Guide

The collar option strategy, measured on real quotes

A collar is 100 shares, a put bought below the price and a call sold above it. Replayed on real bid and ask since 2022, it cut the worst 37-day trade on SPY from -15.0% to -10.4% and on NVDA from -34.2% to -13.5%, and kept 73% of SPY's average gain and 44% of NVDA's.

Short answers

What is a collar in options?

You own 100 shares, buy a put below the price and sell a call above it with the same expiry. The put sets a floor under the shares until expiry, the call caps the gain at its strike, and the premium from the call pays for some or all of the put. It is a protected position with a ceiling, for a fixed window.

Is a zero cost collar really free?

The cash at entry can net to zero, but the cost moves to the upside you give away. On SPY a call about 25 delta paid for a put 10% below with a little left over, yet the average 37-day trade kept 73% of what simply holding earned, and the shares were called away about a third of the time.

Does a collar protect against a bear market?

It protects against a fast drop inside the window. In 2022 the index fell in steps that rarely crossed a put 10% below within one 37-day window, so the collar on SPY averaged -1.44% a trade against -1.32% for the shares. In the fast drop from late February 2025 it lost 10.4% where SPY lost 15.0%.

What a collar is

A collar has three parts: 100 shares you already own, a put you buy below the price, and a call you sell above it, both with the same expiry. If the stock falls through the put strike, the put makes up the rest of the fall. If the stock rises through the call strike, the shares are called away at that price. Between the two strikes you hold the shares as usual.

The call pays for the put. With a call about 25 delta and a put 10% below the price, the call's premium covered the put on SPY, QQQ and AAPL, leaving a small credit of about 0.4% of the price. On NVDA it came out close to even, and on TSLA it cost about 0.4%. That is what people mean by a zero cost collar: the cash nets out, and the price is paid in upside instead.

How we tested it

We replayed 110 entry days per ticker from April 2022 to September 2026 on SPY, QQQ, AAPL, NVDA and TSLA, the expiry nearest 37 days out, held to expiry. The put was bought at the end-of-day ask and the call sold at the bid, with fees on every leg. On the same days we also measured holding the shares alone, a covered call (shares plus the same call), and a protective put (shares plus the same put). Dividends are left out of every column, so the comparison is on price alone.

Results are a share of the stock's value at entry, per trade. The entries overlap, about one every ten trading days against a 37-day hold, so the averages describe the window well but are not 110 independent tries.

What it saved, and what it cost

The worst 37-day result fell sharply. SPY: -15.0% holding, -10.4% collared. QQQ: -17.9% and -11.1%. AAPL: -22.1% and -11.5%. NVDA: -34.2% and -13.5%. TSLA: -32.6% and -13.0%. A collar with the put 5% below cut the worst further, to between -6.3% (SPY) and -9.6% (TSLA), at a higher net cost.

The price was the average trade. SPY averaged +1.31% a trade holding and +0.96% collared, QQQ +1.96% and +1.21%, AAPL +1.79% and +0.94%, NVDA +6.11% and +2.70%, TSLA +2.33% and +0.59%. The best single trade shows why: QQQ's best was +17.0% for the shares and +8.9% collared, NVDA's +50.3% and +16.5%. The shares were called away in about a third of entries on every ticker except TSLA (25%).

The more a stock moves, the more a collar gives away in good months and saves in bad ones, and in this window the good months were bigger. Over these four and a half years the collar kept 73% of the average gain on SPY, 62% on QQQ, 53% on AAPL, 44% on NVDA and 25% on TSLA.

Fast drops, not slow ones

A collar's put pays only when the stock is below the strike at expiry. In 2022 the index fell most of the year, but in steps, so a put 10% below rarely finished in the money inside one 37-day window. The collar on SPY averaged -1.44% a trade that year against -1.32% for the shares, and on QQQ -2.22% against -2.42%. TSLA, which fell fast, is the exception: -4.12% collared against -8.55%.

The fast drop is where it earned its keep. A collar opened on SPY at the end of February 2025 lost 10.4% where the shares lost 15.0%. On AAPL from the same day it lost 11.0% where the shares lost 22.1%.

Collar, covered call or protective put

On the same days, a covered call alone kept more of the gain than the collar (SPY +1.17% a trade, NVDA +3.90%) but protected almost nothing: its worst trades were within about two points of the shares'. A protective put alone kept more upside than the collar (NVDA +4.91% a trade) and cut the worst about as well, but cost 0.3% of the price each time on SPY and 2.8% on TSLA.

A collar sits between them. It is a reasonable way to hold a position you cannot or will not sell for a known window, such as a concentrated stock, a tax lot or a holding through an event you are worried about. As a standing way to hold an index fund every month, it gave up more than it saved in this window.

Before you place one

Pick the window first, then the strikes. A put closer to the price protects more and costs more, and in our replay the 5% put paid off about three times as often as the 10% put on the indexes. A call closer to the price pays more and gets called away more often, which in a taxable account can mean a sale you did not plan. Check the call strike against what you would be content to sell at.

This window covers one bear year and one fast drop, and holds every trade to expiry. Real collars are often rolled or closed early, which changes the numbers but not the shape: a floor, a ceiling, and a cost paid in upside.

Historical replays are not a forecast. Options can lose more than the premium collected. This is education, not a recommendation to trade anything.

Test your own rule on history, free