Understand aggressor sides and sentiment

Start with execution classification

Aggressor categories summarize where an option execution occurred relative to eligible quote evidence. Contract Side Statistics exposes categories such as above ask, ask, midpoint, bid, below bid and unknown. The tape and flow charts can use that evidence to form conventional directional summaries. Read those summaries as interpretations of classified prints, not as a complete record of participant positions or a verified buyer/seller identity.

Inspect one classification before totals

Choose a small symbol scope and match the same filters in the tape and side-statistics tool.

1. Open a tape row and record exact contract, time, execution price and reported side.
2. Open Contract Side Statistics for the same symbol/session.
3. Choose premium, volume or trade count.
4. Inspect midpoint and unknown categories alongside classified bid/ask activity.
5. Compare totals only after matching expiry, right, size and trade-type filters.

Signs and option rights matter

A buyer-initiated call and a buyer-initiated put have different conventional directional readings. Selling a put can be part of bullish exposure, but can also close an earlier long put. A net series depends on which categories and rights its calculation includes. Unknown quantity is not a nuisance to remove; it tells you how much admitted activity lacks usable directional classification. Ratios based only on classified volume should keep that denominator explicit.

Worked example: a mixed-right position

Imagine a call prints near the ask and a put prints near the bid around the same time. A sentiment view can classify them in the same conventional direction. That does not prove they belong to one participant or opened a synthetic position. Conversely, two apparently opposing prints might be legs of a spread. Inspect each exact identity and the available group detail, then state only the observed classification. Do not turn a color agreement into a claim about beneficial ownership.

Troubleshoot side differences

First compare whether one view is premium-weighted and another contract-weighted. Then check whether unknown or midpoint rows are included, and whether a consolidated row represents several constituents. A correction or eligible quote change can alter an earlier classification. For a disputed row, capture exact identity, execution time, reported price and side plus the relevant filters. A screenshot of the aggregate green/red bar alone cannot reproduce the classification evidence.

Align filters, measurements and coverage

Premium, contract volume and trade count measure different things. Ten contracts at a high option price can contribute more premium than many contracts at a low price. Gross premium, signed net premium, call-minus-put drift and an individual time bucket are also different aggregates. Before reconciling tools, match the symbol family, exact roots where relevant, trading date, session, expirations, call/put choice, size and premium thresholds, aggressor filters and any trade-type selection.

A live stream can begin after the session started. Its current rows may be fresh while an earlier prefix is missing. Historical repair can add eligible prints and recompute older totals without representing new live trades. Empty filtered results mean no admitted matching rows, not necessarily no market activity. Read the reported source times, observed bounds and partial-coverage state. For an apparent mismatch, reproduce it with broad filters first, then add one filter at a time and compare the same measurement over the same time interval.

Audit the denominator behind a directional percentage

Suppose a side statistic reports a high ask-side percentage. Determine whether it weights premium, contracts or trade count, then check whether midpoint and unknown observations are part of that denominator. A percentage of classified activity can differ from a percentage of all admitted activity. Two displays can therefore agree on every print while showing different percentages because they answer different questions. Write the measurement and included categories beside any directional conclusion.

Classification at or beyond a quote also requires matching temporal evidence. A quote displayed now is not automatically the quote eligible when a past execution occurred. Comparing an old trade against today's spread can make a valid historical classification look wrong. Use the row's recorded classification and source time as the starting point, and avoid silently recomputing it from a current drawer. When escalating a suspected error, provide the exact print plus its original observed side and the basis of your comparison. That is more actionable than saying the overall sentiment color should be reversed.