Methodology
OpticAlpha's terminal doesn't just mirror raw market data. It also computes a set of derived signals: options flow strength, dealer gamma exposure, short-interest crowding, cross-signal conviction, analyst track records, Fed rate repricing, earnings-season pricing accuracy, and single-name volatility mispricing. Each one turns a lot of raw activity into a tier, grade, or line you can read in seconds.
This page documents, in plain language, what each signal measures, the kind of data it draws on, how to read the tier or grade it produces, and what it does not tell you. That last part is spelled out on purpose, not glossed over. The exact formulas, weights, and numeric thresholds behind each signal are implementation detail, and they aren't published here, for the same reason most quantitative research products don't publish their exact model coefficients. The mechanism behind a score stays internal. What the score means doesn't.
Eight signals, documented
Flow Score
What it measures
Flow Score is a same-day read on options order flow for a given ticker: whether the options activity trading through the market today leans bullish or bearish, and how one-sided that lean is. It's rebuilt fresh every trading day and is meant to be read as today's positioning story, not a multi-day trend on its own, though a rolling sparkline shows how a name's daily score has moved over the past month. The same underlying flow is also grouped by sector and theme into a rollup. That rollup shows which groups of related names are seeing the heaviest bullish or bearish options activity, rather than just one ticker at a time.
What it weighs
It weighs the split between call-side and put-side dollar premium traded on a name today, how directional the trades' own option Greeks make that flow, and how much of the day's volume arrived as urgent, immediate-execution orders rather than resting limit orders. That last figure is a rough proxy for how much conviction is behind the flow.
How to read it
The score renders as a labeled grade on a five-step scale, from Strong Bearish through Neutral to Strong Bullish, alongside a sweep-order percentage, the day's net premium split, and a sparkline of recent daily scores. The sector and theme rollup uses a simpler three-step read for each group: bullish tilt, mixed, or bearish tilt.
What this does not tell you
Flow Score can't tell you whether a trade is opening a new position or closing an old one, who placed it, or why. It says nothing about a trader's cost basis, time horizon, or whether the position is hedged elsewhere. Because it resets each session, a single very large trade can dominate the read for an otherwise quiet name. The score is a same-day positioning signal, not a price forecast: a strongly bullish read describes today's activity, not tomorrow's direction. The sector rollup can also be skewed by one especially active name inside an otherwise quiet group.
GEX (dealer gamma exposure)
What it measures
The GEX profile estimates how much hedging pressure market makers are collectively carrying across the options chain for a given underlying, and whether that pressure is more likely to dampen price moves or amplify them. Alongside the aggregate exposure number, it flags specific strikes ("walls") where hedging pressure concentrates most heavily, a delta-exposure (DEX) read of which direction dealers are already leaning, and the price level where the aggregate exposure is estimated to flip from one hedging regime to the other.
What it weighs
The profile is built from the full options chain for a ticker: every strike's open interest and its live option Greeks, combined with the underlying's current price, feed a model of dealer hedging exposure. That model runs across a range of hypothetical future prices around today's spot, not just today's actual level. A companion setup score for that same name also weighs how close price sits to a wall, whether recent order flow leans the same direction, and whether trading volume is running above or below normal.
How to read it
The output is a small set of directional signals: price movement expected to be dampened or amplified, and specific levels flagged as support, resistance, or a magnet price may gravitate toward. Each one carries a strength qualifier and the level itself. The companion setup score renders as a likelihood bucket, from unlikely through imminent, with a bullish or bearish bias and a plain-language list of what conditions are or aren't currently in place.
What this does not tell you
This is a model built on a standard assumption about how dealers typically hedge a position. It doesn't know any individual dealer's actual book, doesn't account for every hedge a market maker might already have on through other instruments, and doesn't hold with certainty on every name at every moment. It's recomputed on demand rather than continuously, so it can go stale intraday as the chain trades. On illiquid names, thin or missing options data can distort the walls and the flip level. It maps where hedging pressure currently sits. It does not say when, or whether, that pressure will actually be triggered.
Squeeze Radar
What it measures
OpticAlpha surfaces two related squeeze reads. One is a market-wide ranking of which stocks are carrying the heaviest short-interest positioning relative to their available float: a proxy for which names are most crowded on the short side right now. The other lives alongside the GEX profile for an individual ticker. It scores whether dealer positioning, price's distance to an options-driven wall, recent order flow, and trading volume line up toward a fast, forced move in that specific name.
What it weighs
The market-wide ranking weighs each stock's reported short interest as a share of its outstanding float, filtered to exclude names with too little daily trading volume to be a realistic setup, with recent news attached for context. The per-name setup score weighs the dealer gamma positioning regime described in the GEX profile above, the price's proximity to the nearest wall, whether options order flow this session aligns with a squeeze direction, and whether today's volume is elevated versus normal.
How to read it
The market-wide list is an ordered ranking, most-crowded first. There's no numeric score to compare across names, just relative position on the list. The per-name setup score is a likelihood bucket from unlikely through imminent, with a labeled bullish or bearish bias and, where the setup isn't complete, a plain description of what's still missing.
What this does not tell you
Neither reading predicts timing. A name can sit at the top of the short-interest ranking for months without a squeeze ever happening, and a fully-lit setup score describes favorable conditions, not a scheduled event. Short-interest data is reported on a lag, not live. The ranking doesn't say who is short, why, or at what cost basis. It also says nothing about the underlying company's fundamentals. Liquidity filtering can also exclude a heavily-shorted but thinly-traded name entirely, even though it may technically carry more short-interest risk than names that do appear.
Conviction Radar
What it measures
Conviction Radar checks whether several independent reads across the terminal are pointing the same direction on a given ticker at the same time: options flow, social attention, short-interest positioning, congressional trades, insider trades, analyst rating changes, and broad market attention. The idea is that agreement across sources that don't talk to each other is a more meaningful tell than any single source alone.
What it weighs
Each source casts at most one directional vote: bullish, bearish, or no read at all if that source has nothing clear to say about the name right now. A rating upgrade or a purchase-type filing votes bullish; a downgrade or a sale-type filing votes bearish; rising short-interest crowding tilts bearish; rising or falling attention momentum casts a vote in the corresponding direction; and so on, independently, for each of the seven sources. No single source is weighted more heavily than another.
How to read it
A ticker only appears once at least two sources agree on the same direction inside the same window. From there it's graded from Emerging through Extreme Conviction, based on how many of the available sources are aligned, alongside a plain-language line that spells out the exact count (for example, "4 of 6 aligned bullish") and which sources those were.
What this does not tell you
Agreement across sources is not proof of independent discovery. Several of them can move together simply because they're all reacting to the same piece of public news, not because each found something on its own. The read doesn't weigh how reliable or how large any one source's signal is, doesn't know when a source simply has no fresh data for a name, and is recomputed fresh each cycle, so a ticker can drop off the list without any single source having reversed its view. It flags coincidence across signals. It is not a price target or a trade recommendation.
Analyst Credibility
What it measures
This grades a research desk's own track record: whether the direction implied by its past rating changes (upgrades, downgrades, and price-target moves) actually played out in the stock's price afterward. The result is a broker-level tier, rather than treating every analyst call as equally credible by default.
What it weighs
Each rating action is read for the direction it implies: a clear bullish or bearish call. Purely neutral or reiteration-only actions are logged for coverage purposes but never scored, since they don't imply a direction to check. That implied direction is then checked against how the stock's price actually moved after a short cooling-off period following the call. The grade is built from a rolling window of a broker's own past resolved calls, not from any single rating in isolation.
How to read it
Brokers are graded into a small set of plain-language tiers, from "Limited Data" (when too few of a broker's calls have resolved yet to grade fairly) up through "Mixed" and into stronger track-record tiers. Each tier is shown with a short description, a coverage-frequency badge (how often that broker is actively rating names), and a simple visual hit/miss history. The exact hit rate and the underlying call counts behind the grade are deliberately not shown. Only the tier, the description, and the badges are.
What this does not tell you
This is a backward-looking track record, not a guarantee the broker's next call will land. It checks whether price direction matched, not whether the analyst's underlying research or reasoning was sound. A short cool-down period can also mis-time a call that plays out slowly, or one that initially moves the "right" way and then reverses. Neutral and price-target-only notes are never scored, so a broker's grade reflects only its directional calls, not its full body of coverage. And any broker with a small number of resolved calls can look artificially strong or weak until more of its record has had time to resolve.
Fed Delta
What it measures
Fed Delta tracks how much the market's own expectation for the Federal Reserve's next interest-rate decision has shifted over the trailing week, and separately, whether the Fed's own longer-range rate projections currently agree or disagree with where the market is pricing near-term policy.
What it weighs
It weighs the market-implied probability assigned to each possible rate outcome at the next scheduled Fed meeting, and how that probability-weighted expectation has moved over the past week, alongside the median longer-range rate projection Fed officials have submitted, compared against the market's current policy pricing for the same year.
How to read it
The weekly shift renders as a tier from minimal through sharp, with a hawkish or dovish direction line that describes which way the market has repriced. A separate aligned/diverging read compares that market pricing against the Fed's own longer-range median projection. Both render as a single plain-language headline (for example, a note that pricing has moved only modestly, or that market pricing and Fed projections are currently diverging) rather than a raw basis-point number.
What this does not tell you
This reflects what traders are currently pricing, not what the Federal Reserve will actually decide. Market-implied odds move around headlines and can reverse quickly. The Fed's own longer-range projections are a snapshot of individual officials' views as of their last meeting, not a forward commitment, and they're only updated periodically. Because the comparison runs on a trailing weekly window, a genuine shift that happens right at the edge of that window can show up gradually rather than immediately.
Earnings Ledger
What it measures
Across the market as a whole this earnings season, the Earnings Ledger tracks whether options have generally priced in more or less of a move around companies' earnings reports than those companies actually delivered. That means whether options are running rich or cheap heading into earnings, market-wide.
What it weighs
For each company's earnings report it tracks, the ledger compares the size of the price move the options market was implying beforehand against the move that company actually realized once results were released. It then pools that comparison across every reporting company it has tracked recently. A single ticker only reports earnings about once a quarter, too infrequent on its own to show a season-wide trend, so the read is deliberately market-wide rather than per-ticker.
How to read it
Once enough individual reports have resolved to be meaningful, the aggregate renders as a season-level tag (rich, fair, or cheap) that describes the overall relationship between implied and realized moves that season, alongside a trend line of recent individual events. Names still awaiting their own upcoming earnings carry the same rich, cheap, or fair tag, applied to that specific ticker's own reporting history and shown inline in the earnings screener.
What this does not tell you
This is a backward-looking, market-wide seasonal statistic, not a prediction for any specific upcoming report. Any individual company's own earnings-move history can differ a lot from the market-wide trend. The aggregate can also be skewed by a handful of unusually large surprises in one direction during a given season. It also says nothing about which way a move will go, only whether the size of the move options priced in matched what actually happened.
IV Opportunity
What it measures
For an individual ticker, the IV Opportunity tag flags whether the options market's current implied volatility looks stretched relative to that same ticker's own recent realized (actual, historical) volatility. It surfaces names where option premiums may currently be running rich or cheap against their own trading history.
What it weighs
It weighs where a ticker's current implied volatility sits within its own recent historical range (as a percentile) against where its own realized volatility over a recent trailing window sits within its own historical range. It's a purely self-referential comparison, one ticker's volatility against its own history, never a comparison against another ticker's volatility level.
How to read it
A simple two-state tag (rich or cheap) appears inline in the options screeners only when the gap between those two percentiles is wide enough to be notable. Most tickers carry no tag at all most of the time, by design. The absence of a tag is the normal case, not a data gap.
What this does not tell you
The tag says nothing about which direction the stock itself will move. It only means the current implied volatility looks out of step with its own recent realized volatility. Realized volatility is backward-looking and may not repeat, especially around a known catalyst like an earnings date, which can make an otherwise ordinary volatility premium look "rich" for a perfectly explainable reason. Because the comparison is purely intra-ticker, a name tagged "cheap" isn't necessarily cheap in absolute or cross-market terms. It's only cheap relative to its own recent history.
What this page is, and isn't
Every signal above is built from data OpticAlpha already collects for its live terminal, recombined and graded by rules that live on our backend and aren't published in full here. That's the same way most quantitative research shops keep their exact model weights internal, while still explaining, in plain language, what a score is built from and where its blind spots are. If you're citing one of these signals in research or reporting, cite the tier or the plain-language read it produces, and treat the specific weighting as OpticAlpha's own internal methodology rather than a disclosed formula.
None of this is financial, investment or trading advice, and none of it should be read as a guarantee of future performance. For more on who builds OpticAlpha, see the About page; to see any of these signals live, visit the terminal or browse the feature pages.
See these signals live
14-day free trial. No credit card required.