The rulebook is versioned. Today: V5.1.

rulebook changelog

1. V5.1 today · current live

   A sideways-market detector

   June exposed a blind spot. When QQQ chopped inside a tight range (~$709–748) the system still read it as a bull trend and kept buying breakouts at the top of that range. V5.1 adds RangeState (§5.6), the missing regime for “a range inside an uptrend.” It watches for a market going nowhere efficiently — price stuck around its 20-day average with no real direction — and only flips on after the condition holds for a couple of sessions. A 5-year study shows it lights up about 17% of days and stays quiet through genuine trends. When it's on, three things change: the “let the winner run” take-profit carry goes dormant, the most aggressive breakout entries drop to awareness-only, and the desk is instead allowed to buy a defended low at the bottom of the range. It's context only — it adds no score and touches neither the seven-module engine nor any breakout rule; it only changes what's allowed to trade. A same-week refinement (4 Jul) also sharpened the V5.0 options net-drift read — a cleaner Mag-7 signal plus a morning-persistence check — still context-only, still non-scoring.

   + 5.6 RangeState 10.2 refined context only · non-scoring ~17% of days 11B dormant · AGG awareness-only
2. V5.0 18 Jun

   Reading the options net-drift tape

   A new daily read (Step 1f) watches the options net-drift tape — the end-of-day balance of call vs put money — for both the Mag-7 basket and QQQ, over a one-to-three day horizon. The edge is contrarian and narrow: when Mag-7 flow looks bearish (puts outweigh calls), that lean has historically been faded into a mild tailwind; and euphoric one-sided call buying in QQQ is a “do not chase” caution that tends to cool by day two. It's context only — a new 10.2 that adds no score, carries low confidence, and stands aside in a structural bear. Two more pieces lock it down: 5.5 sets the house rules for every context signal (judge it against the market's natural drift, read only the direction not the size, and re-prove it after the next real downturn), and a 10.3 register keeps the five separate net-drift reads from ever being double-counted.

   + 10.2 + 10.3 + 5.5 context only · non-scoring 1–3d horizon Step 1f
3. V4.9 15 Jun

   A macro-decoupling lens on risk

   A new daily monitor (Step 1e) joins the macro read: it tracks the 20-day correlation to QQQ of a 14-name macro-inverse basket — Treasuries, yields, bond-vol, the dollar and energy. It answers one question — is the rates-and-dollar “risk-off” relationship still loaded, or is it releasing? When three or more names resolve out of deep-inverse together (broad resolution), that's historically a medium-term bullish tailwind for QQQ over the next 20–40 sessions; when three or more stay deeply inverse (loaded / risk-off), it's a macro headwind to respect. It's context only — a new 6.2 that adds no score to the seven-module structure and never authorizes sizing or entries. Backed by a 5-year study: broad resolution → +9% over 40 days, 89% positive; no edge at the 5–10 day horizon, so it is never a trade trigger.

   + 6.2 context only · non-scoring 20–40d horizon Step 1e
4. V4.8 5 Jun

   Reading the options tape a day early

   A 21-session study (4 May–3 Jun) found the prior evening's options-positioning dashboard already held the signal the scoring modules were picking up a full session late. The planner now reads that screenshot every morning: Mag-7 options drift can cool — or zero out — the leadership read (new 8.3), and the options heat-map can lift or cut the derivatives read (new 10.1). A stretch sub-flag (5.2), an early-warning note on the take-profit carve-out (11B), and three new conflict checks (18) round it out.

   + 8.3 + 10.1 5.2 amended 11B amended 18 amended
5. V4.7 22 May

   Markov + HMM regime persistence

   A QQQ-calibrated Markov model now anchors the 17 probability framework — replacing the old fixed 50% base with a real 10-step forward distribution. An HMM runs daily as an informational divergence flag; it never touches the scoring math. New 5.4 introduced; 17 and 18 amended.

   + 5.4 17 amended 18 amended
6. V4.6 19 May

   Module-input expansion

   Driven by the 18 May correlation sweep: Macro (6) adds TLT + XLE and a defensive-rotation tag (SPLV/XLU); Breadth (7) adds IWM; Leadership (8) adds IGV/FDN, a semi supply-chain pulse, and a Mag-7 dispersion check. The seven-module SBS architecture is unchanged.

   6 macro 7 breadth 8 leadership
7. V4.5 12 May

   11B take-profit carve-out

   Under Structural Bull only (SBS +5…+7), the planner may emit a stop-sell-at-final-destination per long tier, with a regime-drift kill-switch if SBS slips below +5. Every other execution prohibition stays in force.

   + 11B SBS +5…+7
8. V4.4.1 12 May

   High-impact events only

   The economic calendar is filtered to high-impact releases; medium- and low-impact events are dropped from all inputs, scenario logic, and gap-risk labeling.

   3.1 3.2 15
9. V4.4 baseline

   Swing-mapping baseline

   Pre-Market Planner becomes a pure higher-timeframe engine — regime, location, scenario mapping, activation, invalidation, and zone rationale. Narrow-zone discipline is mandatory: every activation band is tighter than 2.00 QQQ points.

   planner core < 2.00 pt bands

No retroactive edits. Every version is dated and stays accessible for rollback — the changelog above is the complete, public record of how the rulebook has evolved.