What lasts is not just a string of Yes or No calls, but a way of judging events that can be tested and revised.
By OKX
After getting the earnings call right, the harder part is figuring out which good news was already priced in and which changes still deserve a fresh repricing.
In the early hours of Oct. 21 Beijing time, Netflix reported third-quarter results with earnings per share of $0.87. The OKX earnings prediction question settled as Yes. Even so, the related perpetual contract fell about 5.30% from the U.S. stock market close before the report to the close of the first full trading day after earnings. Why would price fall when the company cleared the line?
In its "Prophet Season 3" earnings event, OKX set up two kinds of questions during earnings season. One asked whether a company’s results would meet a preset threshold. The other asked whether the price would rise on the first full trading day after the report. The two look closely linked, but they do not have to produce the same answer. How much a company earned this quarter and how much investors are willing to pay for it are separate questions.
Four companies, three kinds of combinations
The first round of results for Netflix, SpaceX, Tesla and Meta showed several different patterns.
For SpaceX, earnings and contract price moved in the same direction. Quarterly revenue reached $14.6 billion, up 161% year over year, while net profit was $4.4 billion after a loss of about $1.1 billion in the same period a year earlier. Still, revenue growth and price gains are not the same thing. A 161% increase in revenue cannot directly explain a 7.65% rise in the contract.
The other three cases were harder to label simply as good or bad. Netflix lifted EPS by about 49% year over year, but cut its full-year advertising revenue outlook from about $3 billion to about $2.7 billion. Tesla posted quarterly revenue of about $28 billion, almost flat from a year earlier, while operating profit fell about 56% year over year. Meta grew revenue by 26%, but also carried litigation costs and heavy capital spending.
Why price can still fall after earnings clear the threshold
In the event, a result counts as "above expectations" if it clears the threshold set in advance by the question. The market’s reaction depends on something else as well: what investors were already expecting before the release. That includes whether a new business can keep growing, whether margins can improve, and whether management will raise guidance. A single EPS question does not capture all of that.
Netflix’s numbers point to two different time frames. EPS growth of about 49% describes money already earned this quarter. The lower advertising revenue outlook points to business that may come in weaker than previously expected. If investors had bought the stock partly on confidence in ad growth, then meeting the quarter’s profit target would not remove that concern. The earnings question can settle Yes while investors still decide the price needs to be reassessed.
Tesla presents a different version of the same split. The question cleared its threshold while operating profit was falling. Investors still have to judge whether that profit decline will continue when revenue is barely growing, and what level of earnings is needed to support the current valuation. For companies like this, clearing the event threshold only completes part of the earnings analysis.
That does not mean every post-earnings decline can be reduced to "good news was already priced in." It is still necessary to look back at how much the stock had risen before the report, whether earnings forecasts had been revised higher, and whether valuation had become more expensive. Sometimes a stock rises only because the whole sector is moving. Management commentary on the earnings call and broader market moves at the same time can also shape trading. The weaker ad outlook and the drop in profit offer clues, but more information is needed before assigning them a precise share of the decline.
Beyond profit: spending and returns matter too
Meta’s report offers another point worth examining closely. Under U.S. GAAP, its earnings per share were $3.83. Excluding litigation-related costs, EPS was $7.40. The event settled according to the agreed GAAP measure, but company analysis cannot stop there. It also has to ask how much of the profit change came from core operations, how much came from that expense, and how long the effect may last.
The expense created a wide gap between the two EPS figures. The event can produce one answer under its rules, but investors still need to look at both sides. Earnings excluding the charge help show operating conditions. The expense recognized under GAAP also affects shareholder returns. It should not be ignored because it is unusual, but neither should one item be used on its own to conclude that the core business will keep deteriorating.
Meta also reported capital expenditure of about $34.9 billion for the quarter and free cash flow of negative $2.12 billion. That raises another question: with revenue still growing, how much cash is the company actually keeping? Free cash flow is usually calculated as operating cash flow minus capital expenditure, subject to the company’s own disclosure method. Compared with EPS, it gives a more direct view of cash left after the spending needed to support expansion.
When a company buys equipment or builds data centers, cash goes out first, while the related cost is often recognized gradually through depreciation over the useful life of the assets. That helps explain why a company can remain profitable and still face cash flow pressure during expansion. In Meta’s case, the earnings measure and the capital spending line need to be read together. Looking only at EPS can miss the cost of the growth the company is trying to build.
The next question for the market is return on that spending. If the added capacity finds customers and generates enough profit, today’s outlay may support future growth. If revenue does not catch up and more capital keeps being required, investors may need to rethink valuation. The issue left by Meta’s report extends to other companies across the AI supply chain.
How prediction tools can help read the market
Putting the two kinds of questions side by side can help show whether participants see earnings and price in the same way. Prediction readings are formed through trading and reflect the leaning at that moment. They are not the same as the true probability of an event. User choice ratios and long-run accuracy rates also come from different statistical methods. They can point to areas worth studying, but they do not prove that a stock is expensive or cheap.
If the earnings question leans clearly toward Yes while the price question is less optimistic, that gap is worth watching. Participants may think the earnings threshold is easy to clear while worrying more about guidance or spending. It may also mean the prior rally had already reflected much of the good news. If both questions lean Yes, the next step is to ask how much room remains for the actual report to beat what was already expected. If both lean No, it is also necessary to consider whether earlier declines had already absorbed part of the concern.
Timing matters when reading those shifts. Participants know different things one day before earnings, one hour before the release, and during the conference call. If the earnings question stays tilted toward Yes while the price question gradually weakens, the next step is to check what new information appeared during that period. Views formed only after the report should not be counted as pre-release predictions.
Recording those changes together with the reasons given beforehand makes post-event review more concrete. If the earnings question was wrong, the review can focus on operating estimates and accounting definitions. If the earnings question was right but the price question was wrong, the review can ask whether the market’s reaction to good news was overestimated or whether a guidance change was missed. Even when the direction is correct, it still matters whether the company truly met expectations or whether the price simply rose with the sector. That kind of review helps avoid looking only for positives after a rally and only for problems after a drop.
If the analysis is used in actual trading, entry price, upside, downside, costs and leverage still need to be considered separately. A higher probability of a rise does not mean the gain will be large. A decline, if it comes, may still produce a bigger loss. Getting the direction right is only part of the job; position sizing and exit conditions still have to be set.
The winning streak test in "Earnings Winner"
OKX’s "Earnings Winner" event linked earnings questions and price questions into a running challenge. More than 700 participants joined the first round. Of the 150 players who started with two straight correct calls, only three eventually reached six in a row. All three were eliminated in the same round and split a 10,000 USDT prize. The shrinking list added suspense, but the streak data alone are not enough to show how accurate a prediction market is over the long run.
The second round gave participants one revival chance. Under the rules, users could apply for revival by posting a position card and their view within the required time window. But anyone who missed two or more questions on the same day, or missed again after revival, could not continue. Participants had to answer both the earnings question and the price question; they could not choose only the type they felt more confident about.
In the second round, Robinhood and SK Hynix both settled No on the earnings question, but their price questions diverged. Robinhood reported GAAP EPS of $0.51, which did not meet the requirement of being strictly above $0.56. Adjusted EPS was $0.60, but the event still settled according to the specified metric. Robinhood’s price question also settled No, while SK Hynix’s price question settled Yes. Both companies missed the earnings threshold, yet their prices moved in different directions.
Google’s result also differed from what many participants chose. Before the report, OKX Chinese disclosed that about 62% of users selected "above expectations." Alphabet ultimately reported GAAP EPS of $2.12, below the question’s requirement of being strictly above $3.10. Growth in Google Cloud and whether group EPS clears the line are two different matters. The 62% figure here refers to the share of user choices, not a prediction price formed through trading, and it does not show why each participant chose Yes.
As questions for different companies appeared one after another, maintaining a streak became harder. Streaming requires attention to advertising growth. Trading platforms require attention to client activity. Memory chip companies also require attention to demand and pricing. Those operating details then have to be matched back to the exact wording of each question. Participants need to switch research focus quickly and then make a separate call on the market reaction after earnings.
According to the Oct. 30 update, three participants in the second round had reached seven straight correct calls, but all revival chances had already been used. There are still five first-day post-earnings up-or-down predictions left, and the three are not yet final winners. Even after getting earnings right, they still have to judge how the market will react next, with no room left for error.
What to watch in the next earnings reports
In the reports ahead, the AI supply chain remains worth tracking as a group. Cloud business and capital spending at Microsoft, Alphabet and Amazon can help show upstream demand. For cloud platforms, buying equipment is an investment. For chip, memory and optical communications suppliers, the same spending may become orders. The question Meta left on return versus investment can be followed further down that chain.
Cloud platforms need to show whether customer usage and revenue can keep up with spending. Suppliers need to show whether orders can be delivered on time and how much profit and cash remain at the end. For chip and memory companies, revenue growth also needs to be broken down: did they sell more units, raise prices, or shift toward a higher-value product mix? Optical communications, computing infrastructure and energy companies need even closer comparison between delivery progress and capital outlays.
Spending money, taking orders, shipping products, recognizing revenue and collecting cash often do not happen in the same quarter. Connecting the data across upstream and downstream companies makes it easier to judge whether growth can last. It also helps explain why companies that all benefit from AI can still show different earnings and price reactions.
Another group worth comparing is Coinbase, Circle and Strategy. They touch trading, stablecoin business and digital asset prices, and active markets affect each of them differently. When reading their reports, it is necessary to separate growth driven by business expansion from growth driven mainly by market prices at the time. Interest rates and risk appetite form a common backdrop, but each company still has to be judged on its own revenue and balance sheet.
That means the next set of earnings should not be compared only by which companies beat a profit threshold. It also matters whether revenue growth turned into profit and cash. If several metrics move in different directions, the next step is to check where the growth came from, where costs increased, and whether those changes alter the valuation framework that investors had before.
Closing point
From Netflix’s combination of an earnings pass and a price decline to the different outcomes seen in Robinhood and SK Hynix, the same issue keeps coming back: the market is not trading a report card by itself, but how that report card changes views of the future. Event prediction gives one of those judgments a clear threshold. Share prices continue to absorb changes in growth guidance, valuation, funding costs and risk appetite. The two can move together, or they can split.
That is why the value of prediction tools is not that they deliver a guaranteed profitable direction. Their value is that they force participants to separate the question of what earnings will be from the question of how the market will respond, and then check each part after the result is out. As more companies report and sectors become more varied, that distinction matters more: read the wording and the data first, identify what expectations are already embedded in price, and then watch how new information changes the move. What remains is not just a string of Yes or No answers, but a method of judgment that can stand up to review and revision.
