OA Dashboard

OPEN · PREV — today's open and yesterday's close. The most-watched levels of the day; bounces and fights happen here.

SD1 → SD3 (orange to red) — "expected move" bands from option prices. Price tags SD1 all the time; SD2 is stretched; SD3 is an extreme — the engine bets on a snap-back (a fade) at bands like these. The darker orange pair is the same idea from historical volatility (HV).

Dashed versions — yesterday's bands. Still respected early in the session.

⚡hot — a small marker where price escaped the 2σ range the options market was pricing 30 minutes earlier (from the same IV as the bands): the move genuinely outran what options implied. Rare, about 5% of the time when pricing is fair.

ENTRY / STOP lines — while the engine holds a position, a thin dotted line marks the average entry (green long, red short, ×N when pyramided) and a bright red dashed STOP line marks the live stop. The stop moves as the ratchet does (to breakeven at +5, then step-locks past +12), so the line is the exact price a print has to reach to close the trade. Both disappear when the position closes, and both go with the signal arrows when that layer is switched off.

☰ LAYERS — a checklist to hide or show anything drawn on the chart (signal arrows, the Δ DEX profile on the left edge, DMP lines, GEX magnet, walls, zero gamma, state strip and cone, trifecta ribbon, the open/prev, IV and HV bands, yesterday's bands, whale sweeps, ⚡hot). Your choice is remembered on this device; the button turns gold when something is hidden.

GEX M (gold) — the biggest gamma wall. Acts like a magnet: price gets pulled toward it and pins there, especially late in the day. Expect chop, not trend, when price sits on it.

DMP V (purple) / DMP C (green) — the dealers' hedging battle lines. They behave like support/resistance: price often bounces off them, and a clean break-then-retest is meaningful. When the purple line jumps in one direction and the option flow agrees, the engine follows it. With · TICKS on (default) there is one dot per recorded tick (every 3–7 s), not one per candle, so a line that flickers down and back inside a single candle shows up as a visible stutter — that intra-candle move is what the engine measures. Turn it off for the cleaner one-dot-per-candle view.

DMP = Dealer Mid-Price — the balance point of the dealers' hedging book. There are two: DMP V (built from vanna, the purple dots) and DMP C (built from charm, the green dots).

How it's derived — every tick, each SPX 0DTE strike comes with the dealers' signed vanna and charm exposure. For one greek, take the strike with the most positive exposure and the strike with the most negative exposure, anywhere in the chain (not just near price). The DMP is the midpoint of those two strikes: (high strike + low strike) ÷ 2. Do it once with vanna for DMP V, once with charm for DMP C.

Why it matters — those two strikes are where dealers' hedging pressure is heaviest in opposite directions. Vanna is how their hedge changes as implied vol moves; charm is how it changes as the clock runs down. The midpoint is where those forces net out, which is why price treats it like a level.

Why it jumps — the line is anchored to just two strikes, so when a big position makes a different strike the new extreme, the DMP teleports instead of drifting. That fresh jump is what dmp_follow trades (with flow agreeing and enough room to the line). Flicker between two neighbouring strikes is ignored on purpose.

Aligned — when DMP V and DMP C sit within about a quarter of the day's expected move of each other, the two hedging lines agree and the engine treats the structure as one battle line.

QD STATE tile / state strip — the header tile left of ENGINE STATE shows QuantData's current label in its color, and the thin band along the bottom of the price pane is the same label across the session: gold = Pin (rangebound), green = Grind (directional), red = Volatile. Live it changes only when the label changes; on replayed days it is drawn from the 10-minute history. It says what kind of tape this is, not which way it goes. On our own ticks the next hour's range runs about 14 pts in a Pin, 17 in a Grind and 24 in Volatile, and a label holds for the next 30 minutes roughly 7 to 8 times in 10 (Volatile is the stickiest). Tested as an entry filter (f130/f130b): it does not improve our gates, so nothing trades on it — it is context for stops and targets.

Shaded cone — from the viewed candle forward 60 minutes: where SPX ended up 15, 30 and 60 minutes after past minutes in the SAME state. Outer shade = the middle 80% of outcomes, inner = the middle 50%, in the state's color; the chart keeps an hour of room on the right while it is drawn. It is a range to size stops and targets against, not a direction, and it follows the scrub, so you can see what it showed at any past moment.

CALL WALL / PUT WALL / ZERO γ — QuantData's levels, drawn as dotted tracks across the session so you can see them move (the axis pill is the current value): the strikes carrying the largest call and put exposure (commonly resistance and support, but not tied to price) and the price where dealers' net gamma flips sign, which changes how their hedging behaves rather than being a target. The raw zero-gamma level hops strike to strike as the chain updates, so the line drawn is its 15-minute average. Refreshed every minute live; the tracks start the day the engine began logging them.

Gold @lvl — a whale bet AGAINST the current move, placed right at a key level. These usually work — lean WITH the arrow.

Grey — no read. Ignore. (The old pink ⚠fade tag was retired 09-21: regraded on 728 sweeps it was a coin flip, so a whale bet that merely agrees with the flow tells you nothing about the next hour.) 🛡 marks a far-out-of-the-money bet: probably portfolio insurance, not a directional opinion.

CALL vs PUT — today's real option money, added up through the day. Whichever line is climbing faster is in control. The colored bars underneath grade it into phases — tap the legend at the bottom for that guide (bull run / bear run / contention / divergence).

Green — the gamma structure points up, a $2B+ in-the-money strike backs it, AND every delta bar between price and the gamma wall agrees (no red delta sitting in the path): lean long while it's on. Red — the mirror: lean short. Nothing = no backed zone.

STATE — ACTIVE can take trades; READ-ONLY is deliberately standing aside (conflicting signals, extreme tail risk, or near the close); HALTED is outside market hours.

NEXT 30 MIN · HISTORICALLY — the panel above the engine panel. Each row is one plain fact about right now (QuantData's market state, above or below the open, which way dealers' delta leans, where the biggest gamma wall sits and whether it is positive or negative, whether puts or calls are being bought more today, whether the DMP is shifting). The market-state row is different from the others: it shows the 80% range of the next 30 minutes in that state (the numbers behind the shaded cone) and how often the label holds, and it is never counted in the lean. The bar and number beside it are the average move SPX made over the next 30 minutes from that same situation, across every recorded session. Red bars lean down, green lean up, grey means that fact has never predicted anything. The headline is the session-weighted average of the rows that held up out of sample, in points. "Setups like this one" means the current combination of facts, not the current price: the read is recomputed every tick and stays the same for as long as the facts do. Tap a row for the 15 and 60 minute versions, the share of times it went up, the odds of a 5 or 10 point move in each direction, the median, the share of days the setup leaned the way it says, and the session count. Read the probabilities the way they are written: "higher 30 min later" near 50% means direction is a coin flip and the lean comes from the size of the moves, so the numbers that matter are the share of days that leaned this way and how much more often the 10-point move landed on the lean's side. Rebuilt nightly.

DMP VANNA / CHARM — where the dealers' two hedging lines sit right now (the purple and green dots, as numbers). See "What DMP means" above for how they're computed.

TRIFECTA — the ribbon's current state as text: ▲ LONG ZONE / ▼ SHORT ZONE while the backed gamma zone is on, "off" when price is in no-man's-land, "—" when there's no delta data.

NET DRIFT CONV — the flow's conviction, roughly −1..+1. Since 09-08 the drift lines count single-leg electronic flow only (complex, tied, floor and cancelled trades excluded) at ATM and OTM strikes — earlier sessions were recorded on the unfiltered series. Positive = call money winning (bullish). The engine wants at least ±0.15 before it trusts a move.

LAST SIGNAL — the most recent engine entry. The arrows on the chart are these: green = long, red = short, labelled with the strategy that fired.

dmp_follow — the dealers' line JUMPED (a fresh shift) and the money agrees (drift ±0.15 the same way): ride it. Two more rules explain most "why didn't it fire" moments: the line must sit at least 3 pts ahead of price in the trade's direction (it never chases a line trailing behind a move — tested, that entry loses), and a winning position may add a second contract when the signal re-fires with it (shown as "2x @ avg" on the chart and blotter).

deviation_touch / sd3_fade — price tagged a stretched band: bet the snap-back. These fades ALWAYS outrank a conflicting dmp_follow read.

dmp_retest — price broke the dealers' line, came back, and held: enter on the confirm.

trifecta — the ribbon's zone (above) traded directly, exits when the zone turns off.

cross_asset — a correlated symbol (SPY/QQQ) tagged ITS band first: fade SPX with it.

Fade stand-down — when today's flow shows calls AND puts both being net SOLD (the premium-selling grind — most of the time, in fact), all four fade-style strategies stop taking new entries: historically every point they lost, they lost in that state. dmp_follow is unaffected.

Agreement % (on entry cards and arrow tooltips) — how many of the engine's inputs concurred at entry (gate strength, DMP alignment, trend, room, flow). It is NOT a win probability: measured across 268 trades, the high-agreement band has the LOWEST win rate — when everything agrees, the move is usually already late. Descriptive only; nothing trades on it.

Circuit breaker — three consecutive full stops benches the engine for the rest of the day. Tested against looser versions repeatedly; the strict version wins.

What it shows — the full history replayed under TODAY'S configuration (the nightly job writes it), topped up live every 30 seconds with today's actual trades and any open position. "Last update · Resync" at the top tells you how fresh it is.

Stat cards — net P&L in dollars ($50/pt per contract), profit factor (donut), win % (half-gauge: green wins · grey scratches · red losses), and average win vs average loss.

Eval account — the $150k account against the $4,500 trailing end-of-day drawdown rule: live balance, the minimum balance to maintain (ratchets up with each new best day, never down), room left, and the closest call — worst dip on end-of-day marks and intraday (a live dip through the floor fails the account even if the close recovers). The "since" date is pickable; every eval-scoped number recomputes from it.

Calendar — each traded day shows its dollars, trade count and win rate; the right column totals each week. Click a day to filter the trade table to that session (click again or ✕ to clear). A dashed orange outline marks the drawdown stretch behind the closest-call number.

OA score — six health components (targets on the ⓘ) drawn as a radar; the shape tells you WHICH component is dragging, which matters more than the 0-100 number. The toggle scores the whole history or just the eval window.

Daily net cumulative P&L — the eval's running dollars from $0, with the trailing drawdown floor dashed in red. Hover for any day's balance, result, and room to the floor. The line touching the dashed floor = eval failed.

Net by entry hour — which times of day the entries actually pay. Trade efficiency — each row is a trade: the grey bar shows how far it went in our favor at its best moment, the green/red bar what the exit actually kept. Short green next to long grey = the exit gave the move back.

Gate report card — per-strategy stats: trade count, win rate, profit factor (money won per $1 lost), expectancy (average per trade), and totals.

Price-action box (top-left of the chart) — drag it anywhere it isn't in the way; double-click to snap it back. Your spot is remembered.

Scrub bar — replay any moment of the session; every panel follows. 1m/5m/15m/1h — candle size. ⟲ FIT — reset zoom. · TICKS — DMP dots at every tick. ☰ LAYERS — show/hide anything drawn, including the engine's entry arrows and the Δ DEX profile: per-strike delta exposure drawn on the left edge: each bar sits at its strike's price, length = how much delta lives there, green = positive, red = negative. Big bars are the strikes dealers care about; a bright bar with a white tip carries $2.5B+ of delta at one strike — an institutional anchor. The profile follows the scrub bar through the day. The ☰ menu (top left) holds the tabs: Signals — every executed trade and its result across all sessions; Guide — this card; plus the theme picker (Midnight · Slate · Paper · Ocean · Forest · Ember · Terminal — applies live, remembered per device). Tap any sidebar panel title to collapse it. Times show in YOUR timezone (add ?tz=et to force New York).

Nothing here is financial advice — it's a lens on dealer positioning. The engine's live record is in the nightly report, wins and losses alike.