SERVED 2026-10-01 20:41 UTC. This brief is for the open of 2026-10-02; the coming NYSE session is 2026-10-02: FRESH (made 0.3 hours ago). # OPAL brief — what the tested rules hold from the next open Generated 2026-10-01T20:23:22+00:00 · data through 2026-10-01 · for the open of **2026-10-02** · schema opal-brief/1 OPAL records and judges; it never trades. These are research outputs of rules tested on paper, not predictions and not advice. The operator decides whether any money follows them. ## Guardrails for an agent acting on this 1. Weights are fractions of the capital the operator assigns to that rule, never of the whole account. They never sum above 1.0: no leverage, no margin, no shorting. 2. These are next-open orders computed from the prior close. Act at or just after the regular-session open on valid_for_open_of, with market or marketable limit orders. Do not trade them later in the day or outside regular hours. 3. Act only on a brief that is fresh: its valid_for_open_of is the coming regular NYSE session (weekends and exchange holidays counted), which the server states as fresh: true. A brief for any other session is stale, however recent its generated_at. Stale targets are not targets. 4. Trade a rule only when its trade_today is true. On other days hold what you have, even if it has drifted from the targets: the rules were tested trading only on those days (the vol rule when its target moves 10 points, trend at the first open of a month). 5. One exception: a NEW allocation (holding none of a rule's positions) enters once at the rule's targets at the next open, as the paper books did; from then on, trade_today only. 6. When trading, move only the difference between what is held and the target; skip any order under $1 (the broker's fractional minimum). 7. If a rule's forward book is STOPPED, sell that rule's positions and do not re-enter. The stop was set before the book started. 8. Rules listed under do_not_trade have no evidence of an edge. Do not trade them. 9. The QQQ VWAP rule is unvalidated (INSUFFICIENT): if traded at all, $20 a trade and the PLAYBOOK's own limits, never more. 10. Nothing here is a prediction of prices, and past results do not guarantee future ones. The operator decides whether any money follows these targets. ## Volatility-targeted SPY (`vol_managed`) - **Trade today:** no. **Action at the next open:** HOLD — no orders. SPY stays at 100%: the target (100%) is within 10 points. - **Targets** (fraction of this rule's capital): SPY 100.0% - **Evidence:** CANDIDATE on total returns (8/9 gates; failed G3). Beat SPY on risk-adjusted return after publication (Sharpe 0.82 vs 0.75, 2017+); lower return than SPY (12.9% vs 15.3%/yr); worst drawdown -33.5% in 2008, which fails the -30% gate. - **Forward paper book:** since 2026-09-29, 3 sessions, -0.33%, drawdown -0.3%. - **Rule:** After each close, sigma = 20-day realised vol of SPY (annualised). Target = min(1, 0.15 / sigma). Trade at the next open only when the target moves more than 10 points from the last order. - **Stops if:** drawdown past -35%, or a 3-year Sharpe below SPY's. ## Multi-asset trend (8 ETFs) (`trend_tsmom`) - **Trade today:** no. **Action at the next open:** HOLD — no orders. The rule trades only at the first open of a month (next: 2026-11-02). - **Targets** (fraction of this rule's capital): SPY 12.5%, EFA 12.5%, EEM 12.5%, DBC 12.5%; cash/T-bills 50.0% - **Signal from:** the current month's signal (no rebalance until the month turns). - **Evidence:** HIGH QUALITY on total returns (8/8 gates). Passed every gate on total returns, narrowly (Sharpe 0.53 vs 0.5 needed; bootstrap interval 0.14-0.92). 4.95%/yr after 2008 against 6.04% for its basket; worst drawdown -11.7% against -37.6%. - **Forward paper book:** since 2026-09-29, 3 sessions, -0.50%, drawdown -0.5%. - **Rule:** Eight ETFs (SPY EFA EEM TLT IEF GLD DBC VNQ), 1/8 each. At each month-end close a sleeve is held when its total-return close is above the mean of its last ten month-end closes, else it sits in cash/T-bills. Rebalance the whole book at the next open. - **Stops if:** drawdown past -20%, or a 3-year Sharpe below its equal-weight basket's. ## Both rules, half each (illustration) - Targets: SPY 56.2%, EFA 6.2%, EEM 6.2%, DBC 6.2% - Half the capital to each rule. Computed after the results existed, so an illustration, not a tested rule (FINDING.md section 11). ## Do not trade - **QQQ VWAP dump-and-reclaim** (`vwap_reclaim`, INSUFFICIENT): 6 of 100 forward trades; mean +7.8 bp, t = 0.55. Not evidence yet; $20 a trade at most. - **RSI(2) pullbacks** (`rsi2_reversal`, REJECT): failed G1, G8. - **Sector rotation** (`sector_momentum`, REJECT): failed G1, G3, G8, G9. - **Turn of the month** (`turn_of_month`, REJECT): failed G1, G2, G8, G9. - **Overnight only** (`overnight_drift`, REJECT): failed G1, G2, G7, G8. Machine-readable: the same content as `brief.json` (schema above).