conic.ai

CONVICTION PICKS

AVOID / TRIM

• AMAT — RSI 76.6, ATH chase. Fintwit just declared the new ATH — that's the exit, not the entry.
• LRCX — RSI 69.6, pos_52w 1.0, 1m +36.6%. Same chase profile as AMAT.
• KLAC — RSI 74.2, 1m +42%, 1w +20% — pure parabolic, no quality dip to enter on.
• MU — RSI 62, 1m +50%, pos_52w 1.0. Quality (F=7) but valuation chase.
• ARM — 1m +97%. Uninvestable until the air comes out.
• HPE — F=4, spike_only, 3m +127%. No fundamentals.
• SMCI — 1w -24%, hqm False. The fade is working — keep fading.
• VRT — Bounce off a broken trend; F=5, 1m -15.9%. Wait.
• NVDA — F=4 (binding portfolio gate), thesis keeps missing. Downgraded to watch.
• CRCL — F=3, falling knife pattern.
• VCX — fc=0, no F-score, 1w -28% / 1m -44%. Worst name in book.

POSITION PLAN & RISK exposure x1.0 · open P&L +1.3% · realized -5.5% · 2 open / 5 closed

BACKTEST · $100k since inception paper book (with stops) vs the plan basket

VALIDATION / EDGE

BELIEF UPDATES · vs yesterday

• UPGRADE CSCO avoid→watch This is the VRT-style pullback I was waiting for. RSI dropped from 79 to 52, 1w -0.2%, 1m +1.7%. F=8 (quality intact), fc=0.70 (top-6 in book), abs+ts mom still pass. Upgrade to WATCH; need one more constructive session before action.
• DOWNGRADE NVDA accumulate→watch F-score still only 4 (binding yellow flag — under the F<=4 portfolio gate, this would be blocked), hqm_quality now False, fc=0.40, 1m -5.6%, RSI 48. Hit rate 43% on 7 calls. Adriann's H20-China bull case is real but fundamentals don't justify high conviction. Downgrade per the 'thesis that keeps missing' rule.
• KEEP TSM buy→buy Still the universe king: factor_composite 0.81, F-score 9 (top in book), hqm_quality true, abs+ts mom pass, RSI 59 (not stretched), 1w +3.3% / 1m +9.3%. The cleanest quality+momentum stack in semis. Streak still negative (-3) on 3d horizon — beta noise, not thesis decay.
• KEEP AVGO accumulate→accumulate F=8 holds, fc=0.55, amom+ts pass, RSI 43.6 (constructive pullback from prior overbought), 1m -7.3% — the dip is exactly what we wanted. Best fundamentals in semis ex-TSM. Streak -4 is the cost of buying quality before consensus piles back in.
• KEEP VRT avoid→avoid fc=0.42, F=5, 1m -15.9% confirms the downgrade was right. Streak 2 on avoid, hit 33%. Wait for a real quality-floor re-rate, not a dead-cat 1w bounce of +7.7%.
• KEEP META avoid→avoid abs_mom and ts_mom still False, pos_52w 0.26, 1m -3.4%. Streak 6, hit 86%(7) — best avoid call in the book, keep it.
• KEEP MSFT avoid→avoid abs+ts mom False, pos_52w 0.24, 1m -5.0%. Share-loss narrative intact, no inflection.
• KEEP ESTC avoid→avoid spike_only true, abs_mom False, 1m +23% pure squeeze. F=5, fc=0.53.
• KEEP SMCI avoid→avoid 1w -24%, no quality floor (hqm False, fc=0.42, F=5). Avoid streak 7 hit 100% — keep fading.
• KEEP MU avoid→avoid 1m +50%, RSI 62, pos_52w 1.0. F=7 is real but valuation/RSI rules out chasing. Discipline over FOMO.
• KEEP ARM avoid→avoid 1m +97%, RSI 63. Valuation uninvestable. Keep avoiding.
• KEEP AMAT avoid→avoid RSI 76.6 (ATH chase zone), 1m +34%, 1w +17%. Fintwit calling out ATH cluster — exactly when retail piles in. Fade, don't chase.
• KEEP LRCX avoid→avoid RSI 69.6, pos_52w 1.0, 1m +36.6%. Same pattern as AMAT — wait for the pullback we already got in AVGO.
• KEEP ASML avoid→avoid RSI 68.1, pos_52w 0.99, 1m +26%. F=8 is real but RSI says wait.
• KEEP HPE avoid→avoid F=4, spike_only true, 1m +48%, 3m +127%. Pure mania, no fundamentals support.
• KEEP MRVL avoid→avoid RSI 66.8, 1m +75%, spike_only False but the print is mania-shaped. Keep avoiding.
• KEEP SMTC avoid→avoid pos_52w 1.0, RSI 54, F=7 — quality but extended. No entry.
• KEEP NTAP avoid→avoid spike_only true, 1m +35%. Keep avoiding.
• KEEP CRCL avoid→avoid F=3, fc=0.23, 1m -27%. Falling knife.
• KEEP CRWV avoid→avoid abs_mom False, fc=0.37. Hold avoid.
• KEEP WOLF avoid→avoid Fintwit (Ren) flagging post-bankruptcy SiC datacenter optionality — interesting narrative but F-score null, abs_mom null. Story without numbers. Re-evaluate when fundamentals print.
• KEEP AXTI avoid→avoid F=2, 1w +41% pure mania. Avoid.
• KEEP APLD avoid→avoid F=4, fc=0.24. Speculative.
• KEEP ACMR avoid→avoid F=2 floor unchanged. Mania.
• KEEP PL avoid→avoid F=2, fc=0.12. Worst in book ex-VCX.
• KEEP AAPL watch→watch NEW belief candidate: fc=0.68 (top-8), F=8, hqm_quality true, abs+ts mom pass, RSI 40 (oversold), 1m -1.3%. Classic quality-dip setup but not yet in belief ledger — start tracking as WATCH; wants one more session of RSI recovery before promote.

VS PRIOR DAYS

Yesterday's calls (2026-06-12) all still pending maturation on the 3d horizon. Track record overall: buy beat-universe 31% (32 matured) — slightly under coin-flip; high-conviction n=12 at 25% beat-rate is humbling. Edge so far is concentrated in the AVOID book (62% under-universe on 104 matured calls) — fading the mania cohort works, picking the next leader is harder. Lesson today: stay disciplined on quality (TSM/AVGO/AAPL/CSCO) and don't try to be a hero on the ATH-cluster names just because fintwit is screaming.

MACRO & REGIME

X / FINTWIT TRENDS

• AMAT/LRCX/KLAC all at new ATHs (MarketWise) — retail FOMO confirms the late-leg chase; this is the AVOID signal, not the BUY signal.
• Iran war-risk de-escalation removing valuation cap on semis (Casey); secular AI capex thesis intact but priced.
• Morgan Stanley: $41bn/GW NVDA Vera Rubin vs $15bn/GW Trainium3 — hyperscaler capex spend per GW is the structural alpha.
• Datacenter power-semis layer (MPWR, VICR, POWI, NVTS, AOSL, AEHR) flagged as next-gen GaN/SiC beneficiaries — not in watchlist, candidates to add.
• China trade-data beat is defensive (chip front-running + gold reserves), not demand recovery — bearish for FXI/KWEB, neutral-to-bullish for NVDA H20 channel.

FULL WATCHLIST · reference only, descriptive not a buy list