Portfolio Margin, Explained: The Math Behind Your Buying Power — and Why It Rises in a Crash

Most option-sellers meet portfolio margin as a pleasant surprise: they cross some account threshold, switch it on, and suddenly the same positions tie up a fraction of the cash they used to. Buying power balloons. It feels like the broker just handed you a better deal.

It did, and it didn't. Portfolio margin is more efficient, but the efficiency comes from leverage you can't see, and the number moves against you at precisely the worst moment. Here's the math underneath it, and why simulating drawdowns is the only honest way to know where you stand.

Two kinds of margin

Standard Reg-T margin runs on fixed formulas. A cash-secured put ties up the full strike × 100 in cash. A naked short put ties up roughly 20% of the underlying. Simple, conservative, and blind to the rest of your book — each position is charged on its own, as if nothing else in the account existed.

Portfolio margin throws the formulas out. Offered to larger accounts — typically US$100k+ at brokers like Interactive Brokers — it's risk-based. Instead of asking "what's the rule for this position," it asks "what's the most this whole portfolio could lose in a bad-but-plausible move," and charges you that.

That one change is the whole story. Everything else follows from it.

The core idea: margin is your worst-case loss

A portfolio-margin engine (TIMS or SPAN-style, depending on the venue) does something close to a simulation. It shocks each underlying across a range of moves — commonly about ±15% for broad-based names, wider for volatile or leveraged ones — usually with a volatility bump layered on top. It reprices every position at each of those shocked points, and takes the biggest loss across all the scenarios as your requirement.

So the margin number isn't something you can eyeball from a formula. It's the output of a model. Two accounts holding the "same" short puts can carry very different requirements depending on what else sits alongside them.

That's the part worth internalising: your buying power is a model's estimate of how badly you could be hurt.

Netting is why it's efficient

Here's where the capital efficiency comes from. In the model, positions offset each other. Long stock cushions a short call. A long put caps the loss on a short put below it. When the engine reprices your whole book at −15%, the pieces that gain partly cancel the pieces that lose — so the worst-case number is much smaller than the sum of each leg charged alone.

Under Reg-T, a covered call and a naked put are billed separately. Under portfolio margin, the engine sees how they interact and charges the net. A genuinely hedged book can need a small fraction of the Reg-T requirement.

That's the appeal, and it's real. But netting cuts both ways: it rewards hedging, and it quietly tempts you to hold more risk because the headline number looks so tame.

The counter-intuitive part: in a crash, the requirement rises

This is the bit that catches people, so sit with it.

Your margin requirement is your modelled worst-case loss. In a crash, the worst case gets worse — so the requirement goes up, exactly when you can least afford it.

Three things happen at once as the market falls:

• Implied volatility spikes. Think VIX doubling or tripling. The model's volatility bump is now applied on top of an already-elevated base, widening every scenario. This is portfolio vega being repriced across your whole book at once — the exposure that diversifies least, because implied vol on unrelated names rises together.
• Short puts get more expensive to close. The premium you sold for pennies now costs real money to buy back, so the mark-to-market loss on your position grows.
• Strikes move toward and into the money. Puts that were comfortably out of the money are now near or below spot, where their risk is highest.

So the modelled worst-case loss climbs. Meanwhile your net-liquidation value — the actual equity in your account — is falling, because the positions are losing money in real time.

You're squeezed from both sides. Net-liq drops while the requirement rises, and the two lines move toward each other. When net-liq falls below the maintenance requirement, the broker force-liquidates — often at the worst prices of the day, in the names that have moved most. That double-squeeze is the margin call. It isn't a separate event that happens to unlucky people; it's the mechanism, and it's baked into how the math works.

Leverage hides in plain sight

Because portfolio margin only ties up the modelled worst-case — often around 10–15% of notional — a modest premium against it can look like an enormous "return on margin." (This is worth reading next to yield on capital: the denominator you choose changes the story completely.)

But that low requirement is leverage. The tail risk still sits on the whole account, not on the 15%. If the position turns out worse than the model's shock — a gap of 25% when the scenario tested 15% — the loss lands on your full equity, and the margin you posted was never a cap on how much you could lose.

Leveraged ETFs make this vivid. A 3x fund moves roughly three times the index, so the engine margins it far harder — a 15% index shock is a ~45% shock to the fund. Selling options against leveraged products stacks leverage on leverage, and the requirement (rightly) reflects that.

Beta matters in a market-wide drop

One more wrinkle before we simulate. A 20% S&P 500 drawdown doesn't hit every holding by 20%. A high-beta name falls more; a low-beta staple falls less. If your book is tilted toward high-beta stocks, a market-wide drop hurts you more than the index number suggests — and toward defensives, less. The shares you hold outright are part of that tail, not a hedge against it: a concentrated stock position is its own short put struck at zero, and it belongs in the same stress test as everything you've sold.

A useful simulation scales each holding by its own sensitivity — how much that name tends to move relative to the market — rather than assuming everything moves together. A leveraged or high-beta name gets shocked harder than a defensive one. Otherwise you're stress-testing a portfolio you don't actually own.

Why simulate drawdowns

Put all of this together and you see the problem. The margin number is opaque — a model output, not a formula. The crash dynamics are non-linear — IV, moneyness and net-liq all move at once, and they compound. So the questions that actually matter are genuinely hard to answer by staring at a single figure:

• How much more can I safely sell?
• Would I survive a 25% drop?
• How close am I to the edge right now?

Running your book through a ladder of drawdowns — say 5% up to 30% — answers them directly. It shows where the headroom runs out, how many more contracts you could add before a crash would liquidate you, and how the requirement climbs as the shock deepens. You watch net-liq and the maintenance requirement approach each other and see exactly where they cross.

And it forces one honest distinction the single number hides: "survive" means no margin call, not no loss. You can survive a 30% drawdown and still be down a great deal of money. Surviving means you kept control of your positions instead of having them sold for you at the bottom. That's a lower bar than "I'm fine," and it's the right bar to plan around.

This all sits downstream of the same Greeks that drive any short-premium book: negative vega is why the IV spike stings, negative gamma is why the loss accelerates as the market falls. Portfolio margin just prices those risks for your whole account at once.

Where this fits

If you're running the wheel at any size, portfolio margin is likely how your buying power gets set, whether or not you've thought about the math behind it. Understanding it isn't optional risk-nerdery — it's the difference between choosing your leverage and discovering it in a drawdown.

At levelbox.ai we built a risk simulator to make this concrete. It estimates your portfolio-margin requirement under a ladder of S&P 500 drawdowns, lets you add positions from your watchlist and see the impact live, and shows the honest cash-secured yield next to the margin actually used — so the leverage stops being invisible. It shocks each holding by its own sensitivity to the market rather than applying one flat move to every name, so the simulation reflects the book you actually hold — a high-beta or leveraged position gets hit harder than a defensive one. It also flags when a leveraged book stops surviving the deep drawdowns.

One caveat we'd rather state plainly than bury: it's a model estimate of the broker's number, meant to help you reason about risk, not to match their engine to the dollar and not a directive to trade. The point isn't to tell you what to do. It's to make the worst case visible before the market makes it visible for you.