Earnings Options Course: From Implied Moves to a Research Screen

An earnings options course should teach one connected research process. It should not begin with a favorite structure or a rule such as "sell volatility before the announcement." The starting point is more basic: an option price contains a forecast, the earnings release resolves uncertainty, and the gap between forecast and outcome can be measured. Once that mechanism is clear, structures become implementations of a view rather than recipes looking for a reason.

That distinction matters because an earnings trade has several separate questions hiding inside it. What move did the market imply? What move occurred? Does this company have a repeatable history of overpricing or underpricing the event? Is current volatility rich relative to its own recent range? Is the option chain liquid enough to express the idea? A complete curriculum answers them in order and preserves the information that was available at each decision time.

Start with the forecast inside the straddle

Near the money options are the cleanest place to begin. Add the call and put premium around the stock price and you have a practical approximation of the move the market is pricing through expiration. The result is not a promise and it is not a confidence interval with universal coverage. It is a traded forecast shaped by expected movement, supply and demand, time to expiration, rates, dividends, skew, and the exact strikes selected.

A useful course makes the conventions explicit. Choose the expiration that contains the announcement, record the stock price and option quotes before the event, define whether the move is measured close to close or close to next open, and never mix those definitions later. The point is not to find the one true implied move formula. The point is to define a forecast that could have been observed and repeated without hindsight.

Build event rows before building a strategy

The atomic unit is one company earnings event. Each row needs an announcement date, the last observable pre-event forecast, the post-event realized move, and identifiers that survive ticker changes. It also needs a flag for whether the announcement occurred before the open or after the close. Without that timing field, the code can pair a forecast with the wrong return window and produce a precise answer to the wrong question.

Next compute implied minus realized in percentage points. A positive value means the forecast exceeded the outcome under your convention. A negative value means the stock moved more than priced. Keep both the signed gap and the absolute values. The signed gap measures direction of mispricing, while the raw forecast and outcome reveal whether one extreme event is dominating the average.

events["gap_pp"] = events["implied_move_pct"] - events["realized_move_pct"]
summary = events.groupby("ticker").agg(
    events=("gap_pp", "size"),
    hit_rate=("gap_pp", lambda s: (s > 0).mean()),
    avg_gap_pp=("gap_pp", "mean"),
    worst_underprice_pp=("gap_pp", "min"),
)

Make the history point in time

The easiest mistake is to rank a 2022 event using a company record calculated through 2026. That imports future quarters into the signal. For each event, historical hit rate and average gap must use only earlier events for that name. In pandas this is an expanding calculation followed by a one-row shift. The shift is the difference between a feature available before the event and a label that quietly contains the answer.

Sample size belongs beside every average. Three quiet quarters can produce a perfect hit rate and no evidence. A course should force the student to choose a minimum history before looking at results, then show sensitivity across several minimums. It should also display the worst miss. Earnings premium strategies often accumulate small favorable gaps and then surrender them to one large surprise, so an average without its tail is not a usable summary.

Join slow evidence to current state

A per-name record changes only once each quarter. Current implied volatility changes every day. Joining them creates a richer question: is a name that has historically overpriced earnings also expensive now relative to itself? IV rank or IV percentile can supply that state variable, provided the lookback and timestamp are defined. This is where the research moves from a retrospective table to a weekly candidate screen.

The screen should not mechanically turn every top row into a short straddle. It should remove thin chains, exclude events with uncertain timing, inspect skew, and compare the potential premium to the defined loss of the chosen structure. It should also recognize that buying premium can be the coherent expression when a name has repeatedly underpriced the event. The evidence determines the question; the structure controls the exposure.

• Data check. Confirm the announcement timestamp, expiration, quotes, and corporate-action adjustments.
• Evidence check. Require a predeclared event count and inspect the distribution, not just the mean.
• State check. Compare current volatility with the same name and a consistent trailing window.
• Trade check. Model spreads, commissions, assignment, and the defined tail of the structure.
• Regime check. Split results across broad volatility regimes and reporting seasons.

Finish with a reproducible weekly screen

The final project should run from a clean input to a dated output without manual editing. Pull the coming earnings calendar, attach each name's point-in-time event record, join current volatility context, apply declared gates, and save the candidates plus the rejected rows. Saving rejections matters because it shows whether changing data coverage or a filter caused the next run to differ.

This curriculum is deliberately broader than the existing pre-earnings screen article. That article explains one finished screen. This course guide explains the sequence a learner should master before that screen is trustworthy, from the implied forecast through event construction, point-in-time evidence, current state, and risk-aware expression.

The pre-earnings screen lesson is the natural capstone because it makes you join the slow and fast variables into a real research surface. Browse the rest of the quant trading curriculum guides if you are comparing which adjacent skills should come next. The goal is not a list of earnings trades. It is a process that can reject a tempting setup for an explicit reason.