A 15-minute game of wins and losses: Millions of transaction records unveil the "folded world" of the Bitcoin prediction market.

Author: Frank, PANews

In the world of cryptocurrency, what can happen in 15 minutes? For most people, it's just the formation of a candlestick chart; but for participants in the Bitcoin short-term prediction market, it often means "a game of life or death."

Recently, the PANews analysis team conducted a full data review of the recent Bitcoin 15-minute price movement prediction market. In a massive database covering approximately 3 days, 291 short-term markets, and a total of 1.05 million transactions, we see more than just cold, hard numbers; we see a raw and honest battle between algorithms and human nature.

This is not a playground where you can get rich by luck, but a folded world ruled by 3.6% algorithmic robots.

A bustling ant market at a lottery center frequented by individual lottery players.

If you only look at the macro data, this market presents a very lively scene.

Over the course of three days, the BTC 15-minute prediction market generated 1.05 million transactions, with a total trading volume of approximately $17 million. The average trading volume per market was approximately $58,600. However, based on these trading volume figures, the crypto prediction market is still relatively small, far smaller than the trading volume of the traditional crypto market.

During this period, a total of 17,254 unique addresses participated in the market's transactions. The average number of unique trading addresses per market was 881. The average transaction amount was $16.22, which means that the market was not mainly composed of institutional investors betting against each other, but rather of high-frequency "lottery-buying" behavior by tens of thousands of retail investors.

Of these, 8,054 addresses were profitable, while 8,884 unique addresses incurred losses. The ratio of profitable to loss-making addresses was approximately 1:1.1. The market did not experience a one-sided massacre; most losers only suffered minor losses, and this illusion that "it's still playable" retained a large number of users.

However, the limitations of market depth are also glaringly obvious. Data shows that the most profitable address earned a total of $54,531, while the address with the largest loss incurred a loss of $62,184. This data illustrates that market liquidity limits the potential for large-scale trading; it's difficult to earn millions of dollars in a single transaction here because your opponents' pockets aren't deep enough.

The median number of addresses entering the market was 0.544, indicating that buyers generally entered with either a bullish or bearish outlook. However, the median number of exits was 0.247. This means that the vast majority of proactive selling was "panic selling," with an average loss of about 50% per transaction. This also suggests that retail investors often fail to hold onto profitable trades, while frequently trading on losing trades, ultimately handing their shares back to market makers at low prices.

Robots vs. real users: 3.6% of the market is dominated by robots.

If retail investors are playing psychological warfare, then their opponents are launching a ruthless, devastating attack. Data analysis bluntly reveals that in this market, manual traders are facing a complete and utter crushing defeat from algorithms.

First, in terms of results, the robot addresses did indeed completely outperform the real users in terms of earnings.

Although these bots only account for a tiny fraction of the total number of addresses—only 247, or 3.6%—they contributed over 600,000 transactions, representing more than 60% of all transactions. This suggests that a very small number of algorithms dominate pricing power and liquidity, while the vast majority of retail investors are merely providing funds as consumables.

In terms of transaction amount, the ratio of robot transaction amount to that of real users is quite close.

Furthermore, the bots' advantage in profitability is quite significant. Purely bot-based addresses earned approximately $284,000 in total over these three days, while addresses using bot-like trading, human-like trading, and purely human trading all suffered negative returns overall. In contrast, real traders' overall profit and loss was -$154,000. Every penny of excess profit in the market essentially transferred from the pockets of real users to the algorithm's account. Manual trading faces an insurmountable gap in its ability to compete with high-frequency algorithms.

In terms of win rate, bot addresses also performed better, with an average win rate of about 65.5%, while the win rate of real users was only 51.5%.

From this perspective, the analysis reveals that the current crypto short-term prediction market is characterized by bots profiting at the expense of real users, with manual trading yielding significantly lower results compared to high-frequency bots. This also indirectly confirms that it is possible to achieve excess returns in the prediction market through algorithmic optimization.

Smart Money Unveiled: "Speed" is Poison, "Accuracy" is the Antidote

However, if you think you can make money effortlessly just by writing a script and running a bot, you're sorely mistaken. We discovered a counterintuitive phenomenon in the list of top profit earners: the world of bots has also seen dramatic differentiation, and "high frequency" does not equal "excessive profits."

Taking the address 0x5567...a7b1 as an example, it has the most transactions of all addresses. It has conducted over 33,700 transactions in total, averaging over 67 transactions per hour. However, its profit is relatively meager, only $4,989, averaging just $0.14 per transaction.

This is not an isolated case. Data shows that among ultra-high frequency addresses that make more than 50 transactions per hour, only 40% are profitable, and the average return for this group is even -10%. Amidst gas fees, slippage, and extremely fierce competition, bots that blindly pursue speed are ultimately just working for the miners.

Let's look at another case. The address 0x0ea5...17e4 is also a bot address, and its profitability ranks first among all addresses. However, its trading frequency is not as high, averaging only 22 trades per hour, and it only participates in 61% of the market. This means that this address's trading logic does not place orders every second, but rather trades based on specific filtering conditions, only doing so when the market meets the corresponding trading conditions. This address has a win rate of 72%, with total profits of approximately $54,500.

Risk control has become the Achilles' heel of human traders.

In addition, data also offers a glimmer of hope for human traders.

We found that addresses with extremely low trading frequency (less than one transaction per hour) achieved an average win rate of 55%, far exceeding that of high-frequency trading bots that blindly engage in order manipulation. This demonstrates that, without the support of top-tier algorithms, manual judgment based on market intuition and logic yields a higher win rate than algorithm-based bots.

But where did humanity go wrong? The data suggests the answer: risk control.

Low-frequency traders (1-5 trades per hour) have an average loss of approximately $47 per trade, the highest among all address categories. Human traders often correctly predict the direction, but human weaknesses cause them to stubbornly hold onto losing positions and fail to hold onto winning ones. Ultimately, the "small wins, big losses" profit-loss ratio becomes the biggest curse for human traders in this market.

1.05 million transaction records and $17 million in transactions reveal a harsh truth:

The Bitcoin 15-minute market prediction is not a cash cow for retail investors, but a food chain where top-tier algorithms harvest inferior algorithms, and then inferior algorithms harvest humanity.

For the average participant, the data offers a chilling piece of advice: either evolve into a top-tier sniper with a 72% win rate, or become an extremely restrained, low-frequency hunter. Any more frequent actions, any attempt to compensate for a skill gap through "diligence," will ultimately only make you a part of the profit-generating ecosystem.