Trading Through Triple Witching

The setup: a quarterly expiration day is a different microstructure

On Friday, September 18, 2026, the market reaches triple witching, the quarterly simultaneous expiration of stock-index options, stock-index futures, and single-stock futures that falls on the third Friday of March, June, September, and December. It is not a normal session for microstructure. Expiration concentrates an unusual amount of volume as positions are rolled, closed, and settled, it pulls prices toward heavily-traded strikes through dealer hedging, and it loads the final hour, the witching hour, with settlement-related activity. For an automated system, the day is a distinct regime, and settings tuned for an ordinary session do not account for what makes it unusual.

The reason it matters is that a meaningful share of the movement on an expiration day is mechanical rather than informational. When large open interest sits at a strike, dealers hedging their positions can pull the underlying toward that level, an effect known as pinning, and prices can gravitate to round-number strikes into the close for reasons that have nothing to do with a directional view. The volatility regime on a witching day is shaped by that mechanical activity, and a system that reads every move as a signal is exposed to activity that is not one.

Expiration-day moves can be mechanical, not directional

The central risk for an automated system on a witching day is mistaking a mechanical move for a real one. A drift toward a strike into the close, a spike as a large position unwinds, a burst of volume in the final hour, these can look like the start of a trend to a signal-following system, when they are the market settling positions rather than repricing on information. The system cannot tell the difference in the moment, and a move driven by expiration mechanics can reverse the instant the mechanical pressure clears. This is why an expiration day is treated as a microstructure regime rather than a directional one: the tradeable fact is that unusual moves are more likely to be mechanical, and their direction is not reliably informative.

Schedule control around the witching hour

Because the concentrated, mechanical activity loads into the final hour, schedule control that restricts new entries around the witching hour keeps the automation from initiating fresh exposure into the least-informational, highest-volume part of the day. Excluding the closing hour, or the expiration session entirely, is the most direct way to express a decision to sit out the settlement rather than trade it. The honest limit is the familiar one: sitting out the witching hour forgoes the real moves inside it as well as the mechanical ones, and expiration-day volume can present genuine opportunity, not only noise. It is a decision about whether to trade the settlement window, not a claim that the window is unsafe.

Contract-price filters when expiration pricing gets unusual

On expiration day, short-dated option pricing behaves at its most extreme: time value collapses rapidly as expiry approaches, and gamma near the strike makes premiums move sharply on small underlying moves. A max_entry_price ceiling and a min_entry_price floor keep the automation from entering contracts whose pricing has become unusual near expiry, whether that is expensive gamma-inflated premium or a contract about to decay to nothing. The honest limit: expiration pricing is unusual because the mechanics are unusual, not because it is wrong, so a filter that avoids the extremes also filters out setups where the pricing was justified. The filter bounds what the system pays into an unusual pricing environment; it does not distinguish a justified extreme from an unjustified one.

Concurrency, pinning, and sizing into concentrated volume

Pinning concentrates activity at specific strikes and the names with heavy open interest there, so signals can cluster around the same levels for mechanical reasons, and a max_concurrent_positions cap limits how much the system stacks into that clustered, mechanically-driven activity. Because the day's range is widened by the settlement flow, reducing contract count or the max_capital_per_trade ceiling caps the exposure the wider range would amplify. And because the witching hour concentrates volume and moves fast, low-latency self-hosted execution narrows the slippage on a fill, though it does nothing to tell a mechanical move from a real one. The honest limit across all three: caps and smaller size bound how much mechanical activity the system takes on and how much a single move can cost, not whether the mechanics whipsaw the signal.

What no setting resolves

No combination of schedule control, price filters, concurrency, and sizing predicts whether an expiration-day move is mechanical or informational, where a pin lands, or how the settlement resolves, and none of it guarantees a green day or removes downside. These settings bound how much of an unusual microstructure the automation is exposed to and how much a single mechanical move can cost; they do not remove the mechanics or make them readable. The control you have is over whether and how much the system trades into the settlement, not over what the settlement does.

StaxInvesting is self-hosted automation software, not a signal service and not financial advice. Past performance does not predict future results. Every trade runs in your own connected brokerage account under settings you configure: StaxInvesting never accesses member funds, credentials, or accounts, and never places trades on your behalf. No setting, size, or strategy guarantees a profitable session.