Most traders approach earnings like gamblers. They focus on whether the company will beat or miss, whether the move will be big or small, and whether they can sell some juicy premium before the print. That’s not how we approach it in our hedge fund.

The real edge in earnings is identifying where the option surface is distorted, deciding whether the market is overpricing or underpricing the event, and then choosing the structure that matches that exact setup. Sometimes that means selling front-week event volatility. Sometimes it means buying deferred convexity. Sometimes it means doing nothing at all.

The real edge: earnings distort the term structure

Around earnings, the expiration that contains the event often gets pumped full of uncertainty premium. Front-week IV rises sharply because it carries the full earnings jump. The next weekly expiration usually rises much less. Then, once earnings are out, the front-week volatility collapses fast. That front-vs-back distortion is the cleanest and most repeatable feature of earnings options. So the smart question is not: should I sell premium before earnings?, but: which part of the surface is mispriced, and what structure isolates that mispricing best?.

A lot of traders still treat earnings as a simple premium-selling game. That works, until it doesn’t. Oracle in September 2025 is the perfect reminder. Tom Sosnoff sold a large naked strangle position into that earnings event, exactly the kind of trade that looks smart when the market is overpricing noise. But after the September 9 report, the stock ripped 35.9% and closed at $327.76. That was not a normal breach of expected move, that was a full-scale reminder that some earnings events are not overpriced at all.

And once you understand that, you stop asking “what premium can I sell here?”. Before choosing a structure, ask three things:

• Is the implied event move actually rich relative to the stock’s historical earnings moves?
• Does the short-dated surface look smooth, or is it warning about jump tails?
• Is this a one-day event, or a name that tends to trend after earnings? Those three questions determine the trade.

Not every earnings event deserves the same structure

The research strongly supports three filters before you choose a trade.

First, check event richness; the implied event move should be compared with the stock’s own historical realized earnings moves. This matters because earnings uncertainty is not always overpriced. Gao, Xing, and Zhang show that at-the-money straddles opened three days before earnings earned a significant average return of about 3.34% in their sample, which means some earnings events were actually underpriced, not overpriced.

Second, check surface geometry. Alexiou, Goyal, Kostakis, and Rompolis show that short-dated IV curves often become concave before earnings, and that these concave surfaces are associated with larger absolute earnings-day moves and higher realized volatility afterward. They also show worse returns for several delta-neutral option structures in these states. In plain English: when the wings are screaming, short-vol trades become much more dangerous.

Third, check post-event drift. PEAD is real. Some names do not just gap and stop; they keep moving after the print. That is a different edge from the event-volatility trade, and it deserves a different structure.

The default institutional trade: the core term-structure diagonal

When the front event premium is rich, the surface is smooth, and there is no strong continuation signal, the best default earnings trade is still a calendar or diagonal.

This is also the direction ORATS’ large earnings backtest points to: across 5,217 earnings announcements and 20,868 trades, the long calendar was the best-performing strategy in that test, while long straddles and short calendars were weaker. That is highly relevant to my actual execution. My default implementation for a bullish version is:

• Short leg: earnings-week call, around 20-30 delta
• Long leg: next-week call, around 40-60 delta
• Strike relationship: long strike usually 1-3 strikes below the short strike

For a bearish version, mirror the same idea on the put side.

The short leg monetizes the part of the surface most distorted by earnings. The long leg owns deferred vega and enough intrinsic sensitivity that the position does not behave like a disguised naked short option if the stock moves hard.

The advanced sleeve: TSCA / diagonal call ratio / backspread

When the front-week premium is extremely rich, skew is favorable, and the name has real post-event continuation potential, we can upgrade from a plain diagonal to what I would call a term-structure convexity trade.

The idea is to let front-end IV crush help finance a deferred convexity position. A practical bullish version looks like this:

• Short legs: front-week calls in the expiry containing earnings
• Long legs: more back-week calls than front-week shorts

A typical template is to short front calls around 0.10-0.20 delta, ideally outside the market-maker expected move, long back calls one week later or slightly further out, with an anchor that is ATM, slightly ITM, or moderately OTM depending on whether you want defense or convexity. Ratio chosen so that the post-gap “valley” is materially reduced or eliminated.

That last part is critical. The purpose of the structure is not to blindly sell calls and buy lottery tickets. It is to create a payoff where the front-end event IV pays you, the back-week long keeps value after the print, and a normal earnings move does not dump you into a max-loss hole near the short strike.

That no-valley design is not something the literature hands you. It is payoff engineering. The research validates the existence of the term-structure distortion and the jump-risk filters; the exact ratio and strike geometry are still a proprietary implementation problem.

Concavity is the guardrail most traders ignore

This is probably the most valuable research-backed filter in the whole framework. If the short-dated IV curve becomes strongly concave before earnings, the market is telling you that tails matter. Alexiou et al. show that these names go on to exhibit larger absolute returns on earnings day and higher realized volatility afterward, and that several popular delta-neutral option structures perform materially worse in these states.

That gives you a very practical rule: do not run aggressive short-volatility earnings trades into concave surfaces.

A simple diagnostic is to compare the average IV of the 10-delta call and 10-delta put with ATM IV. If that ratio is abnormally stretched, treat the name as a jump-risk candidate, not a casual IV-crush sale.