HYPE Index

Dashboard

Read HYPE Index

Reading guide for the HYPE Index chart — every signal, every lesson from validation, and the trading rules that survived the autoresearch loop.

1 · Chart orientation

The HYPE Index chart layers four kinds of information on a single price line. Knowing what each layer answers makes the rest of this guide click faster:

Layer	What it answers	Visual
Price line	Where HYPE traded	Green→red gradient by 1D regime
Triangles	Did a tradeable signal fire today?	▲ green/cyan/blue (longs) · ▼ red (shorts)
Diamonds	Did a high-conviction divergence form?	◆ purple (+div) · ◆ blue (−div) — late-cycle only
Context markers	What's the higher-timeframe regime saying?	🔥 fresh flip · ELITE 🔥 (purple) · ⚠️ late-cycle warning
Reference lines	What level would form a new pivot if broken?	Dotted ▲ blue (recent high) / ▼ purple (recent low)

The legend at the bottom of the chart counts each marker for the visible window. The hover panel below the chart shows the per-day breakdown when you mouse over.

2 · Real-time vs delayed signals

Not every marker reflects today. Action triangles fire on the day the condition is met (end-of-day close); divergence diamonds and late-cycle warnings have structural lag. The shaded zone on the right edge of the chart marks the unconfirmed-pivot zone — the last 3 bars cannot carry a confirmed diamond.

Signal	Lag	Why
▲ BUY / ▲ PULLBACK / ▲ LIGHT BUY	0 days	Edge event on bb_pos crossing — fires same day on close
▼ SELL WARN / ▼ SELL EXIT	0–3 days	Needs neg-div confirmation; pivot lag drags it
1D regime band	0 days	MA12/20/50/100 on daily fees — today's close, today's classification
3D regime band · 🔥 fresh flip	0–2 days	Aggregates 3 days of fees; current bucket isn't classified until day 3 closes
◆ Divergence diamonds	3 days	Pivot detection requires 3 forward bars to confirm extremum
⚠️ Late-cycle BULL/BEAR	0–6 days	Streak counter ticks only at week-close (Sunday→Monday)
★ Bright-tier diamond (compound)	3 + 0–6 days	Pivot lag stacked with weekly streak lag

💡

Trading-time rule: trade triangles on the day they fire. Treat diamonds and late-cycle warnings as confirmation, not entry triggers — the trade has usually been visible to triangles first.

3 · Regime states (1D / 3D / weekly)

Three timeframes of the same fee-MA classifier stack on top of each other. Each runs the same rule (short MA above/below long MA in two tiers) on a different aggregation:

Timeframe	Aggregation	MA periods	Updates
1D	Daily fees	12 / 20 / 50 / 100	End of every day
3D	3-day fee buckets	4 / 7 / 17 / 33 (scaled)	End of every 3rd day
Weekly	7-day fee buckets	4 / 7 / 17 / 33 (scaled)	End of every Sunday

The 5 states (color-coded on the price line)

BULL — both MA pairs bullish. Fees expanding on both timeframes.
RECOVERY — short MA flipped up, long still down. Early reversal.
TOPPING — short flipped down, long still up. Distribution risk.
BEAR — both pairs bearish. Capitulation territory.
NEUTRAL — MAs incomplete (warmup or transition).

💡

Validated: bull-flavor signals fired with 3D in BULL or RECOVERY family returned +30.6% / 82% win on n=21 vs +9.8% / 50% when 3D opposed (BEAR or TOPPING). Use 3D alignment as a size-up filter, not a skip filter.

4 · Buy signals (▲ triangles)

▲ BUY · solid green

1D = BEAR AND bb_pos ≤ −0.8. Fires on the edge — first day the condition becomes true. Variant-C cluster filter routes only the first BUY of each BEAR run into the actionable count; subsequent re-fires render as faint outline triangles (suppressed).

thresholdbb ≤ −0.8Apr 6 2025 case+97.8% fwd30 (Elite override: when the suppressed BUY day also has a fresh 3D BULL/RECOVERY flip, the suppression is lifted)

▲ PULLBACK · solid cyan

1D = BULL or RECOVERY AND bb_pos ≤ 0 AND RSI < 45. Catches uptrend dip-buys.

▲ LIGHT BUY · solid blue

1D = BULL or RECOVERY AND bb_pos crossed up through −0.5 AND price up DoD. Lower-conviction tier than full BUY/PULLBACK — softer pullback bounce.

▲ Suppressed BUY · faint outline

BUY that would have fired but was gated by variant-C. Cosmetic only — not actionable. Tells you re-fires are happening but the system already gave you the entry on the first one.

▲ Low-conviction BUY · dashed outline

Fired at bb ≤ −1.0 but failed the gap+reset discipline check (price not ≥10% below the previous BUY's anchor, anchor resets on bb ≥ +0.5). Phase 5 evidence: this tier captured the Sep 23 / Oct 11 2025 losers (−12.9% / −11.1% fwd30) while leaving Aug 19 / Nov 22 / Jan 9 / Feb 6 winners as solid BUYs.

📐

Buy-side rule of thumb: all three solid triangles are tradeable. Outline / dashed triangles are diagnostic — they tell you why a BUY didn't promote, not that you should enter.

5 · Sell signals (▼ triangles)

▼ SELL WARN · light red

1D = BULL or TOPPING AND bb_pos ≥ +0.7 AND live neg-div. Tighten stops, partial exits.

▼ SELL EXIT · solid red

1D = BULL or TOPPING AND bb_pos just crossed back below 0.7 (was ≥0.7 in last 5 days) AND live neg-div. Confirmed reversal — full exit.

★ SELL quality tiers

Tier 0 · regular sell — no extra tag
Tier 1 ★ · SOLID — fresh −div within 7d OR consecutive 2+ in past 30d
Tier 2 ★★ · DOUBLE-SOLID — both criteria met

⚠️

Known limitation: SELL signals only fire in BULL / TOPPING regime. Fast crashes (Oct 29 2025 case) can flip 1D to BEAR before the SELL window closes — the trade was a winner (−37.6% in 24d) but no SELL triangle ever printed. Use NEG_DIV bright tier (covered next) as a backstop.

6 · Divergence diamonds (◆)

Regular RSI / MFI / MACD / MACD-Hist divergences detected at price pivots. The chart applies an aggressive filter — only bright-tier divergences render, because validation showed dim-tier diamonds are mostly redundant with the BUY/SELL triangle network.

The bright-tier gate (3 conditions, all required)

Tier-2: divConsec ≥ 2 OR indicatorCount ≥ 2
Led OR bb-extreme: divLedToTrade !== undefined OR (NEG_DIV only) bb ≥ +0.7
Late-cycle weekly streak ≥ 4wk · direction-aware — POS_DIV reads secularBearWeeks; NEG_DIV reads secularBullWeeks

Validated win rates

Bucket	n	Win	Mean fwd30
POS_DIV bright (late-cycle BEAR ≥4wk)	3	100%	+29.9%
NEG_DIV bright (late-cycle BULL ≥4wk)	6	83%	−19.9%
POS_DIV info-only (hidden)	5	20%	+6.3% (noise)
NEG_DIV info-only (1 case shown via bb-extreme)	1	100%	−27.2%

📐

Why the asymmetry? Symmetric application (bb-extreme on both sides) would resurface 4 POS_DIV losers (−31.5% / −17.0% / −17.9% / −8.1%) to gain 0 winners. NEG_DIV info-only only had 1 historical case (Oct 29 2025), which won big. So the bb-extreme alternate gate applies to NEG_DIV only.

Phantom (candidate) diamonds

Hollow dashed-outline diamonds in the rightmost 3 bars (the shaded "pivot pending" zone) mark candidate pivots — bars currently sitting at a local high or low vs the prior 7 bars. If the extremum holds 3 more bars AND a real divergence forms in RSI/MFI/MACD, the phantom solidifies into a real diamond. If a deeper extreme forms, the phantom disappears. Don't enter on phantoms — they're previews.

7 · Context markers

🔥 Fresh 3D flip (orange flame)

Fires on day-0 of a 3D regime change. Validated as the strongest single signal-quality discriminator: bull-side signals fired ≤3 days after a 3D BULL/RECOVERY flip went 9-for-9 / +38.8% mean fwd30; bear-side signals after 3D BEAR/TOPPING flip went −16% (vs −10.8% baseline).

🔥 ELITE trifecta (purple flame)

Fires when a 1D signal AND a fresh 3D flip (≤3 days) AND 3D in BULL/RECOVERY family all align on the same day. The validation's highest-conviction trades:

Apr 6 2025 BUY · +97.8% fwd30
Apr 9 2025 LIGHT_BUY · +71.2% fwd30
Jan 13 2025 PULLBACK · +23.5% fwd30

📐

Variant-C Elite override: a would-be-suppressed BUY day with Elite trifecta is promoted to a real BUY (the Apr 6 +97.8% case). Without this override, the cluster filter would have hidden the entry.

⚠️ Late-cycle BULL warning

Renders when secularBullWeeks ≥ 8 on a bull-flavor signal day. Validated pain: bull-flavor signals fired after 8+ weeks of weekly BULL underperformed earlier-cycle fires by −36.3pp on mean fwd30 (n=3, mean −10.5%).

⚠️ Late-cycle BEAR warning (mirror)

Renders when secularBearWeeks ≥ 8 on bear-flavor signals. Theoretical mirror — limited HYPE history. The MORE interesting mirror finding: when weekly BEAR ≥ 4wk AND a BUY-flavor signal fires, the bear is exhausted — 100% win on n=5, +21.2% mean. This is what the bright-tier POS_DIV captures.

8 · Conviction stacking

The autoresearch repeatedly showed that no single signal is high-confidence in isolation. Conviction comes from layering — and the layers are independent enough that agreement is informative.

The conviction ladder (long side)· validated as-of 2026-05-10

Tier	Signature	Expected behavior
5 · Elite trifecta	1D BUY/PULLBACK/LIGHT_BUY + 🔥 fresh 3D flip + 3D ∈ BULL/RECOVERY	100% / +39.2% on n=8 (range +14% to +104%)
4 · Bright +DIV alone	Late-cycle BEAR ≥4wk + tier-2 +DIV that led to BUY	100% / +29.9% on n=3
3 · 3D-aligned BUY	Solid BUY/PULLBACK/LIGHT_BUY + 3D ∈ BULL/RECOVERY (no fresh flip)	88% / +22.5% on n=16 (gap to opposed has narrowed — see note)
2 · Solid BUY alone	Solid triangle without HTF context	~baseline (mid-teens)
1 · Suppressed / low-conv	Outline / dashed triangle	diagnostic only — don't trade

⚠️

3D-alignment finding has degraded. Original validation (n=21) showed aligned bull-flavor +30.6% / 82% vs opposed +9.8% / 50% — a 20pp+ gap on both axes. Re-run on 2026-05-10: aligned 88% / +22.5% (n=16) vs opposed 71% / +27.5% (n=7). Win rate still favors aligned, but opposed mean PnL flipped higher — driven by Apr 2 BUY (+76.8%), Feb 23 BUY (+52.9%), and the late-cycle-BEAR cluster (Jan 16/20/24, all opposed but caught the bottom). Treat 3D-alignment as a win-rate filter, not a mean-PnL filter, on current data.

The conviction ladder (short side)· validated as-of 2026-05-10 (unchanged)

Tier	Signature	Expected behavior
4 · Bright −DIV + late-cycle BULL	Late-cycle BULL ≥4wk + tier-2 −DIV (5–10wk window optimal)	83% / −19.9% on n=6
3 · ★★ SELL EXIT (double solid)	Tier-2 SELL with fresh −div AND consecutive 2+	strong reversal confirmation
2 · ★ SELL EXIT (solid)	Tier-1 SELL with one −div criterion	moderate
1 · SELL WARN alone	Without divergence quality tier	noisy — wait for EXIT

💡

The 5–10wk sweet spot: on the short side, late- cycle BULL streaks at 5–10wk catch tops as they form (leading by 7–28 days). Above ~12wk, the streak counter is lagging — regime is already rolling over. So bright −DIV at 5–10wk beats bright −DIV at 15wk.

9 · What's next

That covers reading the chart. The trading playbook — synthesized rules, real backtest cases, methodology lessons, and caveats — lives on its own page so this one stays focused on visual interpretation.

💡

→ Continue to How to Trade · 8 trading rules · 11 backtest case studies (the +97.8% Apr 6 trade, the −37.6% Oct 29 short, the LIGHT_BUY cascade disaster, etc.) · methodology caveats & sample-size honesty.

← Back to HYPE Index