A synthesis layer over your terminal. Raw greeks, IV rank, charts and execution stay where they are. What lives here is the math TOS shows you raw but doesn't aggregate.
You're long the stock and write calls above market. Tell the engine where you target · it tunes the cards to your style.
Conviction floor hides weak signals. Slide right to demand stronger consensus.
Paid options through earnings get IV-crushed. How wide should the avoidance window be?
Drill each tier independently. Counts update as you slide.
What it measures: how far apart the 8 factor families are pointing for one name, as (max − min spread) × (balance of pos vs neg), scored 0–100. Why you'd change it: raise to surface only egregious disagreements where a strong family-level signal flatly contradicts another (these tend to be the highest-conviction "look closer" trades). Lower to catch mild tensions. Today's snapshot: AMD at 94/100 (cross_asset +2.0 vs fundamental −0.8) · the stock is leading its sector hard while fundamentals are deteriorating.
What it measures: SEC Form 4 net open-market selling minus buying, trailing 90 days, parsed and netted at the dollar level. Includes only directors, officers, and 10% owners; excludes grants and option exercises. Cluster bonus when ≥3 separate insiders sell. Why you'd change it: $5M is the floor where insider-flow signals tend to be statistically meaningful (Cohen-Malloy-Pomorski found 6%+ annual edge above this band). $25M filters to only flagrant exits. Today's snapshot: NVDA at −$266.4M (10 sells, 0 buys) · one of the biggest insider exoduses in the book.
Note: this one is buildable in thinkScript if you set up a `RelativePerformance()` column. We surface it by default so you don't have to. Raise to filter chronic laggards. Today: NVDA −19% vs XLK over 60d.
What it measures: a composite score for "best premium-selling target right now," computed as composite × (1 − conflict_penalty) + 0.3 × options_family_signal. The options-family signal blends IV-rank, term-slope, skew, and put/call ratio. Why you'd change it: low (0.10) means show every name where the long thesis is intact and options aren't actively screaming "wait." High (0.50+) keeps only names with strong directional conviction and a vol-rich profile suitable for short-premium structures. Today's snapshot: top edge candidate is DELL at 0.45 (cross_asset +2.0, options signal +0.4, low conflict).
SYNTHESIS Composite × (1 − conflict) + 0.3 × options_family. Blends fundamentals + technicals + macro + flow with the conflict penalty and a vol-regime overlay. The lead measure for where to write next · formula breakdown per name on demand.
EARNINGS WATCH Names with earnings inside the avoidance window. TOS has the calendar; the value here is the interactive EV underwriter → · drag strike, expiry, IV-crush, σ, contracts; watch the net EV flip green or red live.
AGGREGATED · PARSED · RATIOED TOS shows raw filings · we net 90 days of open-market sales only (P/S codes parsed, grants and exercises excluded), ratio against market cap and insider hold, flag clustering when ≥3 insiders sell. Cohen-Malloy-Pomorski edge band kicks in above $5M.
UNIQUE SYNTHESIS When 2+ of the 8 factor families flatly oppose each other (high spread × balance), the engine flags it. The math composing this signal is not feasible to reproduce in thinkScript · full 8-family decomposition and score formula visible per name.
RELATIVE STRENGTH Stock 60d − sector ETF 60d. You could build this in thinkScript; we surface it by default. For covered-call writers, sector laggards justify tighter strikes · less upside risk to cap.
EVENT-DAY TRANSMISSION When a name in the book reports, which OTHER holdings move, and by how much. Beta is measured on that name's own past earnings days, not on generic correlation, and baskets deliberately cross sector lines because that is how the desk trades them. Not reproducible in thinkScript: it needs a cross-name event study over an earnings history.
FORWARD SUPPLY Where stock could arrive, and the calendar that makes it likelier. Form 4 is a receipt filed after the sale; this reads concentration and known dates instead, both visible in advance. It does not predict that anyone sells · it ranks where a sale would land hardest.