Back to Trade Craft
Flow & Positioning
Understanding Options Flow: From Noise to Signal
The Problem...
You've heard that institutions leave footprints in the options market. So you pull up an options flow scanner. Thousands of trades scroll past every minute, each with a strike, expiry, premium, and size. A 500-lot call trade at 22,600. A 200-lot put sweep at 22,300. A block print at a strike nobody was watching.
Which of these matter? Which are hedges, which are directional bets, and which are market makers adjusting inventory? The data is there. The interpretation isn't.
Most traders either ignore flow entirely (because it looks like noise) or follow every large trade (because it looks like conviction). Both approaches lose money. Flow data without a filtering framework is a firehose pointed at your decision-making process.
KEY INSIGHT
Options flow is a record of every options trade executed, not a signal feed. The signal lives in four specific patterns: sweep orders (aggressive, time-sensitive execution), block trades (large single prints negotiated off-screen), repeat activity at the same strike over time, and unusual volume relative to existing open interest. Filtering these patterns from the surrounding noise is the foundational skill of flow analysis.
What Changes With Synthesis
The traditional approach to options flow is a spreadsheet exercise. Open a scanner, sort by size or premium, and look for big numbers. A $2 million call sweep looks important. So you go long. Then price drops because that sweep was the closing leg of a spread, not a directional bet.
Sorting by size captures attention. It does not capture intent.
What changes with multi-dimensional synthesis is context. A large call sweep at 22,600 means something different when gamma is positive at that strike (market makers will sell the rally) than when gamma is negative (market makers will buy into the rally). The same trade means something different when CVD is confirming aggressive buying versus when CVD shows distribution underneath. And it means something different on Monday (five days to expiry, time for the thesis to develop) than on Thursday morning (hours to expiry, theta is destroying the position).
Flow in isolation is ambiguous. Flow synthesized with price structure, momentum, gamma regime, and time context becomes directional intelligence. That synthesis is what separates traders who read flow from traders who react to it.
The Method
The options flow method
Step 1: Know what you're reading (the four signal types)
Every executed options trade appears on the flow tape. Most are routine. The signal sits in four patterns. Sweep orders hit multiple exchanges simultaneously to fill a large order as fast as possible, signalling urgency that implies time-sensitive conviction. Block trades are large single-print transactions negotiated privately, indicating institutional size and intent. Repeat activity at the same strike over a session or multiple sessions shows accumulated conviction that a single trade cannot. Unusual volume relative to open interest identifies strikes where fresh positioning is building: if a strike normally trades 2,000 contracts and prints 15,000 before noon, the ratio itself is the flag.
Step 2: Filter noise with three questions
Not every sweep is directional. Before acting on any flow signal, run it through three questions. Is this opening or closing? Compare trade volume to existing OI at that strike: if OI rises the next session, those were new positions representing fresh conviction; if OI drops, someone was leaving. Is this a hedge or a directional position? A $5 million put purchase alongside a $500 million long equity portfolio is protection, not bearishness; check whether the flow is paired with offsetting activity. Is the execution aggressive or passive? Sweeps that pay the offer signal urgency and time sensitivity; limit orders that sit and wait signal strategic patience.
Step 3: Understand what flow cannot tell you
Flow reveals positioning, not outcomes. A fund manager who spends $10 million on call sweeps can be wrong. The signal is that someone with significant capital believes strongly enough to commit real money, and that belief is worth noting, but it is not worth following blindly. Flow also compresses near expiry and around scheduled events as hedging adjustments and gamma-related trades spike, which means your threshold for what counts as a genuine signal must rise with the context.
Step 4: Integrate flow with independent dimensions
The value of flow multiplies when checked against dimensions measuring different things. Flow + CVD: if sweeps are buying calls while CVD confirms aggressive buying in the underlying, two independent sources agree; if CVD shows sellers dominating, the sweeps may be hedges. Flow + gamma regime: call buying in positive gamma faces mechanical resistance as market makers sell into the rally, while the same buying in negative gamma gets amplification. Flow + price structure and momentum: institutional put accumulation while price rises and velocity decelerates is a high-value divergence; the same accumulation while price is already falling is confirmation, not a leading signal. One confirming dimension elevates flow from interesting to actionable. Two make it high conviction.
Step 5: Size your response to signal quality
Not all flow signals deserve the same response. A single sweep at an unfamiliar strike with no confirming dimensions warrants attention, not a position. Repeat activity at the same strike over two sessions, confirmed by CVD and supported by the gamma regime, warrants conviction sizing. Match position size to the number of confirming dimensions, applying the confluence scoring framework directly: flow confirmation is one of the five independent categories.
In Practice
It's Thursday morning, 30 minutes before NIFTY's weekly expiry. You're watching flow for directional clues.
The tape shows heavy activity across dozens of strikes. Most of it is noise: market makers rolling positions, retail closing expiring contracts, delta adjustments as gamma spikes on the final day. This is the compression problem from Step 3. On expiry day, the baseline level of large prints is elevated, so your filters need to be tighter than usual.
You filter. Sweeps only, opening trades, premium above â¹50 lakh. The noise falls away. Four signals remain.
At the 22,800 put strike, you see 1,200 lots swept in two tranches within three minutes. This strike had 8,000 OI yesterday. By mid-morning it has 14,000. New positions, aggressive execution, concentrated at a single strike. That's three of the four signal types firing at one level: sweep urgency, unusual volume-to-OI ratio (nearly 75% of existing OI added in one session), and repeat activity across two tranches rather than a single print. You run the three filtering questions. The OI jump confirms these are opening, not closing. There's no corresponding call buying or futures activity to suggest a spread or hedge. And the execution was aggressive, paying through the offer on both tranches. All three filters point the same direction.
Now you check the independent dimensions from Step 4. CVD on the 15-minute chart shows aggressive selling increasing over the last 90 minutes, even though price has been flat. This is the divergence pattern: surface calm, subsurface pressure building. Velocity is low, near the 20th session percentile, which means price isn't moving much yet, but the selling underneath is steady and growing. The gamma regime is negative below 22,900, meaning if price drops toward 22,800, market maker hedging will amplify the decline rather than cushion it.
The flow says institutional money is positioning for downside at a specific strike. CVD says aggressive sellers are already in control beneath the surface. The gamma regime says the market's structure will accelerate a move lower if it starts. Three independent dimensions pointing the same direction.
Price is still sitting at 22,870. The chart looks neutral. A single-dimensional chart reader sees nothing. The flow reader, with dimensional context, sees a building thesis with conviction-level confirmation across three categories. Whether you act on it depends on your own trade plan and risk rules, but the information asymmetry between "what the chart shows" and "what the flow plus context shows" is the entire value of this method.
Common Mistakes
Treating every large trade as a directional signal.
Institutions hedge constantly. A $10 million put purchase might protect a $500 million equity portfolio, which signals continued bullish conviction, not bearishness. The size of the trade is not the signal. The context around the trade is. Always ask whether the flow could be protective rather than speculative before assigning directional meaning.
Acting on flow without checking other dimensions.
A sweep order at a single strike, with no CVD confirmation, no momentum context, and no gamma analysis, is an observation, not a trade setup. Flow alone has a poor signal-to-noise ratio. Flow plus one independent confirming dimension becomes meaningfully more reliable. The discipline is waiting for confirmation rather than reacting to the first large print.
Forgetting that flow compresses near expiry and events.
On expiry day, hedging adjustments and gamma-driven trades create a surge of large prints that look institutional but are mechanical. Before scheduled events like RBI policy or budget, protective positioning inflates volume across the chain. The bar for "unusual" must adjust to the context. What qualifies as a signal on a quiet Tuesday may be routine noise on expiry Thursday.
FAQs
What is options flow in trading?
How do I tell if a large options trade is a hedge or a directional bet?
Is options flow more reliable for indices or individual stocks?
Can I use options flow as a standalone trading signal?
How far in advance does flow signal price moves?
See what institutional flow analysis looks like in real-time
Try Draconic
This guide is for informational and educational purposes only. It does not constitute financial advice. Trading involves risk. Draconic provides market intelligence; all trading decisions are your own.