Capital withdrawn is different from trading profit

A high fraction of deposited capital being withdrawn describes capital movement. Trading profitability requires a separate outcome ledger. The continuous record companion reports 4,984 ETH deposited into DX Terminal Pro and 92% withdrawn. We should preserve the withdrawal statement as an accounting observation and avoid turning it into an aggregate return.

Applied mechanically to the rounded figures, 92% of 4,984 is approximately 4,585 ETH, leaving approximately 399 ETH outside that withdrawn share. Those balances are derived approximations from rounded published inputs. Allocating the remainder among cash, token inventory and other accounting categories requires a closing valuation ledger absent from the public summary.

A cash-flow identity makes the missing fields visible

For an illustrative account, define deposited capital as D, withdrawals as W and the final marked portfolio value as V. With no other external transfers, a simple net economic change is W plus V minus D, measured in the same currency. If D is 100 ETH and W is 92 ETH, a final value of 8 ETH gives zero change. A final value of 3 ETH gives a loss of 5 ETH; a final value of 18 ETH gives a gain of 10 ETH. The withdrawal fraction remains 92% in every case.

This fixture leaves time-weighted and money-weighted return calculations for a separate question. Their inputs include the timing of deposits and withdrawals. A cumulative withdrawal share lacks that timing and cannot establish an annualized return, a typical owner's experience or a distribution of agent outcomes.

Our audit artifact would be a cash-flow reconciliation with opening balances, external deposits, withdrawals, transfers, final inventory and valuation timestamps. It would reconcile each asset before conversion to a common unit. Trades within a vault move assets and incur costs; they differ from owner capital entering or leaving the measured population. The proposed artifact makes that boundary explicit without inventing the missing historical ledger.

Keep approximate deployment summaries together

The controls paper companion describes more than 5,000 ETH deployed in its scale summary, while the continuous record companion gives 4,984 ETH deposited in its cross-system table. Those labels and rounding differ. We cannot force them into an exact reconciliation by assuming deployed and deposited are identical measures.

The setting was a 21-day real-capital event from February 26 to March 18, 2026, with twelve tokens on Base and a frozen production runtime on one model family. Withdrawal rights and closure controls are part of its operational description. Access to those controls can help explain capital movement. The reason for each withdrawal requires its own evidence.

A reader looking for profit should ask for cash-flow-adjusted valuation and its population. A reader looking for capital recovery should ask for withdrawal mechanics and final reconciled state. The same dataset may eventually support both questions; a single rounded recovery fraction cannot do the work of both.