Options Trading Dashboard

AI approval queue

No trades awaiting approval.

Shared settings

Cross-engine controls consumed by both theta-engine (short premium) and momentum-engine (debit verticals).

Loading shared settings…

These values are deployment-managed and read-only — a tuning change ships as a reviewed preset/config PR, and the startup preset sync applies it on the next deploy.

Cross-strategy guard

When ON, momentum and theta hard-veto each other on the same underlying (one position per symbol across families).

Per-position BP utilization cap (0–1)

Max Reg T margin for any single new position, as a fraction of NLV. Applies to both theta and momentum engines.

Account-wide BP governor cap (0–1)

Max aggregate Reg T margin across the WHOLE account — both engines' filled positions plus every pending candidate — as a fraction of NLV. The per-position cap above bounds each single trade; this bounds the total deployed. Optimum: 0.65 (this small, defined-risk-heavy account — defined-risk BP equals fixed max loss, so it can run above the textbook tastytrade "keep ~50% of net liq in cash"). Set 0 to disable.

Max contracts per position

Upper bound on the contract count either engine may open for one new position. The size itself is derived from the per-position ceilings the entry gates already enforce (per-position BP cap, account-wide governor headroom, structural max loss vs tail-risk %, position-size % of NLV) — this only caps how far those may scale. 1 = every position is a single contract (the historical behavior): dollar P/L is then absolute, so account size changes only which trades pass, never how big they are. Note the projected-loss caps are flat dollars, not a % of NLV, and will pin sizing at 1 contract at any account size until raised.

Risk-free rate (Scenario Grid)

Black-Scholes risk-free rate used by the scenario-grid pricer. Synced automatically each trading day from the current US Treasury short-term rate — no longer operator-editable.

Naked positions

Disallow naked calls

When ON, theta-engine opens no position carrying an unhedged short call — standalone naked calls and short strangles. Defined-risk bear-call verticals are unaffected. Momentum trades only defined-risk debit verticals, so this never changes its behaviour.

Max underlying value for a naked put ($)

Theta-engine drops naked-put candidates whose underlying spot price (USD/share) exceeds this cap. Leave blank for no cap.

Macro delta (account βΔ)

Delta lean (fraction of NLV)

Recenters the NEUTRAL window both engines gate the account's SPY beta-weighted delta against. Negative = standing short-delta lean (the tastylive norm for short-premium books — crash insurance); 0 or blank = strict delta neutrality. Optimum: −0.03.

Updated: —

Alpha settings (momentum engine)

Debit-vertical / directional screener: short-DTE catalysts, low-IV-rank underlyings.

Loading alpha settings…

These values are deployment-managed and read-only — a tuning change ships as a reviewed preset/config PR, and the startup preset sync applies it on the next deploy.

DTE min

DTE max

Minimum reward-to-risk ratio

Deterministic pre-AI gate: a debit-vertical candidate whose reward-to-risk ratio (max gain ÷ net debit) is below this floor is rejected before the LLM advisory runs. Example: 1.5 requires the spread to be able to make at least $1.50 for every $1 risked. Leave blank for no floor.

Long-leg target delta

Short-leg target delta

Strike selection for the assembled vertical: each leg picks the strike whose |Δ| is closest to its target. Long must stay strictly above short (the engine rejects an inverted pair). Deeper long legs (higher delta) pay less embedded-leverage premium; blank uses the engine defaults (0.50 / 0.25).

Thesis envelope (scenario grid)

Records the forward target the directional thesis commits to at the close horizon, and the 2-D price×volatility grid (at the target DTE) used to measure drift against it on each open position. The grid also drives the optional profit/loss screening gate below.

Target DTE

Days-to-expiry at which the directional thesis is reassessed / the position is worked. This is the anchor for the recorded thesis target: at approval, target spot = entry ± 1σ for BULL/BEAR theses (entry for NEUTRAL / premium-decay) and the drift chip on the position card measures from this point. Blank uses the engine default (7). Only applied when no full grid JSON override is set.

Scenario-grid profit/loss gate

Screens new momentum candidates against the Black-Scholes profit grid (priced at the Target DTE above) — the same grid shown in the Live profit preview and on signal-tape approval cards. Leave a field blank to disable that check.

Max projected loss ($ per spread)

$

A floor, not the whole cap — the gate enforces the larger of this and the fraction-of-NLV cap below.

Max projected loss (fraction of NLV)

The gate rejects when the grid's worst cell loses more than max(dollar cap, this × NLV), so the cap grows with the account. Blank or 0 = dollar cap only. 0.05 × $5,000 = $250, the historical flat cap. Leaving the DOLLAR cap blank still turns the gate off entirely.

Position size cap (fraction of NLV)

One new vertical's buying power may reserve at most max($250, this × NLV) — momentum's counterpart of the Theta tab's position-size gate, using the same floor-plus-fraction shape as the loss cap above so the two tie exactly at matching percentages. Unlike the loss cap this is always armed: blank rides the 0.05 default, so clearing the dollar loss cap can never leave a position unbounded. For a debit vertical the BP draw is the whole max loss, so at 0.05 it changes nothing today; set it LOWER to cut contracts per symbol above the floor.

Take-profit target ($ per spread)

$

Reject the candidate when the grid's best-case cell can't reach this gain.

Max bid-ask spread ($ per spread)

$

Reject the candidate when its combined long+short leg bid-ask spread is wider than this — filters out illiquid, wide-market trades that fill poorly.

Max bid-ask spread (% of mid)

%

Companion to the $ ceiling above: reject when the combined spread exceeds this percentage of the sum of the leg mids. A fixed $ cap is 5% slippage on a $3.00 vertical but 30% on a $0.50 one — this catches the cheap, proportionally wide markets the $ cap waves through. Blank or 0 = disabled.

Min expected return on risk (%)

%

Reject the candidate when its directional expected value (the scenario grid repriced under the technical-conviction drift — the same EV the direction picker selects on) is below this percentage of worst-case risk (EV ÷ |max loss|). A quality floor on top of the built-in EV>0 gate. Defaults to 10%; clear the field to disable.

Min expected value after fees ($ per spread)

$

Reject the candidate when its directional expected value, net of estimated round-trip fees (~$2.60 for a two-leg spread), falls below this dollar floor. Catches thin trades that are gross-positive but won't clear commissions — the percentage gate above is risk-normalized and can't see this. Defaults to $5; clear the field to disable.

Min directional forecast probability (%)

%

Reject the candidate when its modeled directional forecast probability — the chance of reaching the profit target by the target DTE under the conviction drift (the "Directional Forecast Probability" shown on the card) — is below this percentage. Distinct from the static long-leg Δ in-the-money proxy shown as POP on the trade card. Catches the long-shot directional bet whose drift-inflated EV clears the floors above while its real chance of working out is low. Defaults to 40%; clear the field to disable.

Min no-drift expected value after fees ($ per spread)

$

Reject the candidate when its no-drift (symmetric) expected value — the honest, drift-free grid EV, net of round-trip fees — falls below this dollar floor. Unlike the two EV floors above (which use the directional drifted EV), this gates trades whose only edge is the assumed move continuing. A directional spread is structurally negative-EV under no drift (the variance risk premium — long premium loses on average), so this is a tail filter, not a near-zero bar: a typical spread's symmetric net EV already sits a few dollars negative. The default is −$5 (active out of the box; rejects the worse-than-typical, egregiously drift-dependent trades); tighten toward −3 to bite harder, loosen toward −8 for extremes only; 0 would gut emission. Clear the field to disable.

Max conviction drift (σ-units)

How far a fully-convinced technical vote (|voteSum| ≥ 4) may drift the modeled spot distribution toward the thesis; lesser convictions scale linearly. This cap scales the directional EV every drifted gate above evaluates — the calibration knob for how much the technical signal is trusted (realized by-conviction win rates are journaled per closed trade). 0 disables the drift (selection falls back to the symmetric zero-drift EV); blank uses the engine default (0.2).

Market regime preset

The regime in force: one setting for the spot sampling and IV offsets together. NORMAL = engine defaults.

Set automatically each trading morning from live macro conditions (moves at most one level/day). Not editable here — a different regime ships in the deployed preset.

Advanced: edit raw grid

Scenario grid

—

Spot axis mode

Spot moves (σ of expected move)

σ

IV shocks (added to entry IV)

Live profit preview (SPY)

—

Cells show dollar P&L at the Target DTE for a one-contract representative trade on SPY (long ATM/+5% call vertical for Alpha). "Use default" on every axis falls back to per-strategy engine defaults.

These checkboxes turn individual screener gates on or off. They do not affect portfolio risk vetoes.

Momentum screener gates

Chain IV rank ceiling gate

Two-tier IVR pre-gate: Tier-1 uses the DB universe IV rank as a fast-path skip; Tier-2 backstops with the live chain IV rank. Momentum favors low IV rank underlyings — symbols above the IV rank ceiling are dropped.

IV rank ceiling (0–100)

Symbols with IV rank above this value are excluded. Blank uses the engine default.

Equity volume minimum

Floor (shares)

When a gate is off, that constraint is ignored during candidate screening. Turning off a gate does not relax portfolio risk vetoes.

Market-volatility entry gate

While the market-vol index (VIX) is at/above the threshold, the screening cycle emits no new debit-vertical candidates — the mirror image of the Theta tab's IVR-floor relaxation.

Gate new entries while market vol is elevated

High-VIX tapes are hostile to short-dated directional debit plays: premium is expensive (worse reward/risk for the same width), momentum signals degrade in panic/rebound regimes, and stop-loss whipsaw rises with realized vol. Entries only — open positions, auto-close, and approvals of already-minted cards are untouched. Fails open if the index quote is unavailable.

Elevated market-vol threshold (VIX)

Blank uses the engine default (25 — shared boundary with the Theta tab's relaxation, so the two regimes flip together).

Release threshold (hysteresis)

Once the gate engages, entries stay halted until VIX falls to/below this level. Blank uses the derived default (threshold − 3, e.g. 25 → 22) — the median above-threshold spike lasts ~2 sessions and hovers at the line, and the immediate post-spike rebound is where momentum crashes cluster. Note: editing the threshold or release starts the regime fresh under the new band (a latched gate does not carry over) — this is also the manual-release lever for a wrongly-latched regime.

Earnings gate

Mirror of the Theta tab's earnings gate, for the opposite reason: a short-DTE long-premium vertical held through an earnings print pays for event IV that evaporates at the report (vol crush), so even a directionally-right thesis can lose.

Skip candidates with earnings before expiration

Drops a candidate when the underlying's expected earnings report (TastyTrade market metrics) lands on/before the expiration plus a 5-day estimate-slack buffer. Fails open when the report date is unknown.

Auto-Close (exit rules)

Overrides for the deterministic momentum exit thresholds. Blank uses the engine defaults shown.

Profit target (fraction of max profit)

Stop loss (fraction of debit)

Time stop (elapsed fraction)

Close once this fraction of the entry→expiration window has elapsed. 0 = off.

Hard close DTE floor

Force-close at/below this DTE regardless of P&L. 0 = off.

Trailing lock arm (fraction of max profit)

Peak profit at which the trailing profit-lock arms.

Trailing lock give-back (fraction of peak)

Close once an armed spread gives back this share of its peak profit — a winner never round-trips to a loser.

Updated: —

Theta Settings

Loading universe settings…

These values are deployment-managed and read-only — a tuning change ships as a reviewed preset/config PR, and the startup preset sync applies it on the next deploy.

IV rank floor

Per-symbol IV-rank gate consulted by the snapshot screener.

IV rank floor (%)

Underlyings whose IV rank is below this floor are excluded from screener output.

Elevated market-vol threshold (VIX)

When the market-vol index (VIX) is at/above this level the tape counts as elevated and the IV rank floor is relaxed — per-name IV ranks lag a broad-vol regime while absolute premium is rich.

IVR floor relaxation (points)

IV-rank points subtracted from the floor while elevated (e.g. floor 45 → 35). 0 disables the market-vol regime check.

Release threshold (hysteresis)

Once elevated, the relaxed floor persists until VIX falls to/below this level. Blank uses the derived default (threshold − 3, e.g. 25 → 22) — keeps the relaxed floor through the post-spike decay (elevated-but-falling vol, the best short-premium entry window) instead of flapping the floor every cycle at the line. Note: editing the threshold or release starts the regime fresh under the new band (a latched regime does not carry over) — this is also the manual-release lever.

Earnings gate

Skip candidates whose underlying has an expected earnings report on or before the candidate's expiration (short premium held over earnings is the classic win-rate destroyer). Fails open when the report date is unknown.

Stage 3: Option Chain Screener (The Hunt)

Filters applied to each option chain during the discovery-orchestrator screening cascade.

All seven option-chain screener gates (chain IV rank, DTE window, |delta| band, option volume, bid–ask spread, probability of profit, equity volume) are always applied — only their thresholds are tunable below. These filters are per-strike vs market snapshot; they do not affect Stage 4 portfolio risk vetoes (concentration, net delta vs regime, BP reserve, position size).

Leave a threshold blank to use server defaults (placeholders). Clearing a saved override sends null so config defaults apply.

Equity volume floor (shares)

Underlyings whose live session share volume is below this floor are dropped before the option-chain fetch. Blank uses the engine default (1,000,000 shares); an explicit 0 disables the floor.

DTE min

DTE max

Delta |x| min

Delta |x| max

Delta |x| min — elevated regime

While the market-vol IVR relaxation is active (VIX elevated + contango), the screener uses this band instead of the normal one: rich premium buys the same credit at lower delta. Both fields must be set for the shift; clearing either reverts to the normal band.

Delta |x| max — elevated regime

Upper edge of the elevated-regime band (must stay above its min). Default pair 0.10–0.22, below the preset normal band 0.16–0.30.

Volume min

Spread max ($)

Spread max (% of mid)

Companion to the $ cap, applied per structure: package spread ≤ this % of the sum of leg mids at packaging/approval, and a naked single leg's spread ≤ this % of its mid. Catches cheap structures where the $ cap is proportionally huge. Blank or 0 = disabled.

Open interest min (short leg)

Short-leg strikes with a known open interest below this are dropped (day volume is spoofable by one print; resting OI is not). Unknown OI passes (fail-open). Blank = disabled.

POP min — verticals & single legs (%)

Strikes/packages with estimated POP below this percentage are dropped before discovery / AI. Applies to single-leg shorts and put/call credit verticals.

POP min — iron condors (%)

Separate, normally lower floor for iron condors: an IC profits only when price finishes between two short strikes, so its probability of profit is structurally below a one-sided vertical's. Using the verticals floor here would reject every condor. Leave blank to use the engine default.

Discovery packaging (theta-engine)

When a screened short is paired with an opposite-type short at the same expiry, the engine generates strangles, iron condors (for each wing width), and a vertical on the screened leg only, then scores each package on normalized net credit, inverse buying-power impact, and POP, and selects the highest composite. Leave blank to use deployment defaults.

IC / vertical wing widths ($)

Comma-separated strike distances between each short and its long hedge (same grid for iron condors and for the screened put or call vertical). Example: 3,5,10 tries $3-, $5- and $10-wide structures when those strikes exist in the chain. Mind the interaction with the minimum-premium floor and credit/width ratio: a $3 wing at a 30% ratio yields ≈ $90 credit, so a premium floor above that silently kills every 3-wide candidate.

Merit weight: credit

Relative emphasis on higher net credit within the candidate set (min–max normalized per evaluation).

Merit weight: BP

Relative emphasis on lower incremental buying power (defined-risk spreads vs naked strangles).

Merit weight: POP

Relative emphasis on higher estimated probability of profit for the packaged legs.

Stage 4: Portfolio Risk Vetoes (Account Protection)

Hard safety stops on every candidate (discovery, rolls, closes). A blank field means the deployment default (shown as the placeholder) is in force.

Max positions per underlying

Gate 1: count of portfolio strategy packages per underlying (same underlying + expiration, e.g. iron condor = 1). Raise for multiple expiries on one symbol.

Delta tolerance (regime window)

Gate 2: portfolio net delta after the trade vs the current market regime (bull / neutral / bear). Wider value = more room. Separate from Stage 3 “Strike |delta| band”, which only filters individual strikes before risk runs.

Neutral window fraction

Gate 2 (NEUTRAL regime): window half-width = delta tolerance × this fraction, e.g. 0.30 × 0.65 = ±0.195 net delta. When the book has already drifted outside the window, candidates that don't push it further out still pass (saturated-book recovery).

Position size cap (fraction of NLV)

Gate 4: one new position's estimated open buying power may reserve at most this fraction of NLV. Size is the primary control on short-premium tail risk — this gate never disables: a value ≤ 0 or ≥ 1 (e.g. a percent-style typo like 10) falls back to the 0.10 default. With the small-account override ON a breach downgrades to a LARGE_POSITION_SIZE warning instead of a veto.

Gate 3: Portfolio BP utilization cap

The buying-power ceiling (fraction of NLV) is a single cross-engine control — set it once as Portfolio BP utilization cap on the tab. Theta's Gate 3 and momentum's risk veto both enforce that same value against each candidate's full-position buying power (defined-risk spread width or naked Reg-T estimate).

Small-account position size

Gate 4: when on, large single-leg BP vs NLV becomes a warning instead of a hard veto.

Max fee % of credit

Gate 5: hard veto (HIGH_FRICTION_COST) when estimated round-trip fees exceed this fraction of the credit.

Stage 5: Pre-AI premium floor

Deterministic gate before the LLM advisory runs.

Minimum net premium per trade ($)

$

A screened short-premium candidate whose net credit (mid × 100 × contracts, USD) minus estimated round-trip fees is below this floor is rejected before the LLM advisory runs — it stops the engine spending an AI call on near-worthless premium. Leave blank for no floor.

Credit % of Wing Width

% of width

Spread candidates whose credit divided by wing width is below this floor are rejected before the LLM advisory runs (e.g. 33 = must receive at least 33¢ per $1 of wing width). Naked positions are unaffected. Leave blank to disable.

IV − HV floor (vol points)

vol pts

Variance-risk-premium gate: a candidate is rejected before the LLM advisory when the short leg's implied volatility sits below the underlying's 20-day realized volatility by more than this floor allows (e.g. 2 = IV must be at least 2 vol points above HV; equality passes). IV rank alone can't tell rich premium from realized vol that has caught up. Fails open when the daily-candle history is cold, stale, or split-corrupted. Leave blank for the engine default (0 = rejected only when IV is below HV).

Thesis envelope & max-loss gate (scenario grid)

The 2-D price × volatility grid (evaluated at the target DTE) records the forward thesis target and measures drift on each open position. The max-loss gate hard-vetoes when the strategy's defined structural ceiling (vertical width − credit, max wing − credit for a condor, strike − credit for a cash-secured put) exceeds the configured % of NLV. Naked short calls / strangles veto unconditionally — no defined ceiling exists.

Max-loss gate enabled

When on, a candidate is hard-rejected if its defined structural max loss divided by NLV exceeds the ceiling below. Sourced from the legs, not from a stress cell of the grid.

Max-loss ceiling (% of NLV)

%

Percent of NLV. A candidate is vetoed when its defined structural max loss exceeds this share of account value. Default 7%.

Risk-free rate —

Synced automatically each trading day from the current US Treasury short-term rate; also shown on the tab.

Target DTE

Days-to-expiry at which the short-premium thesis is reassessed — the classic 45→21 management band lower bound. Anchor for the persisted thesis target grid: drift cells on the Risk Book measure each open position against the 2-D price × volatility matrix evaluated at this DTE. Blank uses the engine default (21). Only applied when no full grid JSON override is set.

Scenario-grid profit/loss gate

Screens new candidates against the Black-Scholes profit grid (priced at the Target DTE above) — the same grid shown in the Live profit preview and on signal-tape approval cards. Leave a field blank to disable that check.

Max projected loss ($ per contract)

$

A floor, not the whole cap — the gate enforces the larger of this and the fraction-of-NLV cap below.

Max projected loss (fraction of NLV)

The gate rejects when the grid's worst cell loses more than max(dollar cap, this × NLV), so the cap grows with the account. Blank or 0 = dollar cap only, which pins position sizing at one contract at any account size. 0.07 × $5,000 = $350, the historical flat cap.

Take-profit target ($ per contract)

$

Reject the candidate when the grid's best-case cell can't reach this gain.

Market regime preset

The regime in force: one setting for the spot sampling, IV offsets and max-loss ceiling together. NORMAL = engine defaults.

Set automatically each trading morning from live macro conditions (moves at most one level/day). Not editable here — a different regime ships in the deployed preset.

Advanced: edit raw grid

Scenario grid

—

Spot axis mode

Spot moves (% of spot price)

%

IV offsets (from entry IV)

Live profit preview (SPY)

—

Cells show dollar P&L at the Target DTE for a one-contract representative trade on SPY (~0.20-Δ short put for Theta). "Use default" on every axis falls back to per-strategy engine defaults.

Auto-Close (exit rules)

Overrides for the deterministic theta exit thresholds. Blank uses the engine defaults shown — keep these consistent with the entry policy above.

Profit target (fraction of credit)

Stop loss (× credit)

Time stop at Target DTE (in profit)

Close any position showing a profit once DTE ≤ Target DTE instead of riding the high-gamma final weeks.

Hard close DTE floor

Force-close at/below this DTE regardless of P&L. 0 = off.

Trailing lock arm (fraction of credit)

Peak captured credit at which the trailing profit-lock arms.

Trailing lock give-back (fraction of peak)

Close once an armed position gives back this share of its peak profit — a winner never round-trips to a loser.

Horizon bank floor (fraction of credit)

Least credit that must already be captured before the in-profit time stop banks at Target DTE. 0 = bank any profit. A value at or above the profit target is clamped to it. Defined-risk structures only. If you set this above 0, keep the min DTE floor above 0 too — otherwise a below-floor winner has no time-based exit left.

Updated: —

Trade performance

Export for Claude Export CSV

Closed round-trip trades with realized P/L (net of fees), their original thesis, and entry/exit conditions. Use the breakdowns to see what works — and when to exit.

Closed trades

Loading…

Backtest

End-of-day replay of the screening engines over archived option chains. Fetch history for your symbols first, then launch a run.

Historical data

Loading coverage…

Symbol	From	To	Sessions	Und. bars	Stats

VIX: —

Symbols (comma-separated)

Start

End

Max DTE

Strike window (± fraction of spot)

Fill gaps only — fetch just the missing sessions (including interior holes), skipping what's already downloaded. The symbols above are gap-filled; leave the Symbols box empty to backfill every stored symbol (new symbols still fetch in full). Widening the strike window or Max DTE also re-fetches stored sessions to add the new strikes/expirations. Uncheck to force a full re-download.

Purge archived data ▸

Symbols (blank = all)

Before date (blank = all)

New run

Engine

Universe

Start

End

Universe size

Min option volume

Starting NLV ($)

Slippage (%)

Risk-free rate (%)

This run only — settings overrides (cleared after each run)

Applied to the settings this run freezes. Your live Theta/Alpha settings are not changed, so an A/B never alters live trading while it runs. Leave blank to replay the saved settings. Only keys the selected engine exercises are listed — the fields below are rendered from the server's override catalog, so the bounds shown are the ones the run will be validated against.

Loading override catalog…

Runs use the engines' CURRENT saved settings (Theta/Alpha tabs), frozen at start — plus any per-run overrides above, which apply to the run only.

Runs

Loading runs…