S&P Daily Predictor Backtest Arena

=== ANALYTICAL STRATEGY INSTRUCTIONS ===

1. MACRO CATALYST EXTRACTION (FILTER AND CONFIDENCE ADJUSTMENT ONLY)
- Build a live catalyst board before 09:30 ET and maintain it until at least 11:00 ET: scheduled U.S. economic data, Fed speakers/FOMC events, Treasury auctions/SOFR operations, major SPY constituent earnings, and geopolitical/foreign news with plausible SPY impact. Refresh every 15 minutes in the first 90 minutes; after that, refresh only when a genuinely new catalyst appears.
- For each catalyst record: Time ET | Event | Consensus vs Prior | Actual (if released) | initial SPY/NQ futures reaction | transmission through VIX/rates/DXY/oil/credit | status | directional bias | SPY bps impact estimate.
- Impact size: HIGH >25 bps, MED 10-25 bps, LOW <10 bps. Judge impact only from market reaction within the first 5 minutes, not from headline tone.
- Status definitions:
  * PENDING: scheduled within next 60 minutes and no market reaction yet.
  * ACTIVE: release/headline occurred and SPY/futures are still reacting with transmission follow-through in the last 20 minutes.
  * PRICED-IN: no fresh reaction for 20+ minutes, initial move retraced <30%, current 5-min SPY volume <=50% of opening 15-min average, and no transmission follow-through.
  * NEUTRAL: no measurable SPY effect.
- Freshness decay: after the initial reaction, cut impact by 50% after 30 minutes and to 25% after 60 minutes unless renewed by transmission follow-through or a valid 5-min close through a pre-identified level in the catalyst direction. Never give stale catalysts full weight.
- Numeric macro lean: HIGH risk-on +1.0, MED risk-on +0.5, LOW risk-on +0.2; HIGH risk-off -1.0, MED risk-off -0.5, LOW risk-off -0.2. Sum only ACTIVE/PENDING catalysts after freshness decay. Macro lean is risk-on if net >= +1.5, risk-off if <= -1.5, otherwise neutral.
- Event-risk rule: if a HIGH-impact catalyst is within 60 minutes, drop any directional contribution from it and reduce confidence by 10 until released. After release, use the catalyst only if the actual surprise is >=0.5 SD versus consensus and SPY confirms with a 5-min close through a pre-identified level in the same direction. Otherwise mark PRICED-IN/NEUTRAL and do not chase.
- Macro is always a filter or confidence modifier, never a standalone trade generator. If macro and technicals conflict, reduce confidence instead of forcing a trade.

2. TECHNICAL LEVEL SIGNALS
- Before 09:30 ET, identify and record: prior day high/low/close, overnight high/low, premarket high/low, anchored VWAP and its 15-min slope, high-volume nodes (>=5M SPY shares per 0.1% price bucket), and the 09:30-09:45 OR high/low. Ignore round numbers.
- Treat tightly clustered levels within 0.15% as one zone. A level is strong if defended/created on above-average volume, tested at least twice, or in confluence with VWAP/high-volume node. A single-touch level is weak and needs one additional confluence before a break is valid.
- Gap size versus prior close: strong gap >40 bps, moderate 15-40 bps, inside <15 bps.
- Valid trigger: a 5-minute close beyond a pre-identified level by >=0.10% (ORB levels: >=0.15%), with RVOL >1.1 (ORB: >1.2), and at least one confluence: another level within 0.15%, VWAP in the break direction, breadth >55% long or <45% short, or an aligned macro lean. A wick or 1-minute close is never a trigger.
- ORB applies only 09:30-11:00 ET. Requires all three: 5-min close beyond OR by >=0.15%, RVOL >1.2, and breadth >55% (or <45%) in the breakout direction. Missing any element means assume fade risk and do not chase.
- VWAP is the control. Above VWAP, prefer longs only on a VWAP hold; below VWAP, prefer shorts only on a VWAP rejection. A VWAP breakout requires a 5-min close through VWAP, RVOL >1.1, and two consecutive 5-min closes in that direction. A 1-minute one-touch is not a break.
- If the nearest pre-identified level is >0.20% away and VWAP slope is flat (<0.05%/hr), stand aside. Do not invent levels.
- After 11:00 ET, discard untested morning levels; use fresh VWAP/afternoon pivots and reasserted prior day/overnight levels only.
- A level is invalidated by a 5-min close back through it; do not count it again until a new valid trigger forms.

3. MOMENTUM VS GAP-FILL BEHAVIOR
- Every 5 minutes update: SPY RVOL versus the same 5-min window of the 5-day average, % of SPX members above VWAP, cumulative NYSE A/D, number of SPY sectors above VWAP, QQQ/SPY and ARKK/SPY relative strength, VIX level and 5-min change.
- Valid up momentum requires breadth >55%, at least 4 sectors advancing, and high-beta (QQQ/ARKK) >= SPY. Valid down momentum requires breadth <45%, at least 4 sectors declining, and high-beta <= SPY. If momentum is valid, do not fade solely on a single divergence; record divergence as a -5 confidence adjustment only.
- Gap-fill protocol:
  * Gap >0.5%: expected to fill unless RVOL >1.2, breadth >60% in the gap direction, and leading sectors agree; if those hold, favor continuation to the next level.
  * Gap 0.15-0.5%: no default assumption; wait for the first 30-minute behavior and require a valid trigger before taking direction.
  * Gap >1%: no mean-reversion and no continuation without a clean 5-min close through a pre-identified level on RVOL >1.2 with breadth confirmation. Otherwise remain neutral.
  * If price reaches the prior close within 60-90 minutes and stalls >10 minutes without snapback, mark the gap filled and use the prior close as the new pivot.
  * If price holds away from the prior close with RVOL >1.2 and breadth in the gap direction, do not expect a fill; look for continuation on a 5-min close through the nearest level.
- Regime classification in the first 30 minutes:
  * Trend: opens near one side, breaks OR quickly, holds VWAP, makes higher highs/lows on 5-min closes. Bias continuation until a valid VWAP/level break against the trend.
  * Trend-with-retest: gap/opening move, one VWAP retracement, then reclaim/rejection with RVOL >1.0 and breadth confirmation. Act only after a 5-min close through VWAP/level.
  * Reversal: loses VWAP after the opening move, fails to hold a breakout level, and breadth flips >10 points. Expect return to prior close/VWAP.
  * Chop: overlapping 5-min ranges, fading RVOL, no clean break. No edge; remain neutral.
- Invalidation: any 5-min close back through the trigger level cancels the setup. Re-entry requires a new valid trigger at another pre-identified level or VWAP retest with momentum confirmation.
- Open-to-close inference: if an unfilled gap remains at 10:15 with strong momentum/RVOL, continuation into the close is more likely. If the gap fills before 10:15 or price stalls near prior close, favor range-bound/neutral unless a new VWAP/level break appears.

4. CONFIDENCE CALIBRATION
- Start at 50% neutral. A directional bias is allowed only if there is at least one valid technical trigger or a high-conviction macro-technical confluence. No confluence = neutral.
- Use integer adjustments for independent, visible evidence; do not double-count.
  * +8 for aligned HIGH ACTIVE/PENDING macro catalyst with technical direction.
  * +4 for aligned MED ACTIVE/PENDING macro catalyst with technical direction.
  * -8 for a HIGH ACTIVE/PENDING macro catalyst directly conflicting with technical direction.
  * -4 for a MED ACTIVE/PENDING macro catalyst directly conflicting with technical direction.
  * +10 for a valid 5-min close through a pre-identified level that holds for at least 5 minutes.
  * +10 for full momentum confirmation: RVOL >1.2, breadth >60% long or <40% short, sector alignment.
  * +5 for ORB confirmation before 11:00 with all three momentum elements.
  * +5 for VWAP direction reinforced by at least one other level.
  * -10 for HIGH-impact catalyst within 60 minutes (event risk).
  * -10 for mid-range price with no broken level and flat VWAP slope.
  * -10 for gap >1% lacking RVOL >1.2 and breadth >60% in the gap direction.
  * -5 when price is within 0.1% of VWAP and no trigger has triggered.
  * -5 when all catalysts are PRICED-IN/stale and no fresh catalyst is developing.
  * -5 for momentum divergence against the direction.
  * -5 when a post-release level break fails after an in-line surprise.
- After adjusting, write the strongest bullish and strongest bearish scenario for the rest of the session. If the counter-story is material or the bull/bear cases are nearly balanced, reduce confidence by 5 and shift toward neutral.
- Cap directional confidence at 75%. Interpretation: 70-75 high confidence with strong directional bias; 62-69 medium-high confidence, trade only if the setup is clean; 55-61 medium confidence, require multiple confirmations; 50-54 low confidence, stay neutral unless a clean setup appears.
- A non-neutral final call requires confidence >=62% AND expected open-to-close SPY move >=15 bps in that direction. Otherwise call neutral.
- Make the final open-to-close call no later than 10:15 ET. If no edge, forecast neutral. Never manufacture a call to match a prior guess.