How RadarPulse scores unusual options activity

The 0-100 unusualness score

Every options trade RadarPulse processes is assigned a single 0-100 unusualness score. The score is a blend of four inputs, each chosen because it captures a different dimension of what makes a print stand out from the thousands of routine, tiny, market-making trades that fill any normal session. The aim is triage: instead of eyeballing a firehose, you get a number that lets the strongest candidates float to the top, so you spend your research time on the 10 prints that matter rather than the 10,000 that don't.

The score is purely descriptive. It measures how statistically unusual a print is within the current data set. It does not predict future price direction, and it does not account for fundamental factors, earnings calendar, or macro regime. The score's job is ranking. Everything after that is your analysis.

The four inputs in detail

Each of the four scoring inputs is asking a different question about the trade. Together, they build a picture of whether this print belongs in the routine pile or the pay-attention pile.

Volume relative to open interest (Vol/OI)

Volume-to-open-interest is the foundational question: is this contract seeing activity it normally wouldn't? Open interest represents outstanding contracts that have not been closed or exercised. Volume on any given day represents fresh trading in the contract. When volume exceeds open interest substantially, it means more contracts are changing hands today than existed in the positions already outstanding, which points to new positioning rather than traders closing old trades.

A Vol/OI ratio of 5x (500 contracts traded vs 100 open interest) is above average but not extreme. At 10x, you're seeing genuinely unusual activity in that contract. At 20x or higher, the contract is seeing a day that has almost nothing to do with its normal trading pattern. The score component from Vol/OI is percentile-ranked, so a ratio of 10x in a contract where this is normal scores lower than a 10x ratio in a contract that typically sees a 1x ratio.

Vol/OI is also resistant to the large-cap bias that affects pure volume analysis. AAPL and SPY options see millions of contracts daily; raw volume on those names is always large. Vol/OI normalizes for base activity, which is why it's more informative than volume alone for detecting concentrated, out-of-character positioning.

Premium size

Premium size is the total dollars spent on the trade: the price per contract multiplied by the number of contracts multiplied by 100 shares per contract. It functions as a proxy for conviction because size is one thing a retail trader genuinely cannot fake. A $10,000 options trade is accessible to almost anyone with a brokerage account. A $2 million sweep requires meaningful capital commitment, which is why large premium is a meaningful signal filter even though it's imperfect.

The thresholds that practitioners commonly use: $25,000-$50,000 is actively traded; $50,000-$250,000 is active-trader to institutional range; $250,000-$1 million is clearly institutional-scale; above $1 million is a significant directional commitment or a major hedge. These are rough categories, not precise rules. A $300,000 premium on a single far-OTM LEAPS contract tells a different story than $300,000 spread across 50 different small orders. But premium is a fast filter for seriousness of intent, and it earns a meaningful weight in the composite score.

Importantly, premium alone does not make a high score. A $5 million trade on an ATM, short-dated SPY put by an institution hedging a large equity portfolio is huge in dollar terms but unremarkable in scoring terms, because it's exactly the kind of hedging activity that appears constantly in large-cap indices. The percentile ranking system adjusts for this.

Days to expiry (DTE)

Days to expiry captures the urgency dimension of positioning. Options trading is always a time-bounded bet, and the distance to expiry signals what kind of event the trader is positioning for.

Contracts with 0-7 days to expiry are betting on something happening almost immediately. At 4 or 5 days to expiry, theta decay is accelerating sharply, which means the buyer is paying a premium that evaporates fast: they need the underlying to move quickly or the option expires worthless. This kind of urgency is a meaningful signal. It suggests the buyer has a specific near-term event in mind, not a slow macro thesis.

Contracts with 8-21 days to expiry are the catalyst-bet range. Earnings typically fall within this window from when the flow appears, as does an FDA hearing date, an FOMC meeting, or a government procurement announcement. Flow in this window on names with known upcoming catalysts often reflects pre-positioning rather than random activity, and the score weights short DTE more heavily to surface this urgency.

Contracts with 60+ days to expiry carry more ambiguity. They could represent a directional bet, a defined-risk spread, a portfolio hedge, or a LEAPS replacement for a stock position. The score does not ignore them, but the DTE component scores lower for longer-dated contracts because the time pressure that makes short-dated flow distinctive is absent.

Out-of-the-money (OTM) distance

OTM distance measures how far the strike sits from the current stock price, expressed as a percentage. An at-the-money option (strike equal to or very near the current price) trades heavily as a matter of course: market makers delta-hedge with ATM strikes constantly, and they're the most liquid contracts in any chain. A deep OTM option with heavy, well-funded activity is more unusual because it requires the underlying to move significantly before the option has intrinsic value.

A call that is 15% OTM on a $100 stock needs the stock to trade above $115 before expiry to be in the money at expiration, ignoring time value. Seeing $500,000 in premium flow into that strike is a meaningfully different signal than seeing $500,000 flow into a $101 call. The 15% OTM trade implies either a significant catalyst bet, high conviction in a specific price level, or a tail-risk hedge from an entity that already holds a position.

The score weights OTM distance to surface these out-of-character prints. Deeply OTM contracts with large premium and high Vol/OI represent a convergence of signals that the composite score captures efficiently.

Why percentile ranking beats absolute thresholds

The most important design decision in the scoring model is the ranking method. RadarPulse ranks each input as a percentile within the current data set rather than applying fixed absolute thresholds. This sounds like a technical detail, but it makes a material difference in signal quality.

Consider the problem with fixed thresholds. A $200,000 premium is a large trade for a small-cap biotech with limited daily volume. The same $200,000 is routine in SPY options, which routinely process billions in daily premium. A fixed "$200,000 = high premium" rule would flag the SPY trade just as prominently as the biotech trade, even though the SPY trade is completely unremarkable in context. Similarly, a Vol/OI ratio of 10x is genuinely unusual for an actively traded large-cap like AAPL. For a thinly traded, speculative name that might have an open interest of 5 contracts, 10x can happen from a single small order.

Percentile ranking solves both of these problems simultaneously. Each input is measured against the rest of the prints in the current data set, so a $200,000 premium in a biotech context scores high on the premium dimension (it's in the top few percent of that data set), while the same dollar amount in SPY context scores lower (it's in the middle of the distribution). The output is a score that reflects relative unusualness within the session's actual data, not against a global rule that doesn't account for market conditions or ticker characteristics.

This approach also handles regime changes naturally. In a high-volatility session where every name is seeing elevated options activity, the scoring threshold shifts up with the data. In a quiet tape, it shifts down. The flags and rankings always reflect the most unusual activity within the actual session, not against a static baseline from a different market environment.

How the composite is assembled

The four inputs are each percentile-ranked within the session data, then combined into a composite score. The combination is weighted: not all four inputs contribute equally. Vol/OI and premium carry the most weight because they are the dimensions most directly associated with out-of-character positioning by an informed or committed actor. DTE and OTM distance are secondary inputs that add specificity about the urgency and target of the trade.

The weighting reflects what experienced tape readers actually emphasize when they evaluate flow manually. A print with extreme Vol/OI and large premium but a long DTE and near-the-money strike is notable. A print that is elevated on all four dimensions simultaneously is exceptional. The composite captures this layering: each additional dimension of unusualness adds to the score, and the combination of all four is what produces the highest scores.

No single input creates a high score on its own. This is an intentional design constraint. A trade that is simply large (high premium, all other inputs normal) scores higher than average but does not reach EXTREME or ELEVATED territory. The score is designed to surface prints that are unusual for multiple reasons at once, because those are the prints worth spending research time on.

EXTREME, ELEVATED, and NOTABLE flags

To make the score scannable at a glance, RadarPulse maps it onto three flags and collects the strongest activity into a daily Top 25 leaderboard. The flags are score bands:

Because the labels track the score, and the score is percentile-ranked within the session data, the labels describe relative standing within the day's activity, not absolute claim about what a trade will do. The bands move with the score, which itself moves with the data. An EXTREME flag on a quiet day represents the most unusual activity in a quiet session; an EXTREME flag on a busy day represents the most unusual activity in an active session. Both are worth attention; neither is a guarantee.

The Top 25 leaderboard surfaces the highest-scoring prints from the current session and rebuilds continuously as new data arrives. It's designed to give you the most unusual activity across the entire tape at a single glance, so you don't have to watch every ticker to catch the day's notable prints.

How bullish or bearish bias is inferred

Direction is the part most options flow tools get wrong. The naive approach, calling every call bullish and every put bearish, is fast and easily understood but frequently misleading. RadarPulse does not use this rule as the primary inference. Instead, bias is inferred from the context of the trade across three layers.

Contract type is a starting point only. Calls have bullish optionality and puts have bearish optionality, but whether the buyer or the seller initiated the trade changes the inference substantially. An institution that owns a large stock position might buy puts to hedge, which is structurally bearish on the options side but doesn't signal bearish conviction on the stock.

Aggressor side is the most informative field. When an order lifts the ask, the buyer is accepting the market maker's offered price and demanding immediate execution. This is an aggressive, urgency-driven action and reads as directional commitment. When an order hits the bid, the seller is accepting the buyer's price, which often reflects a closing trade, premium collection, or spread mechanics. A call lifted at the ask is substantially more bullish in signal than a call sold at the bid.

Sentiment columns, where the underlying data provides them, offer additional directional context. Some data feeds include explicit side annotation (buyer, seller, neutral) that supplements the aggressor side inference. When this context is available, RadarPulse incorporates it.

The combined result is bias as a reasoned interpretation of the print, not a binary label. Even with this context, the same print can be one leg of a multi-leg spread, in which case the directional read is incomplete by design. RadarPulse surfaces the directional inference available from a single print; you cannot know the full strategy from a single leg alone. This is a fundamental constraint of tape-reading any options print, not a limitation of the scoring model specifically.

Common misreadings of the score

After using the score for a few sessions, most users encounter the same interpretive mistakes. Recognizing them early prevents acting on the wrong signal.

Treating a high score as a buy or sell signal. The score measures unusualness, not expected return. A print that scores 90 can still be a hedge, a spread leg, or a roll. The score tells you the print is worth investigating; it does not tell you to copy the trade. EXTREME prints that turn out to be hedges are more common than EXTREME prints that turn out to be pure directional bets. The research step, checking contract context, aggressor side, open interest changes, and the stock's price action, is what converts a high score into a usable thesis.

Comparing scores across sessions. Scores are percentile-ranked within the current data set, which means a score of 82 on a busy, high-volatility session is not directly comparable to an 82 on a quiet, low-volatility session. The flags help here: an EXTREME flag is the top band regardless of session intensity. For cross-session comparison, use the flags and the print's absolute premium and Vol/OI ratio as the benchmark, not the composite score alone.

Over-indexing on single prints. One high-scoring print on a ticker is an observation. Two or three high-scoring prints in the same ticker, same direction, over the same session or across consecutive sessions is a pattern. The repeat-flow detection in RadarPulse is designed to surface this distinction. Single prints are worth logging; patterns are worth acting on.

Ignoring the contract details. A score of 88 on a $50 strike call on a $48 stock (near-ATM) reads differently from an 88 on a $60 call on a $48 stock (25% OTM). Both scored highly, but the first is near-term directional positioning while the second is a larger move bet or a tail hedge. The score surfaces them equally; the contract details differentiate them.

How scores evolve through the session

The percentile ranking is applied to the data set as it exists when each print arrives. Early in the session, when the data set is small, a print's score reflects unusualness within that morning's limited data. As the session progresses and more prints accumulate, the distribution widens and the score reflects unusualness within a larger, more representative set.

This has a practical consequence: the Top 25 leaderboard at 10:00 AM is a different list from the Top 25 at 3:00 PM, not just because new high-scoring prints have arrived, but because the ranking of earlier prints may have shifted as the full session data expanded. A print that was in the Top 25 at 10:30 AM may rank lower by 2:00 PM if subsequent prints were more unusual by comparison.

For most users this is academic; the leaderboard shows the current best candidates regardless of when they printed. For traders who track specific prints across the day, the score evolution is worth understanding: a score that drops as the session progresses doesn't mean the print is less interesting, it means the session was more active than expected.

Confluence: when multiple signals reinforce each other

The unusualness score assesses individual prints. A separate layer of analysis, visible in the confluence panel, tracks when multiple independent signals point in the same direction simultaneously.

Confluence is most meaningful when independent data sources converge. An EXTREME call sweep on NVDA is a single data point. If that sweep appears alongside elevated call buying in AMD and SMH (the semiconductor ETF), the sector-wide pattern suggests a broader thesis rather than a ticker-specific bet. If it appears alongside a congressional disclosure of semiconductor purchases in the prior disclosure window, the cross-domain overlap is worth noting. Each individual signal has its own noise floor; when they reinforce each other, the probability that any one of them is a false positive decreases.

RadarPulse's confluence detection runs automatically on the scored flow, grouping related signals by ticker, sector, and timing. A cluster of ELEVATED prints across tickers in the same sector within a two-hour window flags as a confluence event, separate from each print's individual score. The confluence panel in the Markets terminal surfaces these clusters so you can see them without manually correlating across the flow feed.

Data freshness: delayed vs real-time

The scoring model is identical across tiers; what changes is timing. On Basic and Pro, the scored options flow is 15-minute delayed: real market data, scored and ranked, shown 15 minutes after it hits the exchange. This suits studying activity, building watchlists, and confirming patterns. On Elite, the scored tape is real-time, for traders who need to react to prints intraday before the market has had 15 minutes to respond.

The delay applies to options flow only. Quotes, the S&P 500 heat map, and index charts are live across all tiers. For users on Basic or Pro, this means you have real-time market price context and 15-minute-delayed options flow context simultaneously. The combination is workable for most research and watchlist purposes; the 15-minute gap matters most for intraday execution strategies, not for identifying and researching patterns.

Important: this is research, not a signal

The unusualness score is descriptive analytics for research, not a signal to buy or sell. A high score tells you a print is statistically unusual within the data set. It does not predict the direction of the underlying. It does not account for fundamental factors, earnings surprises, or macro regimes. It does not know whether the print is a directional bet or a hedge leg in a larger structure.

Unusual options activity reflects the full range of institutional options strategies: outright directional bets, portfolio hedges, covered call programs, defined-risk spreads, LEAPS replacements for stock positions, and volatility trades. The scoring model does not distinguish between these structures; it measures unusualness across the four quantifiable dimensions. What comes next, the interpretation, the investigation, the comparison against open interest changes and chart context, is the analyst's job.

RadarPulse provides this data for educational and informational purposes only. Nothing on this platform constitutes financial advice. Trading and investing, particularly in options, involves substantial risk of loss, and most retail options traders lose money. Paper trading is available free of charge and is the appropriate starting point before committing real capital to any strategy based on flow signals.

Frequently asked questions

What does the RadarPulse unusualness score measure?

It measures how unusual an options trade is relative to the rest of the data set, on a 0-100 scale. The score blends four inputs: volume relative to open interest, the premium spent, days to expiry, and how far out-of-the-money the strike is, ranking each as a percentile within the current data set. A high score means a print stands out across these dimensions; it is a descriptive measure of unusualness, not a recommendation to trade.

How does RadarPulse decide if flow is bullish or bearish?

It infers direction from contract type, the aggressor side (whether the order hit the ask or the bid), and sentiment columns where available, not a naive assumption that every call is bullish and every put is bearish. A call bought at the ask reads differently from a call sold at the bid, and puts can be protective hedges rather than bearish bets, so bias is an interpretation of the context, not a fixed rule applied to the contract type alone.

Is a high unusualness score a signal to buy or sell?

No. The score is descriptive analytics for research, not a signal to buy or sell. It tells you a print is statistically unusual within the data set; it does not tell you the direction the underlying will move or what you should do. Unusual activity can reflect hedging, spreads, rolls, or other strategies, and should always be read in context and combined with your own research. Options trading involves substantial risk of loss.

Why does RadarPulse use percentile ranking instead of fixed thresholds?

Fixed thresholds fail across different market environments and ticker types. A $200,000 premium is a large trade for a small-cap biotech and routine for SPY options. A Vol/OI ratio of 10x is exceptional in a high-activity name and unremarkable in a thinly traded contract. Percentile ranking solves both by measuring each factor against the rest of the current data set, so the score reflects relative unusualness rather than an absolute dollar or ratio cutoff that doesn't account for the ticker or market context.

What does a score of 85 mean vs a score of 60?

A score of 85 is in the EXTREME band: unusual across all or most of the four scoring dimensions simultaneously. A typical session produces 3-6 prints at this level. A score of 60 is in the NOTABLE band: above average on some dimensions but not all. It is worth logging but carries less urgency. The difference is breadth of unusualness: an 85 stands out in multiple ways at once; a 60 stands out in one or two dimensions while others are closer to average.

Can a trade score high because of just one very large input?

A single very high input boosts the composite score but does not guarantee a high composite on its own. The score is a weighted blend of all four inputs, so a trade with extreme premium but near-dated, near-the-money execution and average Vol/OI will score higher than average but will not reach EXTREME or ELEVATED territory. The design is intentional: a print that is unusual for multiple independent reasons simultaneously is more informative than one that is simply large in one dimension.

Does the score change after a print is first recorded?

The score can shift during the session as the data set expands. Because the ranking is percentile-based within the current data, a print that scores 78 at 10:30 AM might score 71 by 2:00 PM if subsequent prints were more unusual on the same dimensions. The Top 25 leaderboard reflects current rankings, not the score at the moment of entry. For traders tracking specific prints across the session, the flag (EXTREME, ELEVATED, NOTABLE) is a more stable reference than the exact number, since the flag bands represent the top percentile tiers regardless of small intraday shifts in the exact score.

Keep learning