Net premium, and the limits of all this
The first three chapters explained what each measure sees. This one explains what none of them see. It is the most important chapter in the sequence, because the fastest way to lose money with good data is to believe it answers a question it was never capable of answering.
What net premium measures
Every option trade prints somewhere between the bid and the ask. A trade at the ask means the buyer crossed the spread — they wanted it enough to pay up. A trade at the bid means the seller did. Net premium counts dollars at the ask as positive, dollars at the bid as negative, and ignores trades at the midpoint where neither side was clearly the aggressor.
Accumulated through a session, it becomes a running measure of which side has been more willing to pay for immediacy. Calls and puts are tracked separately, because "calls being bought" and "puts being sold" are different statements that a single blended number would hide.
The reading that matters most is divergence. Price and premium rising together tells you little you could not see on the chart. Premium climbing while price stalls says buyers keep paying up without getting paid for it — which sometimes precedes a move and sometimes precedes nothing at all.
What it cannot see
This list applies to every flow product sold anywhere, including the expensive ones. None of them can see through these, and any tool implying otherwise is overselling.
- Opening versus closing
- A trader opening a new long call and a trader closing a short call both buy at the ask. The tape records an identical print. Half the time the "bullish buying" on your screen is someone exiting a position they no longer want.
- Hedges
- A fund buying puts against a large equity holding shows up as aggressive bearish flow. Their actual position may be net long and getting longer. Protective buying and directional betting are indistinguishable on the tape.
- Spread legs
- Multi-leg strategies print as separate trades. A call spread appears as a call bought at the ask and another call sold at the bid — potentially in different rows, scored separately, pointing in opposite directions. The strategy is invisible; only its fragments are recorded.
- Who is on the other side
- The tape does not label participants. "Institutional flow" is an inference from size, not a fact from the data. Large trades come from funds, market makers laying off risk, and occasionally individuals with more conviction than sense.
None of this makes the data useless. It makes it evidence rather than instruction. A large aggressive sweep into a strike just below the call wall is a real event worth understanding — it is simply not a message from someone who knows what happens next.
Base rates beat examples
The most common failure in flow analysis is not misreading data. It is learning from a filtered sample. Every platform, newsletter and social account in this industry shows you the alerts that worked, because those are the ones worth posting. The distribution they were drawn from is never shown.
The question that matters is never "did this signal work". It is how often does this type of signal work, across every instance, including the ones nobody screenshotted. That number is usually far less impressive than the highlight reel — and it is the only one you can size a position against.
This is why FlowRadar retains ninety days of end-of-day snapshots. Not so you can find the one alert that preceded a large move, but so you can ask what happened after all alerts of a given kind. A method with a modest, honest hit rate that you actually understand is worth more than a spectacular example you cannot reproduce.
Reading the whole picture
When the panels agree, you have a coherent story. When they disagree, you have learned that conditions are ambiguous — which is itself worth knowing, and usually a reason to do nothing.
- Regime first. Above or below the flip? That decides whether moves get absorbed or extended, and it changes appropriate position size before anything else does.
- Then structure. Where are the walls, and are they durable or built from options expiring today?
- Then the tape. Is anything unusual happening — and is it near the levels that matter, or off in a corner of the surface that changes nothing?
- Then premium. Is aggression confirming what the structure implies, or fighting it?
- Then the calendar. Does an event inside your horizon reset all of the above?
The strongest configurations are the ones where independent measures point the same way: a durable wall, unusual activity concentrated near it, aggression on the matching side, and no catalyst due to scramble the positioning. That coincidence is rare, which is the point. Most days the honest answer is that nothing unusual is happening.
A closing caution
Everything in FlowRadar is descriptive. It tells you what has been traded and what that implies about how dealers must hedge. It does not know what will happen next, and neither does anyone selling you a version of it that claims otherwise.
Markets also adapt. Dealer positioning is watched by more participants every year, and behaviour that was reliable when few people tracked it becomes less so as more do. Treat any pattern here as a working hypothesis with a shelf life, not a rule.
FlowRadar is an educational and informational tool. Nothing in these chapters is investment advice, and none of it accounts for your circumstances, risk tolerance or objectives. Options carry a genuine risk of losing the entire amount committed, and strategies that look asymmetric on paper frequently are not once costs and timing are included.