Does following unusual options activity work?

By Faizan Ahmad. Updated .

Not as a standalone buy or sell signal. Research shows options volume can carry real information, but the part that predicts prices comes from knowing who opened new positions, which no public feed shows, and many big trades are hedges, spreads, rolls or closing trades that only look like bets. Used as context, flow is genuinely useful: it tells you where real money is committed, how much and by when, which helps you test an idea rather than replace one.

What does the research say?

The best-known study, a 2004 NBER working study by Jun Pan and Allen Poteshman, used a special Cboe dataset of options volume initiated by buyers to open new positions. Stocks with low put/call ratios on that measure beat stocks with high ones, on a risk-adjusted basis, by more than 40 basis points the next day and more than 1% over the next week. The authors traced the edge to non-public information held by option traders, not to a market inefficiency anyone could harvest.

That detail is the catch. Their data knew who bought and whether the trade opened a position. The public options tape doesn't: it carries price, size, time and exchange. Flow tools infer the side from where a trade filled against the bid and ask, and infer opening from size versus open interest. Cboe sells the real breakdown (buy or sell, open or close, by type of trader) as a separate dataset for its own exchanges.

A second study, of 1,859 US takeovers, found abnormal options volume before about 25% of them. Informed trading exists, but it clusters in a narrow setting, short-dated calls on takeover targets, and you can't tell in real time which prints are the informed ones (more on insider trading and options).

Why do so many unusual trades mislead?

Did the flow work on October 30, 2026?

On October 30, 2026, the whole market's options flow leaned bearish (net -$86.53M) and the S&P 500 fell 1.59%. Here are the names with the strongest leans, and what each stock did that day:

Most bullish net premiumStock that dayMost bearish net premiumStock that day
C $93.80M-2.6%MSTR -$44.61M-6.7%
GOOGL $58.61M-0.8%TSLA -$32.94M-0.8%
MRVL $31.01M-9.4%AMD -$28.20M-10.3%
META $23.40M-1.7%TSM -$26.42M-5.3%
NVDA $19.61M-3.9%CRWD -$20.89M-4.0%
TER $17.47M-7.0%BE -$15.40M-8.0%

The bearish six did fall further on average (5.9% against 4.2%), but all twelve stocks fell. And a same-day score is the wrong test anyway: the biggest Citigroup calls behind that $93.80M expire December 18, so whether they worked won't be known for weeks. Be wary of anyone who grades flow by the next candle, or shows you only the winners.

When does unusual activity help, and when doesn't it?

How you use itDoes it work?Why
Copying single alertsPoorlyYou don't know the trader's full position, cost or exit, and many prints are hedges or legs
Checking your own ideaOften helpfulFresh, aggressive money on your side, or against it, is real information
Reading market moodUsefulNet premium and index put buying show how nervous big holders are
Following repeat buying across daysBetter than one printSeveral sessions of opening trades at one strike are hard to explain away
Earnings playsMixedOptions already price the expected move; a buyer can be right on direction and still lose to IV crush

How to use it without getting burned

  1. Start with your own reason to trade, and use flow to check it.
  2. Favor opening trades (size above open interest) filled at the ask or the bid, not the mid.
  3. Read the whole structure: other legs printed at the same second change everything.
  4. Weigh the dollars against the stock's normal options volume.
  5. Match horizons: a 2027 hedge says nothing about this week.
  6. Don't chase after the stock has already moved (why flow alerts seem late).
  7. Size small. The other side of every trade is often a professional.

How Tradytics tests which flow works

Tradytics reads the full OPRA tape: every trade on every US options exchange, live, about a second after it prints. It stitches fills into orders, tags each as bought or sold, and on the Live options flow page the Opening tag, the side chips and the Spread and Complex tags cover most of the checks above; type what you want to see in plain English and save it as a view. Tradytics' Signals are live orders that match patterns from a study of 560 million pattern and horizon combinations tested on 10.9 million options prints over 378 sessions, joined to real 1-minute stock bars and real option quotes and kept only where they held up in both halves of the period. Only patterns that held up became signal patterns, such as repeat call buying in ETFs and call volume far above open interest, and each signal is tracked to expiry (how they're built: how Tradytics Signals are built). When a print still looks ambiguous, ask the AI analyst: it pulls the trade's context and says whether it reads as a hedge, a spread or a fresh bet. Free shows the previous session, enough to learn how flow behaves; Plus and Pro show it live (is paying for options flow worth it?).

On Tradytics: clean options flow from the full tape, Signals from a study of 560 million pattern and horizon combinations, and every order since January 3, 2022 in the options flow API and the backtester.

Common questions

What percentage of unusual options trades make money?

There is no trustworthy public number. A win rate published without the full list of trades, losers included, isn't evidence.

If following flow worked, why would anyone sell the data?

Fair question. Flow data is a tool, like a chart: it shows you the tape faster and cleaner, but the judgment is still yours. Nobody honest is selling a guaranteed signal.

Is unusual put buying a warning sign?

Sometimes, but index put buying is mostly insurance. On October 30, 2026, index options carried a put/call ratio by premium of 1.26 against 0.91 for single stocks.

Do big traders get it right more often?

They have more information, but they also hedge, and a hedger is happy when the insurance expires unused. Copying a hedge as if it were a bet is a common way to lose.

Sources