Options flow data explained

What options flow data is

Every options trade in the US generates a public record. By law, all options transactions must be reported to the Options Price Reporting Authority (OPRA), which consolidates trade data from all US options exchanges, CBOE, NYSE Arca, Nasdaq PHLX, MIAX, BOX, and others, into a single national feed.

Options flow data is this feed, made accessible. A flow service subscribes to the OPRA feed, parses each print as it arrives, adds derived fields (total premium, Vol/OI ratio, execution classification), and presents the results in a readable format, a live stream of every significant options transaction.

The word "flow" comes from the concept of money "flowing" through the market via options purchases and sales. When a large institution buys $2M in calls, $2M in premium "flowed" into that position. Options flow data captures this movement in real time.

Where options flow data comes from

The data pipeline from trade to your screen has four stages:

1. Trade execution: An institutional buyer places an order. It's routed to one or more exchanges and fills against existing limit orders or market maker quotes.
2. Exchange reporting: Each exchange where any part of the order fills immediately transmits the trade report to OPRA.
3. OPRA consolidation: OPRA publishes a consolidated feed of all options trades across all exchanges, available in near-real-time to licensed subscribers.
4. Data vendor processing: A flow service (like RadarPulse) subscribes to the OPRA feed, parses each print, enriches it with derived fields, and delivers it to users.

The critical point: OPRA access requires exchange licensing. This is why real-time options flow services cost meaningfully more than delayed alternatives, the underlying data cost is substantial. Delayed flow (15-minute lag) uses a cheaper data tier that doesn't require the same licensing overhead.

Every field on a print, decoded

A single options trade generates the following data fields. Understanding each one individually is the prerequisite for reading prints accurately.

How the data becomes a scored signal

Raw flow is thousands of prints per day, most of them carrying minimal directional information. A market maker hedging, a retail trader buying 2 contracts, a fund rolling a spread leg, all appear as prints in the raw feed. The signal is buried in the noise.

Scoring transforms the raw feed into a ranked list. RadarPulse's 0–100 score weights the four most predictive context factors:

The resulting score tiers:

Real-time vs delayed flow

Options flow data has two main delivery tiers:

Real-time (live): Data arrives seconds after the trade executes. Requires full OPRA licensing. Best for active traders who want to act on sweeps as they happen. Price-sensitive: OPRA licensing costs are passed on to subscribers, which is why real-time flow platforms are in the $150–$500/month range.

Delayed (15-minute): Data arrives 15 minutes after execution. Uses a less expensive data license. Still useful for identifying patterns, but by the time you see a sweep, the immediate price reaction (if any) has already occurred. More appropriate for research, watchlist building, and end-of-day review than for same-day reaction trading.

RadarPulse delivers real-time flow to Elite tier subscribers and delayed flow to Pro tier. The scored feed is available on both; the difference is in when the prints arrive.

What flow data cannot tell you

Understanding the limits of flow data is as important as understanding what it shows.

• It cannot identify who placed a trade. Market participants are anonymous in the public tape.
• It cannot distinguish a new position from a closing trade with certainty. Vol/OI helps, but isn't definitive.
• It cannot tell you whether the institution is right. High-score sweeps fail regularly.
• It cannot identify spread legs definitively. Multi-leg strategies route individually and can't always be matched.
• It cannot predict exactly when a thesis will play out, only that a sophisticated actor has positioned for a specific direction by a specific date.

These limits don't reduce the utility of flow data, they define its role: a directional bias indicator, strongest when combined with technical analysis, macro context, and cross-domain data (congressional trading, institutional 13F holdings).

The OPRA feed in detail: what "real-time" actually means

OPRA, the Options Price Reporting Authority, is a joint venture of the US options exchanges, created under a plan approved by the SEC. It is the sole consolidated source for US options trade data, carrying reports from all 16+ active US options exchanges simultaneously.

OPRA publishes two distinct data streams. The first is last sale data: the record of completed trade prints, every executed options transaction, which is what we call "flow." The second is quote data: the current best bid and ask at each strike and expiry across all exchanges. Quote data is what determines whether a trade filled at the ask (buyer was aggressive) or at the bid (seller was aggressive), one of the most important fields on any print.

Both streams are published continuously during market hours. The rate of OPRA data is not trivial: on active days the tape can generate hundreds of thousands of messages per second across all names and exchanges. Processing this in real time requires infrastructure purpose-built for high-throughput market data.

The latency reality

"Real-time" OPRA data arrives in milliseconds at exchange co-location servers in New Jersey, CBOE's infrastructure is in Secaucus; MIAX operates out of Princeton. The physical proximity of a data center to the exchange's matching engine is what determines the first hop in the latency chain. By the time a trade print travels from the exchange matching engine to OPRA's consolidated feed and then to your browser via a flow service, the total elapsed time is typically 100ms to 2 seconds depending on network hops and processing infrastructure. For the purpose of flow analysis and swing trade positioning, this is functionally real-time. You are not competing with high-frequency traders; you are reading institutional positioning signals that play out over days or weeks.

The 15-minute delay tier

The delayed feed uses OPRA's lower-cost licensing tier, which in practice is often processed through FINRA's consolidated tape rather than direct exchange feeds. The 15-minute lag is exactly what it says: a sweep that prints at 10:00:00 AM appears in a delayed feed at 10:15:00 AM. This has a specific practical consequence, the immediate price reaction to an institutional sweep (which can occur in the first 5–15 minutes after the print) has already happened by the time a delayed subscriber sees the signal. Delayed flow is more appropriate for end-of-session review, daily watchlist construction, and multi-day pattern research than for same-day entry timing.

Multi-exchange sweep mechanics: how sweeps actually work

When an institution needs to buy 10,000 call contracts quickly, a smart order router simultaneously routes the parent order to multiple exchanges. In US options markets, the active venues include CBOE, CBOE C2, NYSE Arca Options, Nasdaq PHLX, MIAX, MIAX Emerald, MIAX Pearl, BOX Options Exchange, Cboe EDGX Options, and several others, each running its own matching engine with its own pool of resting limit orders and market maker quotes.

How a sweep appears on the tape

Instead of one print for 10,000 contracts, a sweep generates 8–15 smaller prints in rapid succession, each from a different exchange, each a partial fill of the parent order, together totaling 10,000 contracts. The individual prints might look like: 840 contracts at MIAX, 1,250 at CBOE, 900 at NYSE Arca, 1,100 at PHLX, 2,200 at CBOE C2, and so on, all within the same 200–500 milliseconds. Flow services aggregate these into a single logical sweep record, summing the contracts and computing the total premium. Without this aggregation logic, the raw tape would show 12 seemingly modest prints instead of one large institutional trade.

The timing signature

The mathematical definition of a sweep in flow parsing: prints within 500ms of each other, in the same underlying contract (same ticker, same expiry, same strike), across multiple exchanges, with matching directional aggression. Some services use tighter windows (250ms or 300ms); some use looser ones. The window choice affects how many prints get merged and how the total premium and contract count is summed. RadarPulse uses a multi-exchange detection window tuned to minimize both false merges (treating unrelated trades as one sweep) and false splits (treating one institutional order as multiple separate events).

Why sweeps signal urgency, and what urgency actually means

The multi-exchange routing strategy deliberately sacrifices price improvement for guaranteed immediate execution. A patient institution could post a limit order and wait for sellers to come to them, potentially getting a better fill price. By sweeping multiple exchanges simultaneously, they guarantee they get filled at current market prices, but they accept whatever ask prices are resting across all those venues at that moment, which is usually the worst available fill.

This trade-off is the signal. An institution willing to pay up across multiple exchanges rather than wait for a better price is communicating that time is more valuable to them than price, which means they believe something will happen soon. The urgency embedded in a sweep is the key information content that distinguishes it from a passive block trade at the same total premium size.

Open interest: what it really measures

Open interest (OI) is one of the most misunderstood fields in options data, and it's the denominator of the most important derived metric in flow analysis, the Vol/OI ratio.

OI is the total number of outstanding options contracts at a specific strike and expiry that have not yet been closed, expired, or exercised. It represents the total outstanding position, the cumulative result of all previous trading activity at that contract. When two counterparties open a new position (one buys, one sells/writes), OI increases by one. When an existing holder closes their position, OI decreases by one. When a contract expires worthless, OI decreases by the number of contracts held.

The one-day lag

OI updates with a one-day lag: the OI figures visible in flow data today reflect activity through yesterday's close. If 50,000 new contracts were opened today, those will appear in tomorrow's OI figure. This means the Vol/OI ratio you see in real-time flow is calculated against yesterday's OI, which is the correct calculation, because it tells you how today's volume compares to the existing open position at that strike, not a moving target that includes today's own prints.

Why Vol/OI is the most useful single filter

A strike with 100 existing OI means the market has been mostly indifferent to that contract, very little positioning has accumulated. When today's volume at that strike is 5,000 contracts, the Vol/OI of 50× means an enormous amount of new money is entering relative to the existing position. Nearly all of those 5,000 contracts are opening new positions, there aren't enough existing holders to generate that kind of closing activity. This is a high-confidence signal of new institutional intent.

By contrast, a strike with 50,000 existing OI has seen substantial historical activity. When today's volume is 5,000 contracts, the 0.1× Vol/OI is ambiguous, it could represent new additions, partial closes, rollers, or spread adjustments within the existing position base. No single interpretation is strongly supported. The Vol/OI filter is the mechanism that separates "new money entering a position" from "existing position activity", which is the fundamental question flow analysis is trying to answer.

The OI building and declining patterns

OI trends across consecutive sessions carry their own signal. When OI at a specific strike grows by 5,000 contracts over three consecutive sessions while the stock price is relatively flat, an institution is accumulating a position over multiple days, buying gradually to avoid moving the market. This patient accumulation pattern is different from a single large sweep, but can represent equally substantial institutional conviction. When OI declines at a previously high-interest strike ahead of its expiry, it means holders are closing or rolling the position, which can itself be informative about how the original institution views the remaining thesis.

The dark pool and options flow combination

Options flow and dark pool data are separate feeds from entirely different reporting mechanisms. Options trades go through the OPRA consolidated tape. Equity dark pool trades, large institutional block transactions executed off the lit exchanges, are reported to FINRA's Trade Reporting Facilities (TRF) and appear in a separate equity dark pool feed. They are different instruments tracking different markets, but when used together they produce one of the most powerful multi-signal setups available in public market data.

The cross-instrument confluence pattern

When a stock sees large dark pool prints (big equity blocks trading off-exchange) and unusual options call accumulation in the same name within the same session or adjacent sessions, the two signals together are substantially more convincing than either alone. The dark pool prints say: "A large institution is buying this stock in size, and they are routing it off the lit exchange to avoid market impact, they don't want you to see them accumulate." The options call sweeps say: "A large institution is positioning for upside in this stock via options, committing substantial premium to a directional bet."

The key analytical value of this combination is redundancy across instruments. A single large options sweep might be a hedge, a spread leg, or a mechanical corporate program. A single dark pool print might be index rebalancing, a corporate buyback, or a pension fund's routine portfolio adjustment. But when two independent instruments, in two separate markets, through two separate reporting mechanisms, from what may be two separate institutions, point in the same direction in the same name, the probability that either is purely structural or mechanical drops significantly. The overlap is the signal.

How RadarPulse surfaces this combination

RadarPulse's confluence panel flags names where options flow signal and dark pool activity appear in the same ticker within the same session window. The combination of a high-score EXTREME options print and same-session dark pool block activity in the same direction is one of the highest-conviction multi-source setups the platform can surface, two independent data streams saying the same thing about the same stock.

How to read options flow alongside 13F filings

Institutional investors managing more than $100M in assets must file quarterly 13F reports with the SEC, due within 45 days of each quarter's end. A 13F discloses the institution's current long equity holdings, what they own in stock, in what size. While 13F filings are lagged by up to 135 days from the actual position date, they reveal the known institutional ownership landscape for any given name.

Cross-referencing 13F holdings with current options flow produces a critical disambiguation tool, the ability to evaluate whether a large options print is likely directional (new positioning) or structural (hedging an existing equity position).

The known-holder interpretation

When a large institution's most recent 13F shows a billion-dollar long position in a stock, and put flow appears in that name, the put flow is substantially more likely to be portfolio protection, the institution hedging their own long equity exposure, than a bearish directional bet against the stock. The institution already owns the stock. Buying puts is the textbook way to protect that position against a downturn. This is not a bearish signal; it is a risk management signal from a bullish existing holder.

When a stock has no significant institutional representation in recent 13F filings, meaning major institutions don't appear to hold meaningful equity positions, and a large call sweep appears in that name, the sweep has no obvious structural hedging motivation. It is more likely to be a pure directional bet: an institution opening new exposure to a name they don't currently own, using options rather than equity for the leverage and defined risk profile.

Practical application

Before interpreting large options prints in heavily-owned large-cap names, AAPL, MSFT, AMZN, NVDA, META, it is worth checking the latest 13F filings to understand who is already positioned and in what size. Put flow in names where every major institution holds multi-billion-dollar long positions has a much higher base rate of being protection than directional bearishness. The same put flow in a mid-cap name with minimal institutional ownership is a cleaner directional signal. The 13F context doesn't change the raw print, but it substantially changes the appropriate interpretation of what that print likely represents.

False positives in options flow data: what they look like

Options flow data contains multiple categories of prints that appear unusual or directional on the surface but carry no genuine directional information. Recognizing these categories is a core competency for reading flow accurately, every experienced flow analyst has a mental checklist of structural trade types to rule out before treating a print as a signal.

Using premium decay to time entries around flow signals

Understanding theta, the daily time decay cost of an options position, is essential for determining how to use a flow signal as an actionable entry, and how long you have before the underlying thesis must materialize.

Theta is the dollar amount an options position loses per day purely from the passage of time, holding all else equal. It is not a linear function: options lose time value faster as they approach expiry. An option with 30 DTE has moderate daily theta. The same option at 7 DTE has much higher daily theta per dollar of remaining premium, the clock is accelerating.

DTE as a thesis timeline encoder

When interpreting a flow print, the DTE tells you the institution's thesis timeline. A large call sweep in 30-DTE options says: "I believe this will move meaningfully within the next 30 days." The institution is paying roughly (total premium) ÷ 30 per day in theta, they need the stock to start moving soon, or the position will decay toward zero regardless of whether they are ultimately right about direction.

LEAPS sweeps, options with 180 or more days to expiry, have very low daily theta. The institution has purchased time. They can afford to be wrong about timing while being right about direction. A 6-month LEAPS sweep is a patient, structural, multi-month directional thesis. It will not play out in a week. The urgency of the sweep execution signals conviction about the direction; the DTE signals the expected timeframe.

Matching your entry DTE to the original print

When using a flow signal to inform a personal position, the DTE of the original print is a reference point. If an institution sweeps 45-DTE calls on Monday, and you enter the same name with 45-DTE calls on Tuesday, your theta exposure is similar to theirs. If you enter the same name with 20-DTE calls two weeks later, you have substantially more theta risk per dollar invested, you need the move to happen faster than the original institution did. This DTE mismatch is one of the most common ways flow followers take on more risk than the original signal implied. Matching the DTE range of the original print, or going to longer DTE, keeps your risk profile closer to the institution's original position structure.

The theta decay danger zone

Options lose roughly 50% of their remaining time value in the last 30 days of their life. If a flow signal generates a setup but the stock has not moved meaningfully by the halfway point of the original DTE, the position is in the danger zone, theta is accelerating and the remaining premium can evaporate quickly even if the directional thesis eventually proves correct. Recognizing this time horizon is part of position management: a NOTABLE-tier print in a 45-DTE option has a 45-day window; if the stock is flat at day 35, the position structure is deteriorating regardless of the underlying thesis.

Options data quality: exchange latency, timestamps, and known issues

Flow data has several well-understood quality issues that experienced practitioners account for when interpreting prints. None of them invalidate the signal value of the data, but they affect how prints should be read and how edge cases should be handled.

Timestamp accuracy and sweep timing

Trade prints are timestamped by the exchange matching engine at the moment of execution. Across different exchanges, there can be 10–50ms differences in how timestamps are reported, caused by differences in clock synchronization and reporting latency between exchange infrastructure. When a sweep spans multiple exchanges and all prints are aggregated into a single sweep record, the individual print timestamps span this range. Flow services typically use the earliest print's timestamp as the sweep's reported time. This means the displayed timestamp for a sweep reflects the first partial fill across all exchanges, which is the most accurate representation of when the institutional order began executing.

Erroneous prints and busted trades

Occasionally, exchanges publish trade reports that are subsequently marked as erroneous, so-called "busted trades" that were executed at clearly erroneous prices due to data errors or matching engine issues. These appear in the live flow feed briefly and are then formally withdrawn by the exchange. Well-built flow services handle erroneous print retraction gracefully, removing the print from the feed and adjusting any derived metrics (open interest updates, running totals). Poorly built services display erroneous trades permanently, polluting the historical record and the live feed. The practical impact is small in normal conditions but can be significant around periods of market stress or exchange technical issues, when erroneous print rates increase.

Friday afternoon block activity

The final 15–30 minutes of trading on Fridays generates a disproportionate share of large block options activity as institutions rebalance their books for the weekend. This creates a concentration of large-premium prints in the close window that does not reflect Monday-morning directional intent, it reflects Friday closing mechanics. A $2M call block that prints at 3:55 PM on a Friday is more likely an institutional spread adjustment, a delta hedge rebalancing, or a risk-off position reduction than a new directional bet. Friday close flow should be flagged with appropriate skepticism, particularly block prints (as opposed to sweeps, which retain their urgency signal even at Friday close).

Earnings-adjacent noise

In the session immediately before a company's earnings announcement, options flow data almost always shows elevated call and put volume across multiple strikes. This is standard pre-earnings positioning, institutions buying straddles and strangles (both calls and puts simultaneously) to profit from a large move in either direction, without a directional view. The presence of large call sweeps the day before earnings does not necessarily indicate bullish conviction; it may represent the call leg of a straddle, with a matching put leg visible elsewhere in the tape. Earnings-adjacent flow is the most noise-contaminated period in the options calendar, the signal-to-noise ratio drops sharply in the 24–48 hours before any major earnings event.

Multi-session confirmation: when is a signal "confirmed"?

A single session's unusual options flow print, even an EXTREME-scored sweep, is a hypothesis, not a confirmed signal. The hypothesis is: "A sophisticated institutional actor has positioned for a specific directional move in this name by a specific date." The follow-on question is: how does that hypothesis strengthen or weaken across subsequent sessions?

The multi-session confirmation protocol

A structured approach to confirming flow signals across sessions:

Using RadarPulse for multi-session confirmation

RadarPulse's multi-day flow view shows which names appeared in the top signals across multiple consecutive sessions, enabling this confirmation protocol without manual tracking. Names with four or more consecutive sessions of ELEVATED or EXTREME call flow in the same direction represent the highest-conviction multi-session setups the scanner can identify, sustained institutional accumulation over multiple days, visible in a platform designed to surface exactly this pattern. The confluence panel further enriches this by flagging when these multi-session flow names also have congressional activity, dark pool overlap, or cross-sector co-occurring flow that could act as a catalyst for the thesis to play out.

Putting it together: a complete flow analysis framework

The preceding sections cover the components of flow analysis individually. A complete framework integrates them in sequence:

1. Filter for score tier first. Start with EXTREME prints. Move to high-ELEVATED if the universe is thin on a given day. Ignore NOTABLE for directional purposes, it is a watchlist function, not an action trigger.
2. Check Vol/OI. Any print below 2× Vol/OI warrants skepticism that this is new positioning, it may be activity within an existing position base. Above 5× is strong evidence of new institutional entry.
3. Evaluate execution type. Sweep plus ask-side fill is the strongest execution combination. Block plus bid-side fill is the weakest. Sweeps retain their urgency signal regardless of session timing; blocks at Friday close are structurally suspect.
4. Rule out known false positive categories. Is this a spread leg? A pre-earnings straddle? A VIX settlement week? A name with known corporate buyback + covered call programs? Eliminate structural explanations before treating the print as directional.
5. Check 13F context. For large-cap names, is this print in a name with known massive institutional long exposure? If so, put flow may be hedging. No institutional 13F exposure + call flow = cleaner directional signal.
6. Check DTE for thesis timeline. What is the implied timeframe? Under 14 DTE means the institution expects something in days. 30–60 DTE is the primary swing trade range. LEAPS suggest a patient, multi-quarter thesis.
7. Check for cross-domain confluence. Does this name also have congressional activity? Dark pool prints? Elevated news sentiment? Multiple independent data sources pointing in the same direction increase the probability that the options flow is genuinely informational rather than structural.
8. Confirm across sessions. Check OI the next morning. Watch for accumulation. A hypothesis with single-session evidence is early-stage. The same hypothesis with three sessions of accumulating OI and sustained flow is substantially confirmed.

This framework does not guarantee outcomes, options flow analysis is probabilistic, not predictive. High-score sweeps fail regularly. Confirmed multi-session accumulation patterns reverse on unexpected news. The framework exists to systematically increase the probability that any given print you act on represents real institutional intent, rather than mechanical, structural, or hedging activity that appears similar on the tape but carries no directional information.

Related reading