The Earnings Prediction Was 'Right' — Why Did the Market Still Move Against It? Starting with Two OKX Questions

Author: OKX

After correctly reading earnings, the real challenge is seeing which good news was already priced in and which changes deserve repricing.

Early morning on Oct. 21 (Beijing Time), Netflix reported Q3 earnings with EPS of $0.87. The OKX earnings prediction question settled as Yes, but from the U.S. equity market close before earnings to the close of the first full trading day after earnings, the corresponding perpetual futures contract fell about 5.30%. Earnings cleared the threshold — why did the price still drop?

OKX 'Oracle Season 3' offered two types of predictions during earnings season: one asked whether a company's earnings would meet a preset threshold, and the other asked whether the price would rise on the first full trading day after earnings. Two questions that appear closely related may not produce the same answer. How well a company performed this quarter and how much investors are willing to pay for it need to be viewed separately.

Four Companies, Three Combinations

The first-round results for Netflix, SpaceX, Tesla, and Meta are as follows:

SpaceX's earnings and contract price rose in the same direction. Quarterly revenue was $14.6 billion, up 161% year over year; net income was $4.4 billion, compared with a loss of about $1.1 billion in the prior-year period. However, how much revenue grew and how much the price rose are two different things; the 161% revenue increase cannot directly explain the 7.65% contract gain.

The other three earnings reports were not so easily summed up as 'good' or 'bad.' Netflix's EPS grew about 49% year over year, but it lowered its full-year advertising revenue outlook from about $3.0 billion to about $2.7 billion; Tesla's quarterly revenue was about $28.0 billion, roughly flat year over year, while operating profit fell about 56% year over year; Meta's revenue grew 26%, alongside litigation costs and large-scale capital spending.

Earnings Cleared the Threshold — Why Did the Market Still Fall?

In the campaign, 'beating expectations' is settled according to the threshold set in advance by the question. The market's reaction to earnings also depends on what investors had previously expected: whether new businesses could continue growing, whether margins could improve, and whether the company would raise its forward guidance. These expectations may not be captured by a single EPS question.

Netflix's two data points point to different time horizons: EPS growth of about 49% speaks to money already earned this quarter; the lowered advertising revenue outlook speaks to future business that may not be as strong as previously expected. If investors who bought earlier also expected advertising growth, meeting the quarterly profit target is not enough to remove that concern. The earnings question can settle as Yes, while investors may still have reason to reassess the price.

Tesla's case is that clearing the question threshold and declining operating profit occurred at the same time. Investors also need to judge whether, with revenue nearly flat, the profit decline will persist, and what earnings performance is needed to support the current valuation. For companies like this, meeting the question threshold only completes part of the earnings analysis.

This does not mean that every post-earnings decline can be attributed to 'good news already priced in.' One must also look back at how much the stock rose before earnings, whether earnings forecasts were raised, and whether the valuation became more expensive. Sometimes price gains merely follow the broader sector. Management's explanations on the earnings call and concurrent market changes also affect trading. The advertising outlook and profit decline provide analytical clues, but determining their impact on the drop requires more information.

Beyond Profit, Look at Spending and Returns

Meta's earnings report has another point worth examining closely. Under U.S. generally accepted accounting principles (GAAP), its EPS was $3.83; excluding litigation-related costs, it was $7.40. The campaign settles according to the agreed GAAP basis, but when researching the company, one cannot stop at that answer. Investors also need to see how much of the profit change came from the core business, how much came from this cost, and how long its impact will last.

This cost creates a clear gap between the two EPS measures. The campaign can give one answer by rule, but investors need to look at both sides: earnings excluding the cost help assess operations, while the cost recognized under GAAP did affect shareholder earnings. It should not be ignored simply because it is a special item, nor should one conclude solely from this cost that the core business will continue to deteriorate.

Meta's quarterly capital expenditures were about $34.9 billion, and free cash flow was negative $2.12 billion, raising another question: with revenue still growing, how much cash can the company retain? Free cash flow is usually calculated as operating cash flow minus capital expenditures, subject to the company's disclosed definition. Compared with EPS, it more directly reflects cash after completing the relevant investments.

When buying equipment and building data centers, cash is paid out upfront, while the related costs are usually depreciated gradually over the useful life. This explains why a company can face cash-flow pressure from expansion even if it is still profitable. In Meta's earnings report, the earnings measure and capital expenditures need to be viewed together; looking only at earnings per share can make it easy to miss the costs required for future growth.

For this investment, the market now needs to wait for returns. If new capacity is used by customers and can generate sufficient profit, today's spending may support future growth; if revenue does not catch up and additional funding is required, investors will need to reassess valuation. This question left by Meta's earnings report will also carry over to later companies in the AI supply chain.

How can forecasts help interpret market conditions?

Putting the two types of forecasts together can help observe whether participants' views on earnings and price are aligned. Forecast readings are formed by trading and reflect the prevailing tendency at that time; they are not equal to the probability that an event actually occurs. User selection share and long-term accuracy also have their own statistical measures. They can suggest a research direction, but cannot directly prove whether a stock is overvalued or undervalued.

If the earnings question is clearly tilted toward Yes while the price question is less optimistic, the part worth watching is this divergence: do participants think the earnings hurdle is easy to clear but worry more about subsequent guidance or spending? Has the prior rally already priced in good news? When both questions tilt toward Yes, consider how much room there is for the actual earnings report to exceed existing expectations. When both questions tilt toward No, also consider whether the prior decline has already absorbed some concerns.

When reading changes in these readings, remember when they occurred. The day before earnings, the hour before release, and during the earnings call, participants know different information. If the earnings question consistently tilts toward Yes while the price question gradually weakens, keep investigating what new information emerged during that period. Views formed after the earnings report is released cannot be counted as pre-release forecasts.

Recording these changes together with pre-event reasoning gives you concrete content to check after the release. If the earnings question is wrong, you can examine operating estimates and accounting measures. If the earnings question is right but the price question is wrong, you can check whether you overestimated the market's reaction to good news or missed changes in guidance. Even if the direction is correct, distinguish whether the company's performance met expectations or the price simply rose with the sector. This kind of review helps avoid looking only for positives after a rally and only for problems after a decline.

If the analysis is used for real trading, you also need to separately consider entry price, potential gain and loss, costs, and leverage. The probability of an increase may be high, but the move may be small; if it falls, losses may be larger. Therefore, even after getting the direction right, you still need to manage position size and exit conditions.

The Consecutive-Win Challenge of 'Earnings Big Winner'

OKX's 'Earnings Big Winner' campaign linked the earnings question and the price question into a consecutive challenge. In the first edition, more than 700 participants took part. Among the 150 players who achieved two consecutive wins in the first round, only 3 ultimately reached six consecutive wins. All three were eliminated in the same round and split the 10,000 USDT prize. The list shrinking round by round added spectator suspense, but this set of winning-streak results is still not enough to show how accurate prediction markets are over the long term.

The second edition gave participants a revival opportunity: according to the rules, you could apply for revival by publishing a post with a position card and your view within the specified time. However, if you missed two or more questions on the same day, or missed again after being revived, you could not continue to advance. Both the earnings question and the price question had to be completed; participants could not only choose the category in which they were more confident.

In the second edition's Robinhood and SK hynix cases, the earnings question was No for both, while the price question differed. Robinhood's GAAP earnings per share was 0.51 USD, failing to meet the requirement of 'strictly higher than 0.56 USD'; although adjusted earnings per share was 0.60 USD, the campaign still settled according to the designated measure. Robinhood's price question was also No, while SK hynix's price question was Yes. Neither company passed the earnings question, but their prices moved in different directions.

Google's result also differed from many participants' choices. Before earnings, OKX's Chinese-language disclosure showed that about 62% of users chose 'beat expectations'; Alphabet ultimately reported GAAP earnings per share of 2.12 USD, failing to meet the question's requirement of 'strictly higher than 3.10 USD'. Google Cloud growth and whether the group's earnings per share cleared the threshold are two different things. In addition, the 62% here measures user selection share, not the forecast price formed by trading, and it cannot be used to infer why each participant chose Yes.

Questions for different companies appearing one after another make winning streaks increasingly difficult. For streaming, watch ad growth; for trading platforms, watch customer activity; for memory chips, watch demand and prices. These operating metrics ultimately map to the criteria of each question. Participants must switch research focus quickly and then make choices about the market reaction after earnings.

According tothe October 30 Recap, three finalists in the second round completed seven consecutive correct predictions, but all revival chances have been used up. There are still 5 post-earnings first-day up/down predictions ahead, and the three are not yet final winners. Even if they previously called the earnings correctly, they still need to judge how the market will react, and there is no room for error.

What to look for in upcoming earnings?

In upcoming earnings, the AI supply chain is still worth examining together. Microsoft, Alphabet, and Amazon's cloud businesses and capital expenditures can help gauge upstream demand. For cloud platforms, buying equipment is an investment; for chip, memory, and optical communications suppliers, it may turn into orders. The question of return on investment left by Meta can be followed down this chain.

Cloud platforms need to see whether customer usage and revenue can keep pace with investment; suppliers need to see whether orders can be delivered on time and how much profit and cash remain. For chip and memory companies, revenue growth must be distinguished among higher volumes, higher prices, or a rising mix of high-value products. Optical communications, compute infrastructure, and energy companies need to compare delivery progress with capital deployment.

Spending, booking orders, delivering, recognizing revenue, and collecting cash often do not occur in the same quarter. Linking upstream and downstream data makes it easier to see whether growth is sustainable and explains why companies that all benefit from AI can still show different earnings and price performance.

Another group worth comparing is Coinbase, Circle, and Strategy. They cover trading, stablecoin businesses, and digital asset prices, respectively, and the impact of market activity differs. When reading earnings, distinguish which growth comes from business expansion and which is merely driven by prevailing market prices. Interest rates and risk appetite are the common backdrop, but for each company, one still needs to look at its own revenue and balance sheet.

Therefore, what is worth comparing in subsequent earnings is not only which companies exceeded profitability thresholds, but also whether revenue growth translated into profit and cash. If several metrics move in different directions, continue to investigate where growth comes from and where costs are rising, then see whether they have changed the prior valuation of the company.

Summary

From Netflix's 'earnings beat, price decline' to the two combinations shown by Robinhood and SK Hynix, these cases repeatedly point to the same question: the market is not trading a report card, but how the report card changes expectations about the future. The prediction event gives one of those judgments a clear threshold; the stock price continues to absorb changes in growth guidance, valuation, cost of capital, and risk appetite. The two can move in the same direction or in opposite directions.

Therefore, the trading value of predictions lies not in providing a guaranteed profitable direction, but in forcing participants to separately judge 'how earnings will turn out' and 'how the market will react', then check each item after the results come out. The more companies and the more dispersed the industries are going forward, the more important this distinction becomes: first understand the questions and data, then identify the expectations already priced in, and finally observe how new information changes the market. What can endure is not just a string of Yes or No, but a method of judgment that can withstand testing and allow revision.

Source: TechFlow