Disclaimer

ForkLog assumes no responsibility for the investment decisions of its readers.

In early October 2025, market participants were more optimistic than ever: Bitcoin surpassed $125,000, analysts pointed to a coin shortage on exchanges, and many predicted a rally towards $200,000. Amid this euphoria, it seemed that significant corrections were a thing of the past.

However, on the night of October 11, the price of the leading cryptocurrency plummeted by $20,000 — marking a record daily fluctuation for the market. Exchanges forcibly closed positions worth $19 billion, leading to a nearly $660 billion drop in the total market capitalization of digital assets. Even traders with seemingly profitable positions were affected, as the ADL mechanism liquidated them on both sides of the order book.

This situation raises important questions about the true costs incurred by thrill-seekers and quick profit hunters, the pros and cons of spot trading versus leverage, and what historical data reveals about both market participant categories.

Hidden Asymmetry

When engaging in margin trading, a trader typically considers the most apparent outcome: a price decline that results in a loss. However, losses can occur without any forecasting errors.

A 10% price correction requires an 11.1% increase to return to the original value. A 25% drop necessitates a 33% gain relative to the remaining capital. After a 50% decline, the remaining capital must double, and after an 80% drop, it needs to quintuple.

This asymmetry arises because losses are calculated based on the initial capital, whereas recovery is measured from the diminished amount after a decline. As this remaining amount decreases, a larger percentage increase is required to return to the starting level.

Percentage increase needed to restore an asset's value to its original level. Source: ForkLog.

Leverage amplifies the speed at which account results fluctuate. With a tenfold leverage, a 10% drop can wipe out the entire collateral for a trade. In practice, positions are often closed before prices reach these levels.

Exchanges do not wait for collateral to be exhausted; they require traders to maintain a minimum asset-to-liability ratio and close positions preemptively. The approach to liquidation is influenced by accumulated interest on the loan and fees for forced order execution.

For instance, Kraken charges 2% of the position size. Such orders do not go through the order book, so standard trading fees do not apply.

The timing of forced liquidation ahead of the theoretical level depends on the specific exchange's requirements and accumulated borrowing costs. While prices are far from the liquidation threshold, the difference between the calculated and actual exit points remains unnoticeable. However, during the notorious "Black Saturday," over 1.6 million traders crossed the Rubicon.

In contrast, there is no such threshold in spot trading. As an asset loses value, the account may show losses, yet no one forcibly sells assets at a loss.

This represents the primary difference between the two modes. In leveraged trading, the timing of loss realization is dictated by the exchange's risk mechanism, not the trader. In spot trading, "paper" losses can be weathered while waiting for the asset's price to recover.

Forced liquidation is immediately visible: a notification, a vanished position, and collateral wiped out in seconds. However, borrowed funds gradually erode capital in another way — slowly and steadily, as the trade remains open.

Volatility Takes Its Toll

Let's consider two hypothetical trading sessions. In the first scenario, an asset declines by 10%, then rises by 11.1% the following day, returning to its original price.

A spot buyer retains the same amount: their $100 drops to $90, then back to $100.

With threefold leverage, the arithmetic changes. On the first day, the position loses not 10% but 30% — down to $70 from $100. The next day, it gains 33.3%, proportional to the leverage. However, from $70, it only amounts to $93.3.

The price has returned to its original level, but the account does not: the "paper" loss stands at 6.7%.

With fivefold leverage, after similar fluctuations, the remaining amount is $77.8 (−22.2%). This model does not account for fees or potential liquidations.

The cause of the negative outcome is simple: percentage changes are always calculated from the current amount, not the original. In an open long position, a decline in the asset's price reduces the account value, and subsequent gains are calculated based on the diminished balance. The higher the leverage, the deeper the potential drawdown and the smaller the base for recovery.

The gap between asset growth and the account outcome is calculated using the formula:

r ≈ μ − σ²/2

where μ is the average return of the asset, and σ² is the variance.

Without price fluctuations, capital would grow linearly, reflecting average returns daily. Volatility pulls the actual result below this scenario, and the larger the σ², the wider the gap.

Leverage impacts both parts of the formula differently. The first grows proportionally to the position size, while the second increases quadratically: a trade twice as large loses four times more on "swings".

This leads to a pattern: the more turbulent the market and the longer a leveraged position is held, the more it trails behind spot purchases.

From February 4 to March 4, 2026, Bitcoin moved from $73,095 through a drop to $63,495 and back to $72,667 — a total change of just 0.6%. The chart compares the capital of three hypothetical positions of $100 during this period: without leverage, with 3x leverage, and with 5x leverage, evaluated daily. Source: ForkLog/Coin Metrics.

In a steady upward trend, the results differ significantly. If prices move in one direction without deep retracements, leverage can potentially yield multiplied profits.

Thus, the success of margin trading heavily depends on the smoothness and predictability of price movements, which is rare in the crypto market. In February 2026, Bitcoin's annualized volatility hovered around 81% — roughly double the average for the previous 12 months.

It's hard to quantify how much traders actually lose due to the aforementioned effects.

Cost of Time

In spot purchases, a small fee is charged only during the buying and selling process. With borrowed funds, the situation differs: charges accumulate continuously while the position remains open, and price direction does not affect them.

In margin trading, the exchange lends money to the client at interest. When opening a long position, the trader acquires funds to purchase the asset. For a short position, they need the financial instrument itself — it is sold to later buy back at a lower price.

The interest rate depends on demand, available supply, and the type of borrowed asset. It is calculated hourly on the loan principal.

With Binance, interest begins accruing immediately after the loan is issued. The first payment is calculated proportionally — based on the actual number of seconds from the loan issuance to the nearest hour's calculation. After that, it accrues by full hours until the debt is settled.

Kraken uses a different approach. Instead of hourly accrual, the platform charges a fee for opening a position and then a commission for carrying it every four hours. The rate is fixed at the moment of order execution and is visible directly in the order form, so the total cost is known in advance.

Most platforms have floating rates that depend on market conditions and specific trading pairs. Aggregate data is published by aggregators like CoinGlass, which clearly shows the spread between exchanges.

Perpetual futures have no interest on loans — instead, a funding rate is implemented. This involves payments exchanged between market participants: when there are more long positions, they pay shorts, and vice versa. Calculations occur every eight hours, or three times a day.

The base component of the formula on Binance is fixed at 0.03% per day — 0.01% for each calculation, while for certain contracts like ETHBTC, it is zeroed out. In addition, a variable component is added based on the gap between the contract price and the spot quote.

When one side of the market heavily outweighs the other, the rate hits a limit set by the exchange. The accumulated amount over a week or month can also be viewed on CoinGlass.

Comparison of position holding costs across different trading types. Source: ForkLog.

A unique feature of Coin-M contracts is that the collateral is not a stablecoin, but the traded crypto asset itself. When using such instruments, a price drop affects the position in two ways: losses increase concurrently with the decline in the value of the collateral.

The fundamental difference between the discussed modes boils down to whether a trader pays for an open position. In spot trading — no. In margin and futures trading — yes, regardless of the trade outcome. This is why some exchanges recommend using borrowed funds only for short-term operations.

The extent of these costs depends on the duration of the trade. For day trading, interest on loans or funding rates are unlikely to significantly impact the outcome. However, a position held for a month or more may lose a substantial portion of capital even before the price moves in the desired direction.

Comparative table of various trading modes. Source: ForkLog. 

Hunting for "Stops"

On that "Black Saturday," limit orders on Binance were triggered, some of which had been open for years. Due to insufficient liquidity, prices dropped to levels that had long seemed unreachable.

A massive spike on the hourly chart of BTC/USDT on Binance in early October 2025. Source: TradingView.

The ATOM token from Cosmos fell below $0.01 at one point, while the USDe stablecoin from Ethena dropped below $0.66. The exchange explained the anomaly by pointing to one-sided liquidity, long-standing orders, and a display error.

This is what the spike looks like on the chart:

Hourly chart of USDe/USDT on Binance (early October 2025). Source: TradingView.

Such volatility spikes usually occur at specific points, and the mechanics are relatively straightforward. Stop orders from retail traders are often clustered around round numbers, local minima, and support lines. These accumulations are visible in the order book, and with low liquidity, pushing prices down to the nearest cluster can be achieved with a relatively small volume.

Next comes a chain reaction. Each closed position puts new supply into the market, driving prices down and triggering the next layer of orders.

Collateral adds another layer of complexity. In early October, "wrapped" tokens significantly deviated from their calculated values: BNSOL dropped from about $300 to $35, while WBETH traded at $430 when Ethereum was above $3,800. Positions were closed at distorted prices, leading Binance to pay $283 million to affected users.

When there are few buy orders and many sellers, prices drop to the nearest levels where demand exists. Clusters of stop orders become part of this cycle and amplify it.

For margin traders in such circumstances, the outcome is always a loss realization. A spot position, on the other hand, experiences this situation differently: there is no one to sell it off forcibly, and a pre-set limit order executes at the desired price during sharp movements.

What Spot Trading Does Not Eliminate

In the ten months following the all-time high (ATH), Bitcoin lost almost half its value. The price peak occurred on October 6, 2025, at $126,080, according to CoinGecko. By mid-August 2026, the asset had depreciated by about half.

Bitcoin price dynamics from ATH reached in early October 2025. Source: ForkLog/CoinGecko.

A spot trader who bought near the ATH did not face either a margin call or a "hungry" funding rate. The acquired volume of the asset remains the same, even though it is now worth half.

The absence of forced liquidation reduces risks but does not guarantee capital recovery — it can take years to return to previous levels.

Overall, historical data on spot operations paints a grim picture. The Bank for International Settlements (BIS) studied retail investor outcomes in 95 countries from 2015 to 2022 — between 73% and 81% ended up in the red. New user influxes coincided with price increases.

In a separate BIS bulletin, a curious pattern was noted. Following the collapse of the Terra ecosystem and the FTX exchange, Bitcoin whales rushed to offload coins onto the market, while addresses holding between 1 and 1,000 BTC bought them at a "discount." The median retail investor lost about half their investment by December 2022.

Spot trading also has structural limitations. A transaction requires full coverage, so profit volume is directly linked to deposit size: to make $1,000, one must invest around $10,000 and wait for a 10% increase. Earning from price declines "from the start" is not possible without leverage or derivative instruments, where the calculation is based on price differences, and the asset does not transfer to the buyer.

Counterparty risk remains as well. Unless dealing with decentralized exchanges (DEX), coins are held on a centralized platform rather than a "cold" wallet. As shown by the history of FTX, in cases of platform insolvency, ownership of the asset becomes a claim against the bankrupt entity.

A "paper" loss retains its status only until the funds are needed in real life. Selling at an unfavorable rate for urgent expenses is essentially no different from forced liquidation. The only difference is that the asset holder makes the decision themselves.

While spot trading eliminates the risk of "getting kicked out" of a position, it does not protect against poor asset selection or an unfavorable entry point.

Stubborn Statistics

American retail traders using maximum leverage in the forex market averaged a loss of -44% per month until 2010. This estimate was provided by economists Roli Haimar from Boston College and Alp Simsek from MIT.

That same year, U.S. regulators capped leverage for retail clients. Researchers assessed the consequences using three independent datasets, comparing changes in affected and unaffected groups before and after regulatory changes.

They found that trader turnover decreased by 23%, losses for the most aggressive participants fell by 40%, and brokers’ operating capital decreased by 25%. Importantly, spreads did not widen, indicating that liquidity was unaffected.

Changes in trading results among American financial market participants. Source: Journal of Financial Economics.

European regulators studied the trading of contracts for difference in several jurisdictions and discovered that between 74% and 89% of retail accounts were unprofitable. Average losses per client ranged from €1,600 to €29,000. Following this analysis, on March 23, 2018, ESMA restricted leverage and banned binary options.

A particularly telling study was conducted in Brazil. Economists Fernando Chague, Rodrigo De-Losso, and Bruno Giovanetti analyzed trading results from 19,646 individuals who began trading mini-futures on Ibovespa between 2013 and 2015.

Only 1,551 people lasted more than 300 trading days. Of those, 97% finished the period in the red after accounting for costs. Only 1.1% earned more than the minimum wage in the country, and just 0.5% exceeded the starting salary of a bank operations clerk.

Source: Chague, De-Losso, Giovannetti; ESMA; Bank for International Settlements.

One significant finding relates to trading experience: results did not improve over time. The longer individuals remained in the game, the larger their accumulated losses became. No signs were found that practice translates into skill.

However, leverage is not the only way to incur losses in trading. Financial analysts Brad Barber from the University of California, Davis, and Terrence Odean from Berkeley examined accounts from 66,465 American households between 1991 and 1996. They found that the most active traders earned 11.4% annually, compared to an average of 17.9%. After accounting for fees, the returns of both groups were almost identical — the negative difference was attributed to transactional costs and trading frequency.

One conclusion emerges: fees and turnover gradually erode accounts, leverage accelerates the depletion of deposits, and learning from one's mistakes does not work as well as commonly believed.

Systems Over Intuition

Consider a simple scenario. A trader buys an asset at the lower boundary of a range and immediately places a sell order at the upper boundary. After that, they close the trading terminal and attend to other matters.

The decision is made once — in advance. From that point on, the exchange takes over.

This approach alleviates the principal problem of manually closing trades. Suppose prices surge during the night and approach the designated level. The trader sees a strong impulse, cancels the order, and hopes to sell at a higher price. The momentum fades, the price returns, and the trader ends up selling at a less favorable rate.

A pre-placed order functions differently — in favorable market conditions, it executes at the designated level while the account owner sleeps.

The reverse effect also works. A temporary price spike lasts seconds, making it nearly impossible to catch manually, but a limit order placed in advance executes automatically.

This example illustrates a basic trading system. Arkham identifies several types of systems, where the difference lies mainly in who makes the decision at the moment of the trade:

  • automation executes pre-defined conditions: stop orders, limit orders, or regular purchases on a schedule (DCA). The rule remains unchanged until the trader rewrites it;
  • algorithms are more complex. They receive data streams and make decisions based on the situation — responding to volume, price movement speed, and spread width. The trader still sets the logic, but the timing is determined by the program;
  • bots handle tasks that are unattainable in terms of speed and duration. The digital asset market operates 24/7, and certain strategies measure intervals between operations in seconds;
  • copy trading involves delegating decisions to another participant, whose trades are automatically replicated in the account. Only public performance statistics are available. The logic of entries and exits, leverage size, and allowable drawdown remain with the leader;
  • quantitative approaches begin with data. Patterns are identified through historical data analysis and then tested for statistical significance.

The last two types share a common vulnerability hidden in backtesting.

Testing a strategy on past quotes appears convincing: results are measurable, drawdowns are visible, and returns are calculated. The catch is that obtaining "pretty" results from historical data analysis is much easier than developing a working strategy.

Market strategist and Investopedia author James Chen lists typical pitfalls of such testing. The first is sample composition. If only companies or assets that have survived to the present are included, returns appear artificially inflated: projects that have exited the market are excluded from consideration.

The second pitfall is less obvious. If the same data set is run through a hundred strategies, it’s possible to randomly obtain several profitable variants. Distinguishing luck from patterns based on a single test is impossible.

Thus, a standard requirement is that a strategy is built on one set of data and tested on another. Matching results across both samples serves as a sign that the identified rule is effective.

A third procedure is forward testing, where the system is launched in a live market without real money. Here, honest accounting is crucial — every trade, including losing ones, must be recorded.

Technically, all these issues are described by the term "overfitting." A model memorizes the specifics of the past dataset instead of general patterns. At IBM, a characteristic of overfitting is described as high accuracy on training data with poor results on new data.

To combat this issue, several methods can be employed:

  • the sample is divided into parts and the model is run sequentially on each;
  • training is halted before the system begins to memorize details;
  • the model itself is simplified by removing unnecessary parameters.

None of these approaches negate the costs of using borrowed funds or the disproportionate nature of drawdowns and subsequent recoveries. Systems provide something different — an analyzable history of decision-making and an understanding of under what conditions a rule ceases to work.

Comparison of various trading modes. Source: ForkLog.

Despite the heightened risks, the vast majority of crypto market participants prefer leveraged instruments. By April 2026, spot trading volumes dropped to their lowest since November 2023, while the share of derivatives in total trading volume reached 77.1%.

Source: CoinDesk Research.

***

Spot and leverage serve different purposes. The former limits profit to the deposit size, while the latter restricts the life span of the position. Traders must decide which is more acceptable: a cap on returns or the risk of an early exit from a trade.

Spot trading is not only suitable for beginners. It also attracts conservative market participants for whom a peaceful sleep and "semi-passive" income are more important than rapid capital growth.

The deleveraging in October and the subsequent grueling recession tested both approaches' resilience. The result was that leveraged positions worth $19 billion vanished in less than a day.

The reasons for the clear tilt towards derivatives are largely explained by market participants' psychology. Many newcomers enter trading platforms with inflated assessments of their capabilities. Leverage in this context is seen not as a risk but as an accelerant of "almost guaranteed" success.

When the outcome turns out to be the opposite, explanations are often sought in external factors. Responsibility is shifted onto "devious" whales and "omnipotent" market makers, and the events are described as a conspiracy.

Experience is accumulated only by those who acknowledge their own miscalculations.