Part XI · Swing trading · Chapter 11J

Post-earnings drift

Gap and go

Where we are

Chapter 11I traded into an earnings report and left before it. This chapter is its mirror: it enters after the report, once the numbers are out and implied volatility has crushed, to ride the drift that follows a strong surprise. That drift is one of the most replicated findings in finance.

What you need first

Post-earnings announcement drift, and why behavioral edges decay (Chapter 1G).
The pre-earnings swing, the implied move and the IV crush (Chapter 11I).
R-multiples, time stops and sizing from the stop (Chapter 11H).
Call debit spreads and put credit spreads (Chapter 12C), and days to expiry for swing trades (Chapter 11M).
The Real Value score for single stocks (Chapter 13B).

1 · The mirror trade

Chapter 11I rides the IV ramp and the drift before a report, then exits before the event. This chapter covers the mirror strategy: enter after the event has resolved, and capture the post-earnings announcement drift (PEAD).

It has the stronger evidence of the two. Pre-earnings drift is Tier B/C on Chapter 11B's scale, with mixed academic support and mostly practitioner evidence. Post-earnings drift is one of the most robust anomalies in finance: Tier A, replicated repeatedly across decades and markets. The Dealer Mechanics Lab cites Bernard and Thomas (1989), Bernard and Thomas (1990) and Chordia and Shivakumar (2006), among many later studies. It calls the asymmetry significant: the post-earnings framework has a stronger evidentiary foundation than the pre-earnings thesis, and arguably should be the higher-priority deployment in a swing book.

The Edge Book agrees that the drift is real and adds a warning about its future. Chapter 1G works its example: a stock reports earnings 20% above consensus, jumps maybe 12% on the day, then drifts up another 4% to 5% over the following weeks. PEAD is a behavioral edge, and behavioral edges decay when capital exploits them. Once enough sophisticated investors buy the post-earnings winners, the drift compresses. McLean and Pontiff (2016) document a decline of 26% to 58% in factor returns after publication, depending on the factor. Jensen, Kelly and Pedersen (2023) document a decline of about 47% across hundreds of factors. Chapter 11B makes the same distinction: Tier A grades the kind of source, not whether an effect has survived its own publication.

2 · The mechanism

When a company reports earnings, four things happen in sequence:

The price gaps on the immediate market reaction.
Analysts and institutions begin recalibrating their estimates and positions.
The recalibration takes weeks to months, not minutes, because institutional flow is slow.
During this window, the drift after the event, in the same direction as the initial gap, is persistent and statistically robust.

The drift is not in the first 30 minutes after the report. That time belongs to the gap itself and to partial mean reversion in the opening minutes. The drift unfolds over 5 to 60 trading days after the event, particularly for stocks with four marks:

A large earnings surprise, with the analyst expectation gap above 5%.
A revenue beat in the same direction, not just engineering of earnings per share.
Guidance raised, or lowered for a negative surprise.
Rising institutional sponsorship, seen as rising relative volume and the signatures of accumulation.

The earnings surprise is the gap between the reported earnings per share and the consensus, the average of the analysts' forecasts, as a fraction of the consensus. A company expected to earn $0.80 a share that reports $0.90 has a surprise of $(0.90 - 0.80)/0.80 = 12.5\%$ , well above the 5% bar.

Familiar ground

The drift works like the uptake of a practice-changing trial. The result is public the day it is presented, yet prescribing shifts over months, as guidelines are revised, committees meet and each clinician updates at their own pace. The trial's evidence is complete on the first day, while the change in behavior it causes arrives slowly. An earnings surprise reaches the price the same way: a jump on the day, then weeks of recalibration by institutions that move slowly.

3 · The setup

After a positive earnings event, a long PEAD trade needs seven conditions. A short PEAD trade after a negative event mirrors them.

The quality of the beat. Revenue, EPS and guidance are all positive. Beats on EPS alone, with a revenue miss or flat guidance, often fail to drift.
The gap's size and hold. The stock gaps up more than 5% on the report and opens the next session at or near the post-event level, without gapping back toward the range it held before the event.
The opening range holds. In the first 30 to 60 minutes after the open, the stock holds above its VWAP and above the prior day's close.
Volume confirms. Opening-hour volume is more than twice the 20-day average, and heavy volume is sustained through midday.
Sector and peers confirm. Peers in the same industry are also up, or at least not down, so that a sector tailwind reinforces the drift.
No immediate technical blockers. The stock is not running into the underside of its 50-day or 200-day moving average from below, which would create supply. Nor is it running into a cluster of heavy volume from the prior 6 months.
The IV crush has occurred. Front-month at-the-money IV has fallen to roughly the level of 90-day IV. The event component is out of the options, which are now reasonably priced.

The opening range is the span between the high and the low of the first 30 to 60 minutes of trading. The volume-weighted average price (VWAP) is the average price of every share traded since the open, each weighted by its size. For trades at prices $p_i$ in sizes $v_i$ ,

\text{VWAP} = \frac{\sum_i p_i\, v_i}{\sum_i v_i}

(11J.1)

Worked example: A morning's VWAP

Suppose the first trades after a report are 3,000 shares at $760.00, 1,000 at $764.00 and 2,000 at $758.00. By (11J.1) the VWAP is $(2{,}280{,}000 + 764{,}000 + 1{,}516{,}000)/6{,}000 = \$760.00$ . A stock trading at $766 is above it, so buyers since the open are, on average, in profit. That is what holding above VWAP means.

The skip discipline. If the gap is weak, with only an EPS beat, flat or lowered guidance, or the gap fading in the opening hour, the PEAD framework gives no edge. The drift persists only when the fundamental quality is real.

4 · The structure

The exit-before-earnings problem of Chapter 11I is reversed here. The event is past, IV has crushed and the option chain is cleaner. Three vehicles are preferred:

Shares, the default. Hold long shares for 5 to 30 days, targeting the drift. This is clean directional exposure with no overlay of volatility, best for accounts that do not need leverage.
A long call spread, for defined-risk leverage. Buy a call slightly out of the money or at the money, with 30 to 45 days to expiry. Sell a higher-strike call around the analysts' implied 30-day target, typically 5% to 10% above spot. The call debit spread captures the drift with leverage and a defined risk. A long call alone is acceptable but uses more capital per unit of exposure than the spread.
A put credit spread below the gap. Sell a put at the post-gap support level, often the top of the prior consolidation, which is now a floor. Buy a lower-strike put for defined risk. The spread captures the drift through theta and delta, with the IV crush already locked in. It works best when IV is still slightly elevated after the event and the gap is well supported on the chart.

Worked example: Two option expressions of one drift

Step 1: the call spread. With the stock at $768, buy the $770 call and sell the $820 call, 6.8% above spot, both with 30 to 45 days to run. At expiry the spread is worth at most the $50 between the strikes, so its profit is capped at $50 a share less the debit paid.

Step 2: the put credit spread. Suppose you sell the $740 put and buy the $720 put for a credit of $3.00. The most you can make is the $300 credit, and the most you can lose is $(20 - 3) \times 100 = \$1{,}700$ , below a breakeven of $737.

Step 3: the shape. The put spread wins a little in most outcomes and loses a lot in a few, the shape of short volatility (Chapter 10C). The Edge Book rejects naive short volatility held through a report (Chapter 10I). This spread is sold after the report, but it keeps the shape, so size it by its maximum loss.

Avoid three things in a PEAD trade:

Naked long calls in names with a high IV percentile. Their vega risk is not trivial even after the crush.
Calendars or diagonals. They are trades on the volatility surface before an event, and after the event the surface has already collapsed.
Holding through the next earnings event. That is a new binary catalyst, not part of the PEAD edge.

5 · Size, horizon and exits

Size at 1% to 1.5% of the portfolio at risk per trade, as for the other swing playbooks (Chapter 13C). Hold for 5 to 30 trading days. The tail of the drift extends to about 60 days, but the bulk of the move occurs in the first 4 to 6 weeks.

Exit on the first of four triggers: the target, the stop, a time stop of 60 days at most, or a change in the fundamentals. An analyst downgrade or a withdrawal of guidance after entry counts as such a change.

Take half off at +0.75R, and trail the remainder with a trailing stop. PEAD trades have asymmetric profit distributions: most close near +1R to +1.5R, but the tail extends to +3R to +5R when the drift runs. The rule's arithmetic shows why it suits that shape. The trade's result in R is half of 0.75R plus half of whatever the trailed half makes:

R_{\text{total}} = 0.5 \times 0.75 + 0.5\,R_{\text{rest}}

(11J.2)

If the trailed half is stopped out at the entry price, the trade still makes 0.375R. If it runs to +3R, the trade makes 1.875R. The early half banks something from the many trades that stall, and the trailed half keeps a stake in the few that run.

The figure trades the same made-up paths three ways. Compare the half-off line with the other two: it loses no more often than selling everything at +0.75R, and it keeps part of the tail that holding everything collects.

drift 0.030R a day

trailing stop 0.75R

Exit rule	Mean	Losing trades	90th percentile
all out at +0.75R	+0.28R	27%	0.75R
half off, rest trailed	+0.54R	27%	1.77R
all held to day 60	+1.45R	42%	4.80R

worst once half is off = 0.375R

The half that is sold, and the half that runs. Four hundred made-up trades, each a random walk in units of R, with daily moves of about 0.25R and a drift drawn for each trade: most small, a few large. Each trade is stopped at −1R and taken three ways: all out at +0.75R; half out at +0.75R, with the rest behind a stop that trails the highest price by the distance set; and all held to day 60. The lines sort each rule's results from worst to best, with stops and targets filled at their levels. Set the drift to zero and all three means fall to about zero: an exit rule reshapes the results but creates no edge.

6 · A worked trade in NVDA

The Dealer Mechanics Lab gives this example as illustrative.

Worked example: NVDA after a strong report

Step 1: the report. NVDA reports after hours on a Wednesday. Revenue beats by 8%, EPS by 12%, and guidance is raised by 5%. The stock gaps from $720 to $760 in after-hours trading and opens Thursday at $762, 5.8% above $720.

Step 2: day 1, Thursday. The stock holds above VWAP all day and closes at $768 on three times its average volume. The opening range's high of $764 holds as intraday support. The semiconductor sector, through SMH, the sector's exchange-traded fund, is up 2.5%. Front-month IV has crushed from 48% to 32%, taking the event premium out, and the composite signal is supportive.

Step 3: the Real Value check. Chapter 13B scores single-stock trades on seven factors, each from 0 to 1:

Fundamentals, $F = 0.85$ : a clean beat on all three measures, with raised guidance.
Technical structure, $T = 0.80$ : the gap held, the opening range supports it and the price is above all its major moving averages.
Option flow, $O = 0.70$ : heavy call buying continues after the event, in opening transactions.
GEX path, $G = 0.75$ : post-event GEX is positive, with a call wall at $800.
Volatility value, $V = 0.70$ : IV is reasonable after the crush, so the options are not overpriced.
Execution, $X = 0.85$ : NVDA is liquid.
Sizing, $S = 0.85$ : 1% portfolio risk, with an acceptable correlation to the existing technology book.

Step 4: the product. $\text{RV} = 0.85 \times 0.80 \times 0.70 \times 0.75 \times 0.70 \times 0.85 \times 0.85 = 0.18$ . Every factor looks good, yet the product compresses, which is the multiplicative form's discipline at work.

Step 5: the entry. On Friday morning NVDA pulls back to $755 in the first 30 minutes. You buy 100 shares at $758 with a stop at $748, a risk of $1,000: 0.5% of a $200,000 portfolio, half the usual size.

Step 6: the exit. Four weeks later NVDA is at $815. Half the position was trimmed at $795, and the rest was trailed with a stop at its 10-day moving average and closed at $815. The trimmed half made $50 \times \$37 = \$1{,}850$ and the trailed half $50 \times \$57 = \$2{,}850$ , a total of $4,700 on $1,000 of risk: +4.7R in 28 days.

Three things in this example need care.

The score and the decision disagree. The Dealer Mechanics Lab puts the product at about 0.20 and calls it borderline, at the upper end of the borderline range. It then offers two courses: take the trade at half size, 0.5% risk, or wait up to 5 sessions for a tighter pullback that lifts $T$ to 0.9 or more and $V$ to 0.8 or more. Chapter 13B's own buckets say otherwise. A Real Value from 0.15 to 0.3 is a mixed signal, to be skipped as a common loss-maker for retail, and 0.3 to 0.5 is borderline. Its threshold for taking a trade is 0.4. Even with $T = 0.9$ and $V = 0.8$ the product is only 0.23. On the scorecard as written, this is a skip. The half-size trade that follows shows the management rules at work, not a trade the score passed.

The entry does not match its description. The example says the pullback to $755 holds above the $762 opening-range pivot, but $755 and the $758 entry are both below $762. Read it as a buy on the pullback rather than on the gap, which is what the first failure mode below asks for.

The trim came late against the rule. Taking half off at +0.75R would have sold the first 50 shares at $\$758 + 0.75 \times \$10 = \$765.50$ , for $375. With the rest closed at $815 the trade makes $3,225, 3.2R, against the 4.7R above. At $795 the trim came at +3.7R a share. In a runaway drift like this one the early trim costs money. In the more common trade that stalls near +1R, it is what banks the gain.

The Edge Book reads the GEX factor more coolly. It gives the sign of GEX low confidence and treats walls such as the $800 call wall as no retail edge (Chapters 10H and 10I).

7 · Five ways the trade fails

Buying the gap, not the drift. Entering at the post-event open, at peak euphoria and often near the day's high. The first 30 to 60 minutes are noise, so enter on a pullback to the opening range or to VWAP, not on the gap itself.
Holding through the next earnings. A PEAD trade in a stock that reports again 2 months later carries that report as a binary catalyst. The PEAD edge does not extend through later reports, so exit before the next one or accept that you are now in a different trade.
PEAD on weak beats. Skip beats on EPS alone, missed revenue, and flat or lowered guidance. The drift is conditional on fundamental quality, not on the price gap. PEAD is not "buy any gap up after earnings".
Ignoring the sector. A positive gap into a sector that is already breaking down, with negative sector breadth and peers falling, often fails to drift. The framework needs the broader environment at least neutral and ideally supportive.
The wrong vehicle for the holding period. Calls with 7 days to expiry on a thesis of a 30-day drift lose to theta before the drift develops. The minimum reasonable expiry is 30 to 45 days, and longer is better (Chapter 11M).

8 · Before the report and after it

Factor	Pre-earnings momentum (Chapter 11I)	Post-earnings drift (this chapter)
Evidence tier	Tier B/C, mixed academic support	Tier A, decades of replication
Holding window	1 to 3 weeks before the event	5 to 30 days after the event
Exit discipline	Must exit before the announcement	Hold through the drift, exit on target or stop
Volatility overlay	IV is ramping, favorable to long vega	IV has crushed, so defined-risk plays are cleaner
Default vehicles	Shares, long call, long call spread	Shares, long call spread, put credit spread
Main failure mode	Holding through earnings, a binary outcome	Buying the gap at peak euphoria; holding through the next earnings
Risk character	Continuous and controllable	Continuous and controllable

Two cells need a word from Chapter 11I. There, a long option profits from the ramp only when the market prices the event up, so "favorable to long vega" holds only then. Chapter 11I also lists a diagonal, not a lone long call, as its third vehicle.

The two strategies are complementary, not competing. A well-organized swing book can run both: the pre-earnings trade goes into the event, and the post-earnings trade picks up after the same event resolves positively. The pre-earnings position is closed on T−1. The same stock may then become a PEAD candidate on T+1, if the response to the report is strong enough.

In practice

Read the release, not the headline. The quality condition needs revenue and guidance as well as EPS, and consensus figures to measure the surprise against.
Wait out the first half hour. The drift is not in the opening minutes, and an order at the open buys the euphoria the first failure mode warns about.
Note the next report date at entry. A 60-day time stop can outlast the next report, which must still be avoided.
Count the sector once. Several PEAD trades in one industry, after one strong season, are largely one bet (Chapter 11L).
Expect short-term tax. Holds of 5 to 30 days produce short-term gains on single stocks, and the wash-sale rule applies to quick re-entries (Chapter 2E).

When it fails

The quality was not there. A gap on an EPS beat alone, or against a weak sector, often fails to drift, however strong the day looks.
The gap fades. A stock that falls back below VWAP and the prior close in the opening hour has failed conditions 2 and 3.
The edge has been arbitraged. The Edge Book warns that behavioral drifts compress as more capital buys the post-earnings winners (Chapter 1G), so size for a smaller drift than the history shows.
The next report arrives. Held that long, the trade has become a catalyst gamble.

Decision rule

Qualify: trade only a beat on revenue, EPS and guidance, with a gap above 5% that holds and all seven conditions met.
Enter late: after the first 30 to 60 minutes, on a pullback to the opening range or VWAP, never on the gap at the open.
Choose the vehicle: shares by default, a call spread with 30 to 45 days or more to run for leverage, or a put credit spread below a well-supported gap sized by its maximum loss.
Manage: risk 1% to 1.5%, take half off at +0.75R and trail the rest. Exit on the target, the stop, a fundamental change or 60 days, and always before the next report.

Common mistake: buying the gap

The stock is up 6% at the open, the news is excellent and waiting feels like losing. But the opening minutes are noise, and a buy at the open often marks the day's high. The drift is the edge, not the gap.

Common mistake: treating every gap as PEAD

A gap up on an EPS beat with flat guidance looks like the setup and is not one. The drift is conditional on fundamental quality. Without a revenue beat and raised guidance, skip the trade, whatever the price did.

What to carry forward

PEAD is the stronger earnings edge, Tier A in the Dealer Mechanics Lab's grading, though the Edge Book warns that published behavioral edges decay.
The drift runs over 5 to 60 trading days, not in the first minutes, and is strongest after large surprises confirmed by revenue, guidance and volume.
Seven conditions qualify a trade, and an entry comes on a pullback after the opening range, never on the gap.
Shares by default, a call spread or a put credit spread for options, and no calendars, diagonals or holds through the next report.
Take half off at +0.75R and trail the rest, because most trades end near +1R to +1.5R while a few run to +3R to +5R.

9 · Check yourself

Question 1

Consensus EPS was $1.20 and the company reports $1.29, with revenue slightly below forecast and guidance unchanged. The stock gaps up 6%. Is it a PEAD trade?

Answer

The surprise is $(1.29 - 1.20)/1.20 = 7.5\%$ , above 5%, and the gap is above 5%. But revenue missed and guidance is flat, so the quality condition fails. This is the weak beat the framework says to skip.

Question 2

The first trades of the morning are 2,000 shares at $100.00, 1,000 at $101.00 and 1,000 at $99.50. What is the VWAP, and does a price of $100.40 hold above it?

Answer

The VWAP is $(200{,}000 + 101{,}000 + 99{,}500)/4{,}000 = \$100.125$ , so $100.40 is above it. Condition 3 also needs the price above the prior day's close.

Question 3

You buy 200 shares at $50 with a stop at $45. Where do you take half off, and what does the trade make if the rest is stopped at $50, or runs to $65?

Answer

R is $5 a share and $1,000 in all. Half comes off at $50 + 0.75 \times 5 = \$53.75$ , making $375. If the rest stops at the $50 entry, the trade makes $375, 0.375R. If the rest runs to $65, it adds $100 \times 15 = \$1{,}500$ , for $1,875, 1.875R, as (11J.2) predicts.

Question 4

You sell a $740 put and buy a $720 put for a credit of $3.00. What are the maximum gain, the maximum loss and the breakeven, and why does this position need care?

Answer

The maximum gain is the $300 credit, the maximum loss is $(20 - 3) \times 100 = \$1{,}700$ and the breakeven is $737. It wins a little often and loses a lot rarely, the short-volatility shape, so size it by its $1,700 maximum loss and never hold it through the next report.

Question 5

Why is a call with 7 days to expiry the wrong vehicle for a PEAD thesis?

Answer

The drift unfolds over weeks, while a 7-day call's time value is spent within days, so theta destroys the position before the drift develops. Chapter 11M sets 30 to 45 days as the minimum for a thesis of this length.

Question 6

Seven factors each score 0.8. What is their product, and what does it say about a multiplicative score?

Answer

The product is $0.8^7 = 0.21$ , a mixed signal on Chapter 13B's buckets, although no factor is weak. A multiplicative score falls fast as factors are added, so a single-stock score of seven factors compresses more than the five-factor conviction score does.