Responding to Drawdowns: Reduce Risk, Pause, and Define Recovery

An account rises from RMB 1 million to RMB 1.2 million, then falls to RMB 1.08 million. You still have an RMB 80,000 gain, but are RMB 120,000 below the peak. One person says there is no reason to act before losing principal; another says a 10% decline requires selling everything. Which has the evidence?

The number alone does not decide. The same drawdown could reflect permitted market movement, a duplicate order, or a research assumption that no longer holds. Diagnose before reviewing, reducing, or pausing. Adding money cannot bypass that diagnosis.

State both the loss measure and its base

You will reconcile loss and drawdown denominators, choose a response state for different causes, and prepare a restoration request with evidence, versions, and recheck conditions. No stop percentage here is proposed for every strategy.

C8-80 is an independent unlevered synthetic account, with monetary amounts in units of RMB 10,000. It begins at 100 at the end of October 2026. Subsequent observations are the ends of November, December, and January 2027. Returns include all this example's costs. There are no distributions, taxes, or external flows until the contribution explicitly introduced later. This account does not change Household A's conditions or establish that its approved allowance is RMB 1 million.

November equity is 120; December equity is 108. Relative to the original 100, investment gain is 8, or 8%. Relative to the 120 peak, the decline is 12 / 120, or 10%. Both statements are correct and answer different questions. Changed starting capital, windows, or flows require new calculations.

Drawdown compares a current comparable account-value measure with its previous peak. Time matters too: December is one month-end observation interval after November's peak, without recovery. These endpoints establish observed month-end drawdown, not the maximum decline within the month.

Register the response before the drawdown

For this separate example, register the following policy at the start. Normally inspect month-end values. An observed drawdown reaching 8% enters review. Reaching 12% pauses new risk and requires a disposition request to reduce the risky-asset target to at most 50% of current account equity. A system failure, unexplained holding, or costs outside the prior stress range blocks new commitments for investigation immediately, without waiting for month-end.

These are teaching inputs, not proven optimal thresholds. Preserve the original policy, trigger times, and reasons for revisions. Changing 12% to 20% when the account is down 11.9% does not show that the original rule worked. A monthly strategy still uses its frozen signal and next permitted execution time; this funding policy does not silently add an intraday price stop.

A reduced target is not an accomplished sale. The request must specify timing, quantity, price or other constraints, fee headroom, and treatment of an unfilled order. A policy that changes the original strategy's behavior needs a separate funding-policy version and an implementation-impact check. Do not merge its later returns into the unchanged strategy history.

The same 10% starts three different investigations

Permitted movement. Original prices, corporate actions, signals, orders, and fees reconcile; actual holdings match the frozen rule; household cash needs and the stress table are unchanged. Ten percent triggers the 8% review but not the 12% sizing action. Record “under review; original trading rule unchanged,” investigate, and follow the predetermined observation policy. Neither emotion nor the drawdown alone requires an automatic full sale.

Operating failure. The apparent decline is also 10%, but an order filled twice or a quote is stale. Even the reported drawdown may be unreliable. Block new commitments, reconcile actual positions and live orders, and repair the ledger. Do not wait for profits to return or classify the entire difference as ordinary market movement.

Changed assumptions. Execution is correct, but an essential market rule, data definition, or trading condition changed. Perhaps the original research assumed available liquidity that is persistently absent. Reassess the hypothesis; another run of the same program is not repair evidence. A new version needs relevant research, while the old version retains the reason it ceased to apply.

These conditions can coexist. Establish trustworthy holdings and information before discussing an exact allocation. Earlier household payment needs can also invalidate the funding plan even when the model remains intact; the original observation period is not permission to ignore them.

Make review, reduction, and pause distinct states

Review adds checks and records unresolved questions. Reduction changes an approved target and may produce new exposure after execution. Pausing new commitments prevents increases. Stopping a version also needs explicit treatment of existing positions and open orders. They are different states, not one “high risk” button.

Take a separate approved-disposition snapshot, not an assumed action already taken on the original path. Equity is 108: risky assets 80 and unrestricted cash 28. If the approved cap is 50% of equity, the zero-fee risky target is 54, requiring a sale of 26. With unchanged prices, a full fill, and no fee, assets 54 plus cash 54 still total 108.

Now charge 0.1% of the amount sold. Selling 26 incurs a fee of 0.026 and leaves equity of 107.974. Risky assets of 54 are now slightly more than half the account. Ignoring the fee cannot establish strict compliance with a post-fee cap.

For a strictly satisfied 50% cap, let the sale amount be x. Solve 80 − x = 0.5 × (108 − 0.001x), giving x = 26 / 0.9995 ≈ 26.013007. Remaining risky assets are approximately 53.986993, and equity is approximately 107.973987. Actual quantity increments need rounding in the direction that satisfies the constraint, followed by another check; displayed rounding must not erase a violation.

If the asset is halted or a limit order cannot fill, record “reduced target approved; order incomplete; existing risk remains.” Keep the actual 80 or the position after genuine partial fills. A target of 54 is not proof of a holding of 54, and pausing new commitments does not remove the old exposure.

Costs can trigger another specific action. In a separate policy example, prior stress work considered total cost up to 15 basis points per execution, while the current estimate is 30 basis points. Pause new commitments, verify quotes and cost definitions, and preserve the old version: its evidence does not directly cover the change. Whether to reduce existing risk still depends on execution conditions; do not submit unconstrained orders merely to make the screen display less risk. These are synthetic cost inputs, not a broker's current prices.

A deposit lifts the balance without repairing drawdown

Return to December's original equity of 108. After valuation, a confirmed outside contribution of 20 lifts the balance to 128. A raw comparison with the old peak of 120 would show a new high, although the strategy earned nothing at that instant.

Use internal accounting units to keep a comparable value path. Initially, 100 monetary units correspond to 100 accounting units at a value of 1. November's unit value is 1.2 and December's is 1.08. The contribution at 1.08 adds approximately 18.518519 accounting units, taking the total to approximately 118.518519. Unit value remains 1.08, and drawdown remains 10%. These are internal records, not actual fund shares.

Valuing at external-flow boundaries and distinguishing client money from investment results is a basic performance-calculation idea. A historical CFA Institute calculation-methodology guide explains that principle.1 We use the concept, without claiming this example meets all current GIPS requirements. G93 handles fuller return measurement.

Suppose the entire 128 then earns 4% net in January, without another flow. Ending equity is 133.12 and unit value rises from 1.08 to 1.1232. Relative to the 1.2 unit-value peak, drawdown is still 6.4%. Two observation intervals have elapsed since November's peak, without recovery. A balance above 120 does not change that conclusion.

Maintain both records: actual equity determines current funding and risk amounts; a comparable value measure adjusted for external flows supports performance and drawdown analysis. New funding can change future constraints. It cannot mark a duplicate order, excessive cost, or broken assumption as resolved.

Restoration needs evidence tied to the original cause

A restoration request should identify the pause reason and time; rule and funding-policy versions; original and corrected records; evidence addressing the cause; recalculated costs, liquidity, and household/account budgets; the proposed scope; and conditions for stopping again. A legitimate decision may allow limited observation, continue the pause, or retire the old version.

Duplicate orders call for identifiers, fill reconciliation, and verified duplicate-submission protection. Costs outside the tested range call for updated execution and fee evidence. A changed hypothesis calls for newly registered research, including unfavorable results. G52 addresses the operating evidence; G57 addresses behavioral plans. Here, determine whether the materials support a funding decision.

“It rose again for two weeks,” “the model says it is fixed,” and “the deposit restored the balance” do not answer those causes. SEC performance guidance asks readers to examine methods, fees, and selected presentation periods rather than treating past performance as a guarantee.2 The reason for restoration must be inspectable.

Decide the next state yourself

Explain. Why can neither “still up 8” nor “down 10%” alone justify selling everything from the 120-to-108 path?

Save a response table and restoration request

Keep starting equity, comparable peak, observation dates, flow times, monetary profit/loss, comparable drawdown, trigger-policy version, current cause, approved targets, actual positions, incomplete orders, missing evidence, review date, and restoration decision. Preserve the original and withdrawal branches separately.

Check that a pause was not shown as completed liquidation, approved targets were not substituted for fills, and deposits neither reset the old peak nor cleared a failure. Every restoration reason should address the pause cause. Mark missing evidence as unverified rather than asking AI to invent a plausible fill.

First verify the numbers and timeline, then diagnose, then change commitments and execution states under the policy. A rebound can improve unit value without repairing a failure. A contribution can improve the balance without establishing that more risk is justified. Keeping original and corrected versions together makes those distinctions visible.

References

1. CFA Institute: Guidance Statement on Calculation Methodology, adopted September 28, 2010, effective January 1, 2011; printed pages 1–2. Historical guidance on external cash flows and valuation, not a complete 2026 compliance instruction.
2. Investor.gov: Investor Bulletin—Performance Claims, September 15, 2022. Fees, methodology, presentation periods, and limits of past performance.

Editorial status

AI-assisted draft synchronized to the accepted second-round Chinese source. C8-80 accounts and policies are synthetic. See the author review. A separate AI reviewer has completed an independent full-text and bilingual review. Human financial review and target-reader testing have not been performed and are not scheduled for this edition.