desk: house-delta date: 2026-10-05 forecasts: delta: p_trade: 0.15 direction: long conviction: low
Briefing reference: /Users/dadbot/Desktop/ClaudeBod/projects/dadbrain/Analysis/briefings/2026-10-05-pm.md PM: DELTA (independent read of my own desk — D3 superseded 2026-09-23; this is House Delta's own opinion, not a blend)
data-infra/regime/current.json (read 2026-10-05, status: ok): as_of 2026-09-22 — SPY day flat (open-to-close −0.08%), above 50d, drawdown bucket <2% (−0.58%), VIX 14.21 LOW (<18). The file now lags the live tape by nine sessions (09-23, 09-24, 09-25, 09-28, 09-29, 09-30, 10-01, 10-02 and today are not reflected). It is a dated read, not a stale-missing one — I name it as behind the tape rather than writing unavailable, the convention my recent plans used.
Consistency with House Delta's ## Regime Charter (desks/house-delta/CLAUDE.md, header still DRAFT — not ratified): consistent; no non-baseline tag is active. Walking the charter's own indicator list (item 5) against the file and today's live tape: spy_drawdown ≥ 5% — no (<2% at the file's read; the recent tape does not approach it); vix_regime MID or HIGH — no (file 14.21 and today's live 16.30 both sit inside LOW (<18)); spy_vs_sma50 below 50d — unmeasured: the file's dated tag says above 50d (as of 09-22), and today's briefing carries SPY at 769.65 against SMA20 764.83 (SPY above its SMA20) but no current SMA50 level, so proximity to that tag is unmeasured, not "close". Today the plan expresses the standing posture: long-only, intraday, flat by 15:30, one position — the design, unchanged.
One reading worth naming precisely, not a shift: VIX at 16.30 (+0.99 vs prior close 15.31) remains deep inside LOW (<18) — 13.7 points below my 30 hard skip, and inside LOW (<18), my best-backtested LONG bucket (own tabulation: <18 PF 2.22, n=62; 18–28 PF 1.62). A tag change needs the same non-baseline tag for 5 consecutive sessions; the tag is still baseline.
Own-desk charter watches: rolling 6-month PF reads 1.648, alert OFF — carried as the demonstrated instrument finding (a fixed trailing-180d window holding the whole book is cumulative PF under another name; the on/off state is one trade's gross P&L against a fixed loss base); no escalation filed on the OFF print. Per-pair PF: AMD→TSM n=11, WR 45.5%, PF 1.346 (above 1.0 for the first time, no recovery declared — one of its three recent wins, 09-22, is measured sector beta), NVDA→ASML 2W/0L, +$264.42 (unseen since 08-03 — ~9 weeks). Watch and report only — a persistent shift becomes a Desk Review proposal (recalibration / sit-out / new-mechanism), never an automatic change, and I touch no gate.
Call: p_trade 0.15, direction long, conviction low. Setup present possible but unlikely — the session is open to me (no hard skip) and VIX sits in my best bucket, but nothing in the pre-open state supplies an idiosyncratic dispersion reason to expect a ≥1.5% first-hour leader impulse, and the tape's recent character argues against one.
My signal is a leader's first-hour return off the 09:30 cash open, read at 09:55, versus ±1.5%: NVDA ≥ +1.5% → buy ASML at the 10:00 bar (primary, edge-carrying cell, live 2W/0L +$264.42, dormant since 08-03 — ~9 weeks), AMD ≥ +1.5% → buy TSM at the 10:00 bar (secondary, n=11, WR 45.5%, PF 1.346). A leader ≤ −1.5% is my disabled-SHORT branch (ENABLE_SHORT = False) — no trade by rule. If both leaders clear up, the larger |return| wins (single position, no stacking).
Why 0.15, and why low conviction. p_trade is the probability the engine logs ≥ 1 real trade today, not whether it should. intelligence/plan-calibration.csv, my rows as of today (54 scored days, 9 fires, unconditional realized fire rate ~0.167, mean forecast ~0.175):
| p_trade bucket | scored plan days | fires | realized fire rate |
|---|---|---|---|
| 0.00–0.10 | 13 | 0 | 0.000 |
| 0.12–0.15 | 16 | 4 | 0.250 |
| 0.18–0.25 | 14 | 2 | 0.143 |
| 0.30+ | 11 | 3 | 0.273 |
| all | 54 | 9 | ~0.167 |
The five hard-skip sessions in the 0.00 bucket (07-14 CPI, 08-12 CPI, 08-26 PCE, 09-04 NFP, 09-30 PCE, 10-02 NFP) never fired, exactly as the rule guarantees. Today is not a skip, so 0.00 would be dishonest — the session is live and the up-branch can fire. But the recent 1 fire in 6 tradeable sessions (09-23→10-01: the only ±1.5% clear was 09-28's disabled-SHORT AMD −2.07%; the best LONG reading was 09-29's AMD +0.97%, 0.53pp short) says the impulse has been scarce. 0.15 sits inside the 0.12–0.15 bucket (realized 0.250) — held below its own realized rate, per the discipline I set after the 09-25/10-01 over-forecasts, and it is not lifted by the soft-jobs / rates-relief story, because a broad beta bid is exactly the shape my own stand-down argument classifies as not-edge. direction: long because LONG is the only direction I am permitted to take; conviction: low because the number is near the base rate and the call would move on any real pre-open dispersion signal — which today is absent.
Sizing vs. event risk: standard — 0.75% risk, single position, if a signal fires. No size-down and no headline buffer is invented off a morning read. One genuine caveat I name rather than adjust for: the ISM print lands at 10:00 AM ET, the exact bar I would enter on — a first-hour clear would be taken into a scheduled macro release, not after it. That is a real execution risk, but it is neither a hard skip (ISM is not FOMC/CPI/NFP/PCE) nor a reason to change the SOP: my exit is a 2.0× ATR stop plus the hard 15:30 flat, and a whipsaw into the print is inside the loss distribution the backtest already prices. I adjust nothing; I name it.
Session-character fit: marginal. My mechanism needs a large idiosyncratic first-hour leader impulse that the laggard has not yet followed. The pre-open state supplies the bucket (VIX LOW — my best) but not the trigger: futures are mixed-to-slightly-lower with no confirmed gap, the tape's recent character is sub-ATR, and the one complex-wide bid the last two weeks produced (MU/AI-memory, 10-02) was the beta shape my stand-down argument says cannot transmit leader-to-laggard cleanly. A LOW-VIX, low-dispersion, non-directional Monday into a 10:00 macro print is the tape where a leader most often fails to clear ±1.5% — which is why the number is low, not because the mechanism is impaired.
Key levels (mine only): the only load-bearing numbers are each leader's first-hour return off the 09:30 cash open, read at 09:55, versus ±1.5% — NVDA ≥ +1.5% → buy ASML (primary, edge-carrying, live 2W/0L +$264.42 — unseen since 2026-08-03, ~9 weeks); AMD ≥ +1.5% → buy TSM (secondary, n=11, WR 45.5%, PF 1.346). Index gap-framing only: SPY 767.16 / 772.65 (range), SMA20 764.83, ATR 7.12; QQQ 747.63 / 754.51 (range), SMA20 727.77, ATR 9.70. Carried data gap: the non-firing pair's first-hour value is recorded on traded days only as the blocking note, and on hard-skip days no leader value is recorded at all — a live-dispersion read today is available from the pipeline only as the no_signal blocking note if it does not fire.
Invalidation / no-trade reasons (binding today):
no_trade:no_signal (the modal outcome; ~0.167 base rate).ENABLE_SHORT = False) → no trade by rule — the branch the tape would land in if the soft-jobs read reverses and semis sell off.Desk book state (carries forward, unchanged by today's plan). Cumulative +$566.45 across 13 LONG trades, 7W/6L, WR 53.8%, gross PF 1.648; SPRT n=13, LLR +0.976, CONTINUE (+1.969 to CONSISTENT-WITH-BACKTEST, −3.920 to DEGRADED). NVDA→ASML 2W/0L, +$264.42 — the edge-carrying cell, starved ~9 weeks (unseen since 08-03), regime-shaped, not a gate problem (VIX < 18 is my best bucket and inside it the tape has been quiet). AMD→TSM n=11, WR 45.5%, PF 1.346 — above 1.0 for the first time on three consecutive wins, no recovery declared (one of the three, 09-22, is measured sector beta); the 09-05 stop-calibration diagnostic read NO and remains pending TEMPER cadence-ratification via the 2026-09 Desk Review. Rolling 6-month PF 1.648, alert OFF (instrument artifact, above). SHORT re-enablement rejected 2026-07-15 — live gate stands; the governing path (30 LONG trades at PF ≥ 1.5) is at n=13, 17 trades from being decidable. Stand-down argument live and provisional; its withdrawal trigger is re-registered on the leader-vs-basket residual, not the PF ratio (last reading: 09-24, +2.28pp vs SMH, the first case against the argument). Most recent live sessions: 10-01 no_signal (AMD −0.56%, NVDA +0.39%), 09-30 PCE hard skip, 10-02 NFP hard skip.
Forecast-methodology note (the open item I am carrying, not a gate change). My last two reflections flagged a twice-measured forecast-layer error: pricing 0.22 (09-25) and 0.30 (10-01) on complex-wide-bid sessions, reading sector beta as edge. Today I anchor to the mechanical dispersion state (VIX bucket, leader leg moving pre-market, session containment) and hold the number below its bucket's realized rate — which is the re-anchor already with TEMPER (DC-1). This is a p_trade methodology input only; no gate, floor, window, stop or sizing parameter is touched, and nothing is applied unilaterally.
Operational item (desk-top, carried): runner-health-2026-09-25 recorded delta: completed, daily artifact present, state ok — clean sessions after six September watchdog flags; 09-28 through 10-02 all produced normal rows. But the ledger-freshness post-run check is still unwired, so a clean run is being observed by me, not verified by the pipeline. Today's forecast is 0.15, so if a signal fires the EOD score depends on a live row landing — I note the check remains unwired.
p_trade 0.15, direction long, conviction low. Session is live (no hard skip) and VIX sits in my best LONG bucket, but the pre-open state supplies no idiosyncratic dispersion trigger; the recent 1-fire-in-6-tradeable-sessions tape and the 10:00 ISM collision argue the leader most likely fails to clear ±1.5%. Standard 0.75% single-position sizing if a signal fires; no adjustment — the ISM-at-entry collision is named, not traded around. No gate, floor, window, stop or sizing parameter touched; the p_trade dispersion re-anchor is a TEMPER item (DC-1), filed as a note only.No trades taken.