Most Options Flow Data Is Noise
Disclaimer: All ticker prices, premiums, and return calculations shown are examples for educational purposes and reflect market conditions at the time of writing. They are not trade recommendations. Options trading involves significant risk of loss. Past performance of any strategy does not guarantee future results. Consult a licensed financial professional before trading.
Every day, millions of options contracts trade on US exchanges. A small percentage of those are institutional size. Sweeps, blocks, dark pool prints. The stuff that might tell you where the big money is going before it gets there. Flow scanners surface these trades. The problem is they surface all of them. On a normal trading day, a typical scanner spits out 200-500 "unusual" options trades. You're obviously not supposed to follow all of them. But which ones matter? I spent a year tracking institutional flow and building a system to answer that question. It comes down to one thing: conviction.
Raw Flow Alerts Are Mostly Useless
Here's what a typical alert looks like:
AAPL 05/16 $230 Calls. 5,000 contracts. $2.3M premium. Sweep.
Looks bullish. Someone just dropped $2.3 million on AAPL calls. Must know something, right?
Without context, this alert is close to meaningless.
It might be a hedge. An institution sitting on 500,000 shares of AAPL might buy calls as part of a collar or as upside protection against a short position elsewhere. The trade looks bullish by itself but the overall position is actually bearish or neutral.
It might be one leg of a spread. That $230 call purchase could be paired with a $240 call sale that shows up as a completely separate alert. The actual position is a vertical spread with capped upside, not an aggressive directional bet. But the scanner shows you two separate alerts and you have no idea they're connected.
5,000 contracts sounds massive until you realize AAPL trades 500,000 call contracts on a normal day. It's a rounding error. On a name like GTLB or IONQ, 500 contracts would be the story of the week.
And timing changes everything. A sweep at 9:35 AM during the opening chaos means something completely different than the same order at 2:45 PM after a trader has watched the tape all day and decided to commit.
This is why most people who try to "follow the flow" end up frustrated. They see a wall of alerts, chase the biggest dollar amounts, and get random results.
What Conviction Scoring Actually Looks Like
Instead of treating every large trade like it matters equally, conviction scoring assigns a number based on how many factors line up to suggest this is a real directional bet from someone with an informational edge.
Here's what I look at:
Premium relative to the stock, not the dollar amount. A $1M sweep on a $500 stock is a different conversation than $1M on a $20 stock. The absolute number matters less than what that premium represents relative to just buying the shares outright. Bigger relative premiums mean more skin in the game.
How the order was executed. Sweeps route across multiple exchanges simultaneously because the trader needs to get filled before the market moves. That urgency is a conviction signal. Block trades are negotiated on a single exchange, which suggests less time pressure. Orders split across hours suggest even less. The execution method tells you how badly someone wanted this position right now.
Volume versus open interest. If today's volume on a specific strike is 5x the existing open interest, those are new positions being opened. Not someone closing out or rolling. A V/OI ratio above 3-5x is when I start paying attention.
The Factors That Actually Matter
How far out of the money. Deep OTM options are cheap and risky. An institution buying them is making a leveraged bet that requires a bigger move to pay off. The further from the current price, the more conviction it takes to pull the trigger, because the probability of profit drops with every dollar of distance.
Time to expiration. Weekly options have brutal theta decay. Buying them means you need the move to happen fast. An institution buying weeklies is expressing urgency. Someone buying LEAPs six months out might just be establishing a slow position. Different conviction levels entirely.
Repeated hits at the same strike. This is the single strongest signal I've found. One sweep could be anything. Three sweeps at the same strike from different times of day and different exchange routes? That's not a coincidence. That's multiple desks arriving at the same conclusion independently.
IV rank context. When IV rank is elevated, options are expensive relative to their own history. An institution paying above-average premiums is saying "I'm willing to overpay because I'm that confident in the direction." That means something.
What the sector is doing. A bullish call sweep on a tech stock while the whole sector is selling off is a much louder signal than the same sweep during a broad rally. Positioning against the prevailing trend requires conviction because you're disagreeing with the crowd.
Putting It on a Scale
When you stack these factors together, each trade gets a composite score. I use 1-10.
Scores 1-3 are noise. Big trades that only check one or two boxes. Probably hedges, spread legs, or routine rebalancing. I ignore these completely.
Scores 4-6 go on a watchlist. Interesting enough to monitor, especially if the same ticker keeps showing up. Not enough to act on alone.
Scores 7-10 are where I focus. Multiple factors lining up. Large relative premium, sweep execution, high V/OI, meaningful OTM distance, and ideally repeated hits throughout the day. These are the trades worth digging into.
The Part Everyone Skips
Even with conviction scoring, you need to know if it actually works. This is where almost every trader drops the ball.
The answer is paper trading. Not the kind where you open a spreadsheet, log three trades, forget about it for two weeks, and never look at it again. We've all done that. It doesn't work.
Automated paper trading. High conviction alerts get tracked automatically with preset take-profit and stop-loss levels. You don't have to remember to check. You don't have to manually calculate P&L. After 30-60 trades, you open the journal and the data is just there.
What percentage of 7+ conviction trades were profitable? What was the average return? Do sweeps outperform blocks? Do certain sectors produce cleaner signals? Is the system actually better than random?
That's what turns flow analysis from gambling with extra steps into a research process with measurable results.
I built this into Flow Proof because I couldn't find it anywhere else. Every trade that scores 7+ gets paper traded automatically. After a month you have a real journal showing what worked and what didn't.
What Six Months of Data Taught Me
Some patterns that showed up after tracking hundreds of high conviction trades:
Repeated hits are the strongest signal by a wide margin. Trades that scored high primarily because the same strike got hit multiple times in a day had a meaningfully better win rate than trades scoring high for other reasons. If I could only look at one factor, it would be this one.
Earnings plays are a coin flip no matter what the conviction score says. Institutional flow before earnings is so heavy and contradictory that even scores of 8-9 don't reliably predict direction. I filter out any trade within 5 days of an earnings announcement now and my results improved immediately.
Sector rotation context is huge. High conviction bullish flow in a sector that money is rotating into performed dramatically better than the same signal in a sector seeing outflows. The macro backdrop either amplifies or cancels the signal.
The best signals come from boring stocks. AAPL and TSLA have so much options volume that even "unusual" activity is just Tuesday. Mid-cap stocks with thinner options markets produce much cleaner signals because any institutional-size trade is proportionally significant. GTLB at position 1 in my search console data is a perfect example. Low volume name, clean signals, easy to read.
Where to Start
If you want to try this:
Start with any free flow tool. Barchart, the free tier of Flow Proof, or Unusual Whales free. Just watch raw flow data for a week and get a feel for what shows up.
Before you act on any alert, ask yourself four questions. Is this a sweep? What's the volume relative to open interest? How far out of the money is the strike? Has this same strike been hit more than once today?
Paper trade your first 30 signals. Track the results honestly. Most people skip this and it's the most important step in the entire process.
After 30 trades, look at the data. What worked? What didn't? Adjust your filters. Run another 30. That's how you build a system instead of guessing.
The goal is not to follow every big trade. It's to find the small percentage that represent genuine conviction and then prove to yourself with data that your framework for identifying them actually holds up.
Key Takeaways
Related Research & Tools
Try the Free Put Scanner
Flow Proof's conviction scoring does this automatically. Every trade scored 7+ gets paper traded so you can see what actually works before risking real money.
Open Scanner →Frequently Asked Questions
How many flow alerts per day should I actually look at?
Most scanners show 200-500 unusual trades per day. After conviction scoring, maybe 10-20 are worth monitoring and 3-5 are worth digging into. If you're looking at more than that, your filters aren't tight enough.
What's the difference between a sweep and a block?
A sweep routes across multiple exchanges simultaneously to get filled fast. The trader is paying more for speed because they need the position now. A block is negotiated on a single exchange, usually at a discount. Sweeps express more urgency and generally indicate higher conviction.
Can retail traders really benefit from following institutional flow?
Yes, but not by chasing every alert. Retail traders benefit from institutional flow as a confirmation signal. If your own analysis says a stock is a good put sale and institutional flow agrees, that's a stronger setup than either signal alone. Use flow to confirm, not to generate ideas from scratch.
Why do you filter out earnings plays?
Institutional flow before earnings is massive and contradictory. You'll see heavy call buying and heavy put buying on the same stock in the same week. The flow is pricing in uncertainty, not direction. After tracking hundreds of pre-earnings conviction signals, win rates were basically 50/50 regardless of score. Filtering them out immediately improved results.