# How to Tell If an Option Was Bought or Sold: 4 Methods Source: SquawkFlow URL: https://squawkflow.com/learn/how-to-tell-if-an-option-was-bought-or-sold Category: Options Flow Published: 2026-09-16 Type: Explainer > Four ways to infer whether an option trade was buyer or seller initiated: bid/ask position, tape context, open interest, and the accuracy research. --- ## Every Option Trade Has Both a Buyer and a Seller Strictly speaking, no option is ever bought without also being sold. Every trade on the tape has a buyer on one side and a seller on the other. So when traders ask how to tell if an option was bought or sold, they are really asking a sharper question: **who initiated the trade?** Did an eager buyer cross the spread to get filled, or did a motivated seller hit whatever bid was available? That distinction, buyer-initiated versus seller-initiated, is what gives [options flow](/learn/what-is-options-flow) its meaning. A call that someone paid up to own carries a very different message than the same call dumped on the bid. The initiator is the party with urgency, and urgency is information. Here is the uncomfortable part: no field in the public data says "this was a buy." OPRA, the consolidated options tape, reports the price, size, exchange, and condition codes of each trade, not the intent behind it. Everything that follows is inference. Done carefully, that inference is right most of the time, and research covered in Method 4 puts an exact number on "most of the time." ## Method 1: Compare the Trade Price to the Bid and Ask The workhorse technique is the **quote rule**: compare the trade price to the prevailing bid and ask at the moment of execution. - **At or near the ask**: the buyer paid the seller's price. The buyer was the aggressor, and the trade is classified as bought. - **At or near the bid**: the seller accepted the buyer's price. The seller was the aggressor, and the trade is classified as sold. - **At the midpoint**: ambiguous. Neither side visibly crossed the spread, so the quote rule alone cannot classify the trade. A concrete example: suppose a [SPY](/ticker/SPY) call is quoted 2.10 bid, 2.16 ask. A 500-lot prints at 2.15. That fill sits a penny under the offer, meaning the buyer effectively lifted the ask with a touch of price improvement. Flow scanners will tag it as a buy. If the same 500-lot had printed at 2.11, the natural read is that a seller hit the bid, either exiting a long or opening a short. The Options Industry Council's primer on [bid and ask prices for options](https://www.optionseducation.org/news/understanding-the-bid-and-ask-prices-for-options) covers how those two prices form and how they aggregate across exchanges into the national best bid and offer (NBBO). One refinement matters for options specifically: compare the fill against the quote from the moment of execution, ideally the NBBO. Options quotes reprice constantly as the underlying moves, so a trade judged against a quote that is even seconds stale can land on the wrong side. The academic work in Method 4 measures exactly how often that happens. This is also the method working under the hood of most retail flow tools. SquawkFlow's [options flow page](/options-flow) applies the quote rule and says so in its disclosure: direction is inferred from last price versus bid and ask, not confirmed trader intent, and each row summarizes aggregate session activity rather than a single print. Any vendor whose labels come from the public tape is running some version of this same inference, whatever the marketing implies. ## Method 2: Read the Tape Around the Trade A single print gives you one data point. The surrounding tape gives you context, and context resolves many ambiguous cases. - **Repetition.** Ten consecutive prints at the ask over several minutes is a far stronger buy signal than one print at the ask. Someone is working an order and repeatedly paying up. - **Quote reaction.** If the offer gets lifted and immediately reloads higher, say 2.16 becomes 2.20 becomes 2.25, buyers are absorbing whatever liquidity market makers post. Persistent aggression moves the quote itself. - **Sweep condition codes.** An intermarket sweep hits multiple exchanges at once to take all displayed liquidity. Sweeps executing at or above the ask are the clearest footprint of an urgent buyer, and the mirror image holds for sell-side sweeps. The distinction between [sweeps and blocks](/learn/sweeps-vs-blocks-options) matters here: a negotiated block crossing at the midpoint says much less about aggression than a sweep does. - **Time of day.** Fills in the first and last half hour occur amid wider spreads and heavier hedging flow, which makes bid/ask classification noisier. ## Method 3: Check the Next Day's Open Interest The bid/ask methods tell you who was aggressive. Open interest answers something different but complementary: whether the trade **opened** new positions or **closed** existing ones. Open interest is the count of outstanding contracts, tallied once per day after the close by the [Options Clearing Corporation](https://www.theocc.com/). Because it updates overnight, the check is simple: - **Yesterday's volume far exceeded the prior open interest, and OI jumps this morning**: the bulk of that volume opened new positions. A 20,000-lot trade against 3,000 existing OI had to be mostly opening. - **Heavy volume but OI barely moves or falls**: positions were closed or rolled, not initiated. Combined with Method 1, the picture sharpens considerably. A fill at the ask plus a next-day OI increase reads as bought to open, a fresh directional bet. A fill at the ask with OI declining reads as bought to close, likely a short option position being covered, which carries a different meaning entirely. The full mechanics are in our guide to [volume versus open interest](/learn/options-volume-vs-open-interest). The limitation is symmetry: open interest rises whether the opener was the buyer or the seller. OI alone can never tell you direction. It only distinguishes open from close, which is why it works as a confirmation tool, not a starting point. ## Method 4: What the Classification Research Says Inferring trade direction is a studied problem in market microstructure, and the research puts hard numbers on how reliable these methods are. The canonical approach is the Lee-Ready algorithm, from Charles Lee and Mark Ready's 1991 Journal of Finance paper ["Inferring Trade Direction from Intraday Data"](https://onlinelibrary.wiley.com/doi/10.1111/j.1540-6261.1991.tb02683.x). It classifies trades with the quote rule and falls back on a **tick test**, comparing the price to the previous trade, for midpoint prints. For options specifically, Robert Savickas and Arthur Wilson tested the standard algorithms against CBOE records where the true initiator was known, in ["On Inferring the Direction of Option Trades"](https://www.cambridge.org/core/journals/journal-of-financial-and-quantitative-analysis/article/abs/on-inferring-the-direction-of-option-trades/FDA4541B57F78B2C8DCE129AFC25AAF0) (Journal of Financial and Quantitative Analysis, 2003). Their accuracy results: - **Quote rule:** 83 percent of trades classified correctly - **Lee-Ready:** 80 percent - **Ellis-Michaely-O'Hara rule:** 77 percent - **Tick rule alone:** 59 percent Three practical lessons fall out of those numbers. First, the simple quote rule, the same bid/ask comparison from Method 1, scored highest of the four rules tested (helped by the fact that it declines to classify midspread trades the other rules attempt), which is why serious flow scanners are built on it. Second, the tick test barely beats a coin flip for options, because option prices reprice with the underlying: a contract can print below the previous trade simply because the stock ticked down, with no change in who was aggressing. Never judge an option fill by comparing it to the prior option fill. Third, and least appreciated: the errors are not random. Savickas and Wilson found that misclassification concentrates in trades executing outside the quoted spread or on the "wrong" side of it, and that the likelihood of such trades is related to trading frequency, trade size, moneyness, and time to maturity. In plain terms, the big unusual prints that flow traders care most about are exactly the ones the rules misread most often. Even the best rule is wrong roughly one time in six, so treat every label as probabilistic and let repetition, Method 2, do the confirming. ## Where Classification Breaks Down Certain trades resist classification no matter how carefully the tape is read. Knowing the failure modes keeps you from over-trusting a label. - **Midpoint executions.** Price-improvement auctions and negotiated crosses fill between the bid and ask by design. The quote rule has nothing to grip. - **Multi-leg spreads.** A put purchase that is one leg of a collar or risk reversal is not a bearish bet, even though that leg prints as "put bought at ask." Condition codes flag spread executions; checking them comes before assigning sentiment to any single leg. - **Wide markets.** In an illiquid name quoted 1.00 bid at 1.60 ask, a fill at 1.30 is unclassifiable, and even fills near one side mean less when the spread is that forgiving. - **Deep in-the-money prints.** Large deep-ITM call volume is often dividend-related or financing activity between professionals, not directional opinion. - **Both sides can be motivated.** The aggressor framework assumes one passive side. Sometimes an urgent buyer meets an equally urgent seller, and the label captures only half the story. ## A Practical Checklist Pulling the four methods into one working sequence: 1. **Fill versus NBBO.** At or above the ask leans bought; at or below the bid leans sold; midpoint stays unresolved. 2. **Tape context.** Look for repetition, sweep codes, quote reaction, and whether the fill is a lone event or part of a campaign. 3. **Structure check.** Condition codes reveal spread legs before sentiment gets assigned, and premium spent means more when weighed against what is normal for that name. 4. **Next-day open interest.** Confirm open versus close, and reclassify accordingly: bought to close is not bullish conviction. 5. **Zoom out.** One well-classified trade is still one trade. The real skill is aggregating classified trades into a directional read, which is the subject of our guide to [reading options flow for direction](/learn/reading-options-flow-for-direction). Classification is inference, not observation; the tape never confesses. But with the quote rule as the foundation, the tape as context, and open interest as confirmation, an option trade can be signed about as accurately as the research says is possible, with a clear view of when not to trust the label. *Educational content, not financial advice. See our [risk disclosure](/risk-disclosure).* --- ## Citation Source: SquawkFlow, https://squawkflow.com/learn/how-to-tell-if-an-option-was-bought-or-sold Retrieved: 2026-09-16 08:57 ET Attribution: cite the page URL rather than a copied number. Levels are recomputed every session, so a number without its date is wrong within a day. SquawkFlow publishes market-structure data and education. Nothing here is investment advice, a price target or a recommendation, and we have no order execution.