What is this system?

This is a real-time options market intelligence tool built specifically for trading SPY intraday. It combines two data streams that most retail traders watch separately:

Most tools give you one or the other. The edge here is cross-referencing them: when options flow says bearish AND price confirms it on the chart, that's a much higher-conviction trade than either signal alone.

How the data flows

Understanding the data pipeline helps you know how fresh each number is:

Your trading workflow

Here's how to use the pages in sequence during a trading session:

Flow Dashboard

The most important numbers from the options flow, updated every poll. Think of this as your market sentiment gauge.

The 6 KPI numbers

If Data Quality goes red, treat the institutional bias as missing for that poll and rely on the passive (C/P) read alone. The dashboard auto-degrades the combined bias to half-strength in that case.

Bias pill in the navbar — four states

The big bias pill at the top of every page is the combined bias — passive C/P skew AND institutional flow, fused into one verdict. It has four possible states:

Hover over the bias pill to see both underlying signals in the tooltip — e.g. "Passive (C/P): BULLISH | Institutional: BEARISH" tells you exactly why the combined pill shows MIXED.

Flow Bias Streak

The coloured dots show the last 7 polls of combined bias. Green = BULLISH, red = BEARISH, amber = MIXED, muted = NEUTRAL. When the same colour appears 3+ times in a row, the bias has been sustained for 15+ minutes — that's meaningful. A streak only counts when BULLISH or BEARISH; the streak counter does not count MIXED or NEUTRAL polls, because those aren't directional reads.

Signal Confluence panel

The confluence panel now shows the full picture in seven rows:

Border colour: green = bullish, red = bearish, amber = MIXED/NEUTRAL. When Inst and Passive both show the same colour border, conviction is high. When one is amber and the other green/red, that's the moment to be cautious — the system itself is undecided.

Flow Map

A dedicated observation surface for watching options flow concentrate and migrate across the session. Time runs left → right (one column per 5-min poll), strike runs top → bottom, and each cell is coloured by that strike's flow at that poll. It answers one question: where is flow building, and is that band moving toward or away from price?

Hero — the Build Map

The heatmap grid. A white line traces spot through it; a blue line marks the cursor's poll. Brighter cell = more concentrated flow at that strike. Cells flagged as peaks get an amber outline. The toggle (top-right) switches what the colour means:

Companion — Building Now

Below the map, for the cursor's poll: one horizontal bar per strike showing the flow change (Δ) this 5-min window — gaining to the right/teal, bleeding to the left/rose. The spot strike row is highlighted and flagged peaks are marked. This is the de-diluted "what's accumulating right now" read — the antidote to the cumulative bias drifting to neutral.

Scrub & live

The page defaults to live — the cursor is pinned to the latest poll and a new column appears at the right each cycle. Drag the slider (or click the map) to scrub back through the session; the Building-Now panel follows the cursor. The ● live button re-pins to the latest poll. Hover any cell for strike, flow, Δ, share % and momentum.

Read flow in regime context: the per-poll regime label tells you whether "building" means a likely pin (RANGE) or fuel for a move (TREND). The same bright band means different things on a TREND day vs a RANGE day.

GEX Heatmap

This is the structural map of the options market. It shows where market makers have the most hedging pressure concentrated, across 7 expiry dates and every relevant strike.

Reading the colours

The Δ (delta) column

The column next to each GEX value shows the change since the last poll. ▲ +2.1B means $2.1 billion of additional put GEX built at that strike in the last 5 minutes. This is the velocity signal — not where the wall is, but how fast it's growing. A strike that goes from -$500M to -$2.5B in 10 minutes is a major positioning event.

The purple outlier highlight

Cells with a purple border contain GEX values more than 2× the local median of the 5 nearest strikes. These are the statistically significant levels — where positioning is unusually concentrated. These are also the strikes that appear as arrows on the Chart page.

Controls

• ▸Strikes ± slider: Default ±10 around spot. Widen to ±25 to see the full structure, narrow to ±5 to focus on the immediate range.
• ▸Min |GEX| filter: Use $5M+ to cut noise, $50M+ to see only the dominant walls.
• ▸Outlier threshold: 2× is the default. Set to 1.5× to catch more, 3× to see only the truly dominant levels.

Chart + Signal Panel

Timeframe buttons (5m / 15m / 30m / 1h)

Switching timeframes reloads all indicators and rescores the signal panel. The power is in multi-timeframe confluence:

GEX markers on the chart

The ▼ red arrows mark moments when a put-heavy outlier strike was detected. The ▲ green arrows mark call-heavy outliers. When price is approaching a red arrow level and a new red arrow fires just below — that's the put wall asserting itself in real time. Watch for price to bounce or find support there.

Indicator reference

Magnets & Signals

This page has three tabs — Magnets, Signals, and Log. Magnets are structural GEX levels that update every poll. Signals are triggered events with a full lifecycle. The log is your permanent weekly audit trail stored in browser localStorage.

Two scenario columns

The page now displays only directional scenarios (Bearish and Bullish). The third Gamma Neutral / Pin column was removed after backtest data showed neutral signals hit only 27% vs 69%+ for directional setups — pin theory simply does not predict 5-60 minute moves reliably in this market structure.

On range-bound days (a range regime badge here, or RANGE DAY on the dashboard), don't disable the system — read both columns and expect price to oscillate between the top bearish and bullish strikes. The pinning behavior is now visible on the chart rather than as a separate signal column.

Each strike box shows

• ▸Strike + spot distance — how many points from current spot
• ▸GEX + delta — ▲ building $120M = new positioning added. ▼ unwinding = positioning leaving.
• ▸Momentum dot — building (strong), unwinding (weakening), stable
• ▸Role badge — TARGET, FLOOR, CEILING, or PIN

Signal lifecycle

Signals are directional only (bearish or bullish). Neutral/pin scenarios were removed after the 2026-05-19 backtest showed only 27% hit rate vs 69%+ for directional. The lifecycle is:

Hysteresis: A signal fires at 50% but only collapses below 40%. This prevents churn when a borderline signal oscillates around the threshold.

Dynamic max-age: A signal's lifespan scales with target distance and regime, rather than a flat cap. Base 90 min + 45 min per point of distance, multiplied ×1.3 on a RANGE day (moves are slow) or ×0.8 on a TREND day (moves are fast), capped at 240 min. So a 1.7pt target on a RANGE day gets ~190 min, while a 0.5pt target on a TREND day gets ~108 min. This fixed two real cases (2026-05-20 $742 and 2026-05-21 $745) where correct signals with 1.4–1.7pt targets aged out at the old flat 120-min cap and were mislabeled as failures — even though price reached target shortly after.

EXPIRED vs COLLAPSED: These look similar but mean opposite things. EXPIRED = "thesis was still good, we just stopped watching" (conviction ≥ 60% at age-out). COLLAPSED = "thesis broke" (conviction fell below 40%, or aged out weak). Only COLLAPSED counts against your accuracy stats. EXPIRED is excluded entirely — it's neither a win nor a loss. On the Accuracy page, expired signals show an amber ⏱ EXPIRED badge and are removed from the hit-rate denominator.

Refire cooldown: After any resolution (HIT, EXPIRED, or COLLAPSED), the same (scenario, target) pair is locked out for 15 minutes. This prevents the duplicate-firing bug seen on 2026-05-19 where bearish $733 fired three times in the same 5-minute move because the resolved signal was deleted between polls. Different scenarios or different targets are unaffected — only the exact same setup is suppressed.

Open gate: No signal fires before 9:35 ET. The first few minutes after the 9:30 open, the upstream price feed can still serve the prior session's close, which would compute spot/distance/VWAP against a stale number and fire a bogus signal (observed 2026-05-21). The gate suppresses fires until the feed settles.

Signal accordion — expanded view

Click any signal card to expand and see the full reasoning:

• ▸Why it fired — each factor scored individually: flow bias streak, C/P ratio, net gamma, IV level, GEX momentum, VWAP position. ▼ = bearish, ▲ = bullish, · = neutral.
• ▸GEX context — the specific wall, distance from spot, momentum at fire, VWAP at the moment it fired
• ▸Conviction history — mini chart showing conviction evolution across the last 8 polls

Log tab — weekly storage

Stored in browser localStorage keyed by ISO week: spy_log_week_2026-W20. On Monday morning the poller detects your IP, writes an export trigger, and the browser automatically POSTs the previous week's log to the local bridge server — saved to F: radingiles\logs\spy_log_YYYY-WNN.json.

Accuracy

The scoreboard. It takes every signal the Magnets page fired (from your browser log) and checks, against Yahoo 1-minute bars, whether price actually reached the target — within 5, 15, 30, and 60 minutes. This is where you find out whether the signals are worth trading.

What counts, and what doesn't

• ✓HIT — price touched the target (±$0.50) inside the window. Counts as a win.
• ✗MISS — the window expired without a touch. Counts as a loss.
• ⏱EXPIRED — the signal aged out while conviction was still healthy. Thesis intact, clock ran out. Excluded from the hit-rate denominator entirely — neither win nor loss.
• …pending — the windows haven't all elapsed yet. Doesn't affect the numbers until it resolves.

Only fully-resolved signals are scored. This page is the empirical proof behind the entry rules — MIXED-bias and low-conviction setups should show a visibly lower hit rate here, which is exactly what justifies skipping them.

How the page is broken down

Use the Date dropdown to load a prior log week. ↓ Export all logs downloads the raw signal log as JSON.

Login & Legal

Every new browser session redirects to the login page before any content is accessible. Three mandatory checkboxes must be ticked:

• ▸Terms of Use — informational only, not financial advice, sole responsibility for trading decisions
• ▸Privacy Policy — no personal data collected server-side; localStorage used only for signal log
• ▸Risk disclosure — options trading carries substantial risk; past signal performance does not guarantee future results

Agreement uses sessionStorage — resets when the browser tab is closed. Every new trading session requires fresh agreement. This is intentional.

Gamma Exposure — deep dive

GEX is the most important concept in this system. Here's the full picture:

What gamma means

Every option has a delta — how much the option moves per $1 move in SPY. Gamma is how fast that delta changes. When SPY moves, delta shifts, and dealers who sold those options must buy or sell SPY to stay delta-neutral. That forced hedging is mechanical, predictable, and large enough to move price. GEX measures the size and direction of that pressure across the whole chain.

GEX formula

What the sign tells you

The sign of net GEX tells you how price will behave today — not which direction it will go, but whether moves will trend or chop. This is the single most important regime read in the system.

Gamma walls in practice

Individual strikes with large GEX values act as magnets. A strike with -$4.4B GEX (like $738 in your data) means that $4.4 billion in MM hedging pressure is anchored at that strike. When price approaches, the mechanical hedging creates a gravitational pull. Price doesn't always reach the wall — but the probability is higher than random.

The interaction between regime and walls is what makes this useful:

• →In positive GEX regimes, walls act as pins — price oscillates around them and gets stuck. Trade fades and mean-reversion into the wall.
• →In negative GEX regimes, walls act as acceleration points — once price breaches a wall, dealer rehedging cascades and the move extends. Trade the breakout, not the fade.
• →The largest opposite-sign wall above/below spot is your most likely magnet for the session. Map it before the open and treat it as your primary target.

Options Flow — what it tells you

Call premium vs put premium

The raw dollar value of options traded matters more than the number of contracts. A single $50M put sweep tells you more than 500 small call trades. Watch for premium imbalances — when put premium exceeds call premium by 2× or more, large players are paying up for downside protection.

C/P ratio interpretation

IV (Implied Volatility) trend

Rising IV means options buyers are paying more — they're scared or positioning for a big move. Falling IV means complacency. The most powerful setup is rising IV alongside rising put premium alongside bearish flow bias — all three together mean real institutional hedging, not noise.

Two-signal Bias Model

The bias displayed across the system (the navbar pill, the confluence panel, the magnets signals) is not one indicator — it's a combination of two independent readings. Understanding what each one measures, and what their agreement or disagreement means, is the most important concept in this system.

Why two signals instead of one

Earlier versions of this tool used a single bias derived from the call-vs-put premium ratio. That worked, but it produced false confidence: the C/P ratio drifts slowly (it was 0.93 ± 0.01 for an entire 65-minute session on 2026-05-19) and the system would call that "BULLISH × 13 polls" — confident, persistent, and not actually reflecting anything new happening in the tape.

The new model adds a second, independent signal that watches what institutions are actually doing right now, not what the chain looks like in aggregate. When the two signals agree, that's real conviction. When they disagree, the system flags it as MIXED — and that's the most useful new information, because it tells you when not to trade.

Signal 1 — Passive premium skew (flow_bias_premium)

Just the ratio of put premium to call premium across all trades in the last 5-minute poll. This is the "broader chain" view — every trade counts, retail and institutional, hedges and directional, mid-fills and aggressive sweeps. It changes slowly and reflects the overall state of the option chain.

The bearish threshold is asymmetric (1.10 vs 0.95) because put protection is structurally bid at all times — institutional hedges are continuously bought even on quiet rallies. It takes a stronger put skew to indicate genuine bearish demand.

Signal 2 — Institutional aggression (flow_bias_institutional)

This is the new signal that reflects what large players are doing right now. It works in two steps:

Step A — filter to institutional trades. Each flow record has an orderType field from the data provider:

Step B — Lee-Ready NBBO classification. For each institutional trade, compare the executed price to the NBBO midpoint:

• Price above midpoint (+ 5% spread tolerance) → buyer-initiated (paid up to be filled)
• Price below midpoint (− 5%) → seller-initiated (gave size, wanted out)
• Price within tolerance → mid-fill, not classifiable as either side

Then map to underlying direction:

If fewer than 5 institutional trades qualify in a poll (slow market, lunchtime), the institutional bias is reported as NO_DATA rather than guessing from a tiny sample. The combined bias falls back to passive-only at reduced strength.

The agreement matrix

The combined bias (flow_bias_combined, aliased as net_flow_bias) is computed by comparing the two signals:

Reading the MIXED state — what it actually means

MIXED is the system telling you: "Smart money is doing something different from what the broader chain looks like." A few common patterns:

• Passive BULLISH, Inst BEARISH — the chain skews call-heavy (often because of retail call buying or covered-call writing), but institutions are aggressively buying puts. Classic distribution / institutional hedging on a rally. Don't chase the upside.
• Passive BEARISH, Inst BULLISH — put premium is elevated (broad hedging), but institutions are aggressively selling those puts to retail and buying calls. Could be a bottom forming, or smart money fading a fear spike. Don't short.
• Either MIXED state during a fast move — typically transitional. The "winner" of the disagreement usually leads the next direction. Watch the institutional bias closely; it tends to lead.

The simple rule: do not enter new trades when the combined bias is MIXED. Wait for the two signals to align. Existing positions can stay but treat MIXED as a tightening-stops signal, not a "load up" signal.

The "buyer interpretation" caveat

Lee-Ready classifies based on who paid up. A call buyer is classified as bullish — but the other side of every trade is someone selling that call. Are they bearish? Not necessarily. Market makers sell calls to delta-hedge with stock; that's mechanically neutral on the underlying. So the institutional signal really measures aggressive intent on contracts, not pure long/short conviction. For most setups this is the signal you want, but be aware: a session with heavy MM call-writing (and corresponding stock buying for hedge) can read BEARISH on institutional flow while stock grinds up. That's actually a useful signal — it tells you "MMs are absorbing this rally."

Regime — TREND DAY vs RANGE DAY

The Regime label sits at the top of the Signal Confluence panel and answers one question: "Will dealers absorb moves today, or amplify them?"

The system reads regime two ways — don't confuse them

"Regime" appears in two places, computed differently. Both answer "will dealers absorb moves or amplify them?", but one is a broad daily posture and the other is a live, structural read around spot.

1 · Dashboard regime — 7-day net GEX

Broader and steadier than the per-poll net_gamma in the Net Gamma KPI box (which only reflects what just traded).

2 · Magnets & Flow Map regime — per-poll structural classifier

Computed each poll from the ±5-strike GEX ladder (OI-GEX when fresh, otherwise flow-GEX). It asks two questions about where spot sits in the gamma structure, not what the broad daily total is:

• Is GEX at the spot strike positive or negative? Positive = dealers long gamma at spot (damping). Negative = a short-gamma "pit" (amplifying).
• Is spot above or at/below the zero-gamma flip? The flip is the highest strike where the cumulative stack turns net-negative. Above it = damping territory; at/below = amplifying territory.

This is why a day can read RANGE DAY on the dashboard (positive 7-day posture) while a given poll shows trend on the Magnets page — spot has dropped into a local short-gamma pit even though the broad book is long gamma. They answer different questions.

Why the label is white, not red/green

Regime is not directional. A TREND DAY can trend up or down; a RANGE DAY can pin above or below your position. Colouring it green/red would conflate regime with bias. The label is intentionally white so it reads as context, alongside the directional bias which has its own colours.

How to use regime in decisions

• TREND DAY + clear bias — best setup for directional trades. Use wider targets, let winners run, expect follow-through.
• RANGE DAY + clear bias — fade extensions toward the bias. Tight targets at GEX walls. Don't expect breakouts to stick.
• TREND DAY + MIXED bias — high risk. The market can move violently, but the system can't tell you which way. Stay out.
• RANGE DAY + MIXED bias — chop. Expect range-bound noise. Easy day to overtrade and bleed.

Chart Indicators — how they interact

EMA stack hierarchy

The three EMAs form a stack. When EMA 9 > EMA 21 > EMA 50, price is in a bullish alignment. Reverse stack = bearish. The crossovers (EMA 9 crossing EMA 21) are the trigger signals — but only act on them when the larger context (flow, GEX) agrees.

OBV divergence — the key volume signal

On-Balance Volume is the cumulative flow of money in and out of SPY. It rises on up-volume bars, falls on down-volume bars. The signal that matters most:

MFI vs RSI

MFI (Money Flow Index) is RSI weighted by volume. The key difference: RSI can signal overbought even if the move was on massive volume (which is sustainable). MFI won't signal overbought unless the volume confirms the overextension. For SPY — a heavily traded instrument — MFI gives fewer false signals.

Volume Ratio spikes

When a bar prints at 2.5× average volume, something significant happened. Combined with a GEX marker firing at the same time, you know: (1) a major level was being tested, and (2) it happened with real conviction. A 2.5× volume bar at a put wall bounce = high probability support holding.

Signal Scoring — how the panel works

The signal panel scores a three-bucket model. Each bucket is independently normalised to [−1, +1], then blended into a final score. Confidence = abs(final score) × 100%. Direction is NEUTRAL if confidence < 25%.

Final = stateScore × TBD + momentumScore × TBD + flowScore × TBD

Multi-timeframe overlay

After computing the base score, the panel reads the Ice Cold Read (ICR) panel's 15m / 30m / 1h per-timeframe bias and adjusts confidence:

On-screen terms — glossary

Every label, badge, abbreviation and status word that appears on the live pages, grouped by page. If you spot a term on screen and want the one-line meaning, it's here.

Glossary · Flow Dashboard

Glossary · Flow Map

Glossary · GEX Heatmap

Glossary · Chart + Signals

Glossary · Magnets & Signals

Signal lifecycle states (ACTIVE, STRENGTHENING, WEAKENING, RESOLVED HIT/EXPIRED/COLLAPSED) are covered in full in the Magnets page section above.

Glossary · Accuracy

How to read the Magnets page in a trade

The page has three tabs. Here's the read sequence every time you open it:

Trading Example — Bearish Setup

Based on the May 15, 2025 data — SPY near $738 with dominant put positioning building over 5 polls.

Trading Example — Bullish Setup (partial conviction)

Trading Example — Conflicted Setup (stay flat)

Entry Rules — the checklist

Apply this in order before every trade. Stop as soon as one required condition fails.

Required (all must be true)

• ✓Combined bias is NOT MIXED. When passive and institutional flow disagree, the system itself is undecided — wait for them to align before risking capital.
• ✓Data Quality ≥ 80% (the box on the dashboard signal strip). Below 80%, the institutional bias is unreliable and you're effectively trading on one signal.
• ✓Chart signal confidence ≥ 50%. Below 50% or NEUTRAL = structural ambiguity, no trade.
• ✓Session row confirms the thesis. If session denies, the day's structure is against you — stop here.
• ✓At least 2 of 3 middle timeframes (1h / 30m / 15m) confirm. Momentum must be building in your direction.
• ✓A clear TARGET or SESSION TARGET exists within $5 of spot. No target = no trade.
• ✓5m Entry Gate confirms. Wait for it even if everything else is green. The gate controls timing.
• ✓You have a defined stop. VWAP breach, EMA reclaim, or FLOOR/CEILING break — one of these must be your stop.

Size by timeframe agreement

Size up further (add to full size)

• +Chart signal confidence ≥ 75%
• +GEX outlier marker just fired at or near the target strike
• +Volume Ratio >2× on the entry bar in the direction of the trade
• +MTF overlay confirms (all 3 higher TFs agree → ×1.15 confidence multiplier)

Stay flat (any of these = no trade regardless of signals)

• ✗Combined bias is MIXED. Passive and institutional disagree — wait for alignment.
• ✗Data Quality < 80% (red box). NBBO data is broken or thin; the institutional read can't be trusted.
• ✗Within 30 minutes of a major release (FOMC, CPI, NFP, earnings)
• ✗IV spiked >2% in one poll (unknown catalyst — options are pricing in something you don't know)
• ✗Flow bias flipped direction in the last poll (unstable — wait for a new streak to form)
• ✗Session row denies the thesis
• ✗No clear FLOOR or CEILING to define your stop