compression, distillation, and mixture-of-experts routing). Falsification Ledger at Closure
|
||||||