Paper Profit vs Operating Loss: Anatomy of PnL Architecture

Introduction

The third quarter 2026 report from Twenty One Capital — the investment fund managed by Tether — shows $1.09 billion in net profit and $4.8 million in operating loss. At first glance, this is a paradox. In practice, it reflects the same principle found in algorithmic trading systems: a bot can display profit on screen while the actual balance remains zero. Understanding this mechanism is critical: confusing paper profit with real profit is one of the main reasons trading bots show gains in backtests but blow up real accounts.

Paper profit mechanics

$1.09 billion represents unrealized appreciation of portfolio asset values. Assets are revalued at current market prices daily, but actual funds do not hit the balance until a sale occurs.

The mark-to-market mechanism works as follows: if an asset rises from $60,000 to $90,000, the fund records $30,000 in profit per unit held — even if no units were sold. This creates an illusion of profitability. When prices fall, paper profit disappears from the report instantly.

Historical data shows: unrealized profit can turn into unrealized loss within weeks during a sharp market reversal. The 2022 period, when Bitcoin — Twenty One Capital’s primary asset — dropped from $69,000 to $16,000, demonstrates the scale of this risk.

Operating loss: infrastructure costs

The $4.8 million operating loss represents actual expenses independent of price movements:

• Infrastructure and security. Private key storage, smart contract audits, real-time transaction monitoring.
• Personnel. Engineers, analysts, legal counsel, accountants.
• Regulatory compliance. Compliance, regulatory reporting, licensing.
• Operating commissions. Hosting costs, API subscriptions, audit fees.

Operating loss alongside paper profit is not anomalous for a fund in the scaling phase. It is an expected outcome, provided unrealized profit covers potential risks and exceeds operating costs over the long term.

PnL architecture in trading systems

Any algorithmic trading system must separate PnL (profit and loss) into two independent modules: realized and unrealized.

Realized PnL — profit or loss from closed positions. Funds have actually been credited to or debited from the balance. This is the equivalent of operating results in a crypto fund’s report.

Unrealized PnL — estimated profit or loss on open positions. It depends on current market prices and can vanish with a market move. This is the equivalent of Twenty One Capital’s paper profit.

The error of mixing these two categories: a false signal about system profitability. A bot can show profit in a prototype because open positions are valued at current market prices. The first anomalous move turns that into a loss.

Consider two scenarios.

Scenario 1. An algorithm opens a long BTC position. Unrealized profit stands at +$3,000. Based on this signal, the system increases position size. An hour later, the market reverses — the recorded result becomes a real loss. Module separation prevents decisions based on unrealized profit.

Scenario 2. A backtest shows positive expected value for a strategy. Upon review, the profit was calculated by treating unrealized profit as realized. In live conditions, this causes slippage and invalidates the result. A strategy that appeared profitable turns out to be losing.

Architectural solution: a separate unrealized PnL module with a consolidation threshold. Profit transfers to the realized module only upon meeting strict conditions: position closure, target price reached, stop-loss triggered.

Engineering pattern: state separation

The pattern revealed in the crypto fund report applies to trading system design: separate and control.

The accounting module must maintain two independent accounts — realized and unrealized PnL — and prevent one from influencing decisions based on the other.

This means:

• Separate database tables for each PnL type.
• Separate monitoring metrics: realized PnL displays in the trading bot statistics window, unrealized PnL in the position evaluation panel.
• Separate alert triggers: realized loss above threshold triggers system pause. Unrealized loss above threshold triggers notification only, not a shutdown.

Strategy backtests must be calculated using realized PnL only. If the statistics window shows positive expected value but unrealized profit was included in the calculation — the result will not reproduce in live trading.

The database must guarantee that realized PnL data is not overwritten during unrealized recalculation. Kill-switch (emergency system shutdown) is a separate mechanism, triggered by critical realized loss, not by open position drawdown.

Summary

The crypto fund report demonstrates a pattern familiar to anyone designing trading systems: paper profit does not equal operating results. $1.09 billion on paper and $4.8 million in operating loss — this is a consequence of mark-to-market accounting, not an anomaly. The key lesson applies to trading bot architecture: realized and unrealized PnL must be tracked in separate modules. Mixing these categories creates a false sense of profitability that vanishes at the first market reversal.

FAQ

What is the difference between paper and realized profit?

Paper (unrealized) profit is the estimated increase in value of open positions at current market price. Realized profit is actual income from closed positions, credited to the balance. Paper profit can disappear with a price change. Realized profit cannot.

Why do stablecoin issuers create investment funds instead of just holding assets in reserve?

Reserve diversification reduces dependence on a single asset. A stablecoin is pegged to fiat currency, but holders may lose confidence during crypto asset declines if the entire reserve is concentrated in one class. An investment fund allows a portion of reserves to function as capital with a separate risk architecture.

Can a fund be profitable while showing an operating loss?

Yes, if unrealized profit covers operating costs and exceeds potential risks. Practical examples show: paper profit significantly exceeds operating loss, but during a sharp market decline, the balance can shift rapidly.

How does an algorithmic trading system account for unrealized PnL?

In a correct architecture, unrealized PnL is stored in a separate accounting module that does not influence trading decisions. Trading signals are generated based on realized PnL, risk metrics, and market indicators. Unrealized PnL is used only for monitoring and evaluating the current portfolio state.

How to verify correct PnL calculation in a trading bot?

Compare the realized PnL total from the accounting module with actual balance movements. If the bot shows profit but the balance has not changed — unrealized profit was included in the calculation. A correct bot separates these values and does not use unrealized PnL for generating trading signals.