Earnings Beat Expectations, Yet Why Did the Market Reverse? Starting with Two Cases from OKX

Author: OKX

After correctly anticipating earnings, the real challenge lies in clearly seeing: which good news has already been priced in, and which changes warrant revaluation.

At dawn on Oct 21 Beijing Time, Netflix reported its Q3 earnings with an EPS of $0.87. The OKX earnings prediction question settled as Yes, yet from the US market close before the report to the close of the first full trading day after it, the corresponding perpetual contract fell by approximately 5.30%. If the performance crossed the threshold, why did the price still drop?

In the earnings season, OKX's "Prophet Season 3" sets up two types of predictions: one asks whether a company's financial report meets a preset threshold, and the other asks whether the price will rise on the first full trading day after the report. Two questions that appear closely related do not necessarily yield the same answer. How much a company earns in a given quarter and how much investors are willing to pay for it need to be viewed separately.

Four Companies, Three Combinations

The preliminary results for Netflix, SpaceX, Tesla, and Meta are as follows:

SpaceX's performance and contract price moved in the same direction upward. Quarterly revenue reached $14.6 billion, up 161% year-over-year; net profit was $4.4 billion, compared to a loss of approximately $1.1 billion in the same period last year. However, how much revenue grows and how much prices rise are two different matters; one cannot directly use the 161% revenue increase to explain the 7.65% contract price surge.

Summarizing the financial reports of the other three companies as simply "good" or "bad" is far less straightforward. Netflix saw YoY EPS growth of approximately 49%, yet lowered its full-year advertising revenue guidance from about $3 billion to $2.7 billion; Tesla posted quarterly revenue of roughly $28 billion, nearly flat YoY, while operating profit fell YoY by about 56%; Meta reported 26% revenue growth amid litigation expenses and large-scale capital expenditures.

If Earnings Cross the Line, Why Does the Market Still Fall?

The "beat expectations" metric in the activity settles based on the preset threshold established beforehand. The market's reaction to earnings reports also depends on what investors had anticipated prior to the report: Can new businesses continue to grow? Will profit margins improve? Will management raise forward-looking guidance? These expectations cannot necessarily be captured by a single EPS question.

NNetflix's two data points point to different timeframes: the ~49% EPS growth reflects money already earned this quarter, while the lowered ad revenue guidance suggests future business may underperform initial estimates. If investors bought in previously anticipating ad business growth, then clearing the quarterly profit target is insufficient to dispel those concerns. The earnings question can settle as Yes, yet investors still have reason to reassess the price.

For Tesla, crossing the question's threshold coincided with declining operating profit. Investors must also judge whether profit declines will persist despite negligible revenue growth, and what level of earnings performance is needed to justify current valuations. For such companies, meeting the question's threshold only completes part of the financial analysis.

This does not mean that all post-earnings drops can be attributed to "the good news has been realized." One must also look back at how much the stock rose before the report, whether earnings forecasts were raised, and if valuations became more expensive. Sometimes price increases merely follow the broader sector. Management's explanations during conference calls and concurrent market shifts also influence trading. Advertising outlooks and profit pullbacks provide analytical clues, but confirming their impact on the decline requires more information.

Beyond Profit, Examine Inputs and Returns

There is another aspect of Meta's earnings worth examining closely. Calculated under US GAAP, its EPS is $3.83; excluding litigation-related expenses, it stands at $7.40. While the activity settles according to the agreed GAAP metric, researching a company shouldn't stop at this answer. One must also see how much of the profit change stems from core operations versus these expenses, and how long the impact will persist.

These expenses create a clear divergence between the two EPS figures. The activity can provide an answer per the rules, but investors need to look at both sides: profits excluding expenses help observe operational health, while GAAP-recognized expenses undeniably impact shareholder returns. They cannot be ignored simply because they are special items, nor should ongoing business deterioration be assumed based solely on them.

Meta's quarterly capital expenditure was approximately $34.9 billion, with free cash flow at -$2.12 billion, raising another question: as revenue keeps growing, how much cash can the company actually retain? Free cash flow is typically calculated as operating cash flow minus capital expenditure, subject to specific company disclosure metrics. Compared to EPS, it more directly reflects cash positions after accounting for relevant investments.

Purchasing equipment and building data centers requires upfront cash payments, while related costs are usually depreciated gradually over their useful lives. This explains why a company might face cash flow pressure due to expansion even while remaining profitable. In Meta's report, profitability metrics and capex must be viewed together; looking solely at EPS makes it easy to miss the costs required for future growth.

For these investments, the market now waits for returns. If newly added capacity is utilized by customers and generates sufficient profit, today's spending could fuel future growth; if revenue lags and requires constant additional funding, investors will need to reconsider valuations. The question left by Meta's report will also extend to subsequent AI supply chain companies.

How Predictions Assist in Reading the Market?

Placing both prediction types side by side helps observe whether participants' views on earnings and prices align. Prediction readings are formed by trades, reflecting prevailing sentiment at the time rather than the actual probability of an event occurring; user selection ratios and long-term accuracy rates each have their own statistical methodologies. They can indicate research directions but cannot directly prove whether a stock is overvalued or undervalued.

If the earnings question heavily leans Yes while the price question is less optimistic, the focus should be on this divergence: Do participants believe the profit threshold is easily met but remain more concerned about forward guidance or capex? Has the previous rally already priced in the good news? When both lean Yes, one must assess how much actual earnings can exceed existing expectations. When both lean No, it's necessary to consider whether the prior downturn has already digested some worries.

Tracking reading changes also requires noting when they occur. Participants possess different information on the day before the report, an hour before publication, and during the conference call. If the earnings question consistently leans Yes but the price question gradually weakens, further investigation into new information revealed during that period is warranted. Views formed only after the report is released cannot be counted as pre-release predictions.

Recording these changes alongside pre-event reasoning provides concrete content to verify upon release. If the earnings question proves wrong, check operating forecasts and accounting metrics; if the earnings question is right but the price question is wrong, check for overestimation of the market's reaction to good news or missed guidance shifts. Even when the directional guess is correct, distinguish whether company performance matched expectations or if the price simply coincidentally rose with the sector. Such review prevents cherry-picking highlights after rallies and focusing only on problems after dips.

If applying this analysis to live trading, entry price, potential gains and losses, costs, and leverage must be considered separately. A higher probability of an upside might still result in minimal gains; conversely, a downside could lead to larger losses. Therefore, even after getting the direction right, position sizing and exit conditions must still be properly planned.

The Streak Challenge of the "Earnings Big Winner"

The OKX "Earnings Big Winner" activity strings the earnings and price questions into a continuous challenge. Over 700 participants joined the first phase; among the initial 150 players securing back-to-back wins, only 3 ultimately achieved six consecutive wins. All three stopped in the same round, splitting the 10,000 USDT prize. The progressively shortened list adds a spectator appeal, though these streak results alone are insufficient to demonstrate the long-term accuracy of prediction markets.

Phase 2 granted participants a comeback opportunity: players could apply to revive by publishing posts with position cards and viewpoints within the specified timeframe per the rules. However, failing two or more questions in a single day, or missing again after reviving, disqualifies them from advancing further. Both earnings and price questions must be attempted; participants cannot simply pick whichever category they feel more confident about.

For Robinhood and SK Hynix in Phase 2, both earnings questions resolved to No, but price questions diverged. Robinhood's GAAP EPS was $0.51, falling short of the requirement to be "strictly above $0.56"; although adjusted EPS stood at $0.60, the activity still settled according to the designated metric. Robinhood's price question was also No, while SK Hynix's price question resolved to Yes. Neither cleared the earnings hurdle, yet prices moved in opposite directions.

Google's outcome also differed from many participants' choices. Prior to the report, OKX China disclosed that approximately 62% of users selected "beat expectations"; Alphabet ultimately reported a GAAP EPS of $2.12, missing the question's stipulated requirement of being "strictly above $3.10". Growth in Google Cloud and whether the group's EPS crosses the line are two separate matters. Additionally, the 62% here counts user selection ratios, not trade-derived prediction prices, so one cannot deduce why each participant chose Yes based on it.

Successive questions across different companies also make maintaining a winning streak increasingly difficult. Streaming platforms require monitoring ad growth, exchanges track client activity, and memory chip makers watch demand and pricing; these operational details ultimately map to their respective question metrics. Participants must rapidly switch research focus and then choose how the market will react post-earnings.

According to the October 30 battle report, the three advancing competitors in Phase 2 achieved a "seven-streak," but their comeback opportunities have been exhausted. Five more first-day post-earnings price movement predictions remain, meaning these three are not yet the final winners. Even having correctly forecasted performance earlier, they must still anticipate market reactions going forward, with zero margin for error remaining.

What to Watch in Upcoming Earnings Reports?

Upcoming earnings still warrant viewing the AI supply chain as an integrated whole. Cloud services and capex from Microsoft, Alphabet, and Amazon help illuminate upstream demand. For cloud platforms, purchasing equipment represents an investment; for chip, storage, and optical communication suppliers, it translates into orders. The input-output question Meta leaves behind can be traced further down this chain.

Cloud platforms must monitor whether customer usage and revenue keep pace with investments; suppliers need to track on-time delivery fulfillment and eventual profit/cash retention. For chip and storage firms, revenue growth requires dissecting whether it stems from higher volume, price hikes, or a rising share of high-value products. Optical communications, compute infrastructure, and energy enterprises must particularly align delivery progress with capital outlays.

Spending money, receiving orders, delivering goods, recognizing revenue, and collecting payments often span different quarters. Linking upstream and downstream data makes it easier to see if growth can sustain and explains why financial and price performances still vary across companies despite all benefiting from AI.

Another group worth comparing includes Coinbase, Circle, and Strategy. They respectively involve trading, stablecoin operations, and digital asset pricing, meaning active market conditions impact them differently. When reading earnings, distinguish which growth comes from business expansion versus being driven by prevailing market prices. Interest rates and risk appetite form a common backdrop, but for each specific company, revenue and balance sheet details still take precedence.

Therefore, upcoming earnings warrant comparing not just which companies crossed profit thresholds, but also whether revenue growth actually translated into profits and cash. If several data points diverge, continue investigating where growth originates and where costs rise, then determine if they alter prior valuation assumptions for the company.

Summary

From Netflix's "earnings beat, price drop" to the two divergent combinations shown by Robinhood and SK Hynix, these cases repeatedly point to the same issue: markets trade not on a report card itself, but on how that report card alters future judgments. Predictive events assign a clear threshold to one judgment; stock prices continue absorbing shifts in growth guidance, valuations, cost of capital, and risk appetite. The two can move in tandem or in opposite directions.

Therefore, the transactional reference value of predictions does not lie in prescribing a guaranteed profitable direction, but in forcing participants to separately judge "how earnings will pan out" versus "how the market will react," verifying item by item after results drop. The more companies and fragmented industries come next, the more crucial this distinction becomes: first comprehend the questions and data, then identify expectations already baked into prices, and finally observe how new information shifts trends. What endures is not merely a string of Yeses or Nos, but a methodology of judgment that withstands scrutiny and remains open to correction.