FIELD NOTES / EDITION 0116 OCTOBER 2026 · ILLUSTRATIVE RESEARCH OPERATIONS
Investment Model
Monitoring
Outcome review

Read drawdown from the full wealth path

Include the initial wealth peak and distinguish recovery arithmetic from a decision to trade.

Investment Model Monitoring · 16 October 2026 · Reference edition 1

Drawdown measures the decline from the highest wealth reached up to each point. Maximum drawdown depends on the order and sampling of returns, so a final cumulative return cannot substitute for the path.

Include the starting point

The report begins wealth at one and includes that initial value in the running peak. After each period, drawdown is current wealth divided by the running peak minus one. The most negative value is maximum drawdown. This convention records a loss immediately after inception even if no later period reaches the initial wealth level.

Consider invented returns of minus ten percent, plus five percent and plus ten percent. Wealth is 0.9, then 0.945, then 1.0395. Maximum drawdown is minus ten percent from the initial peak. Initializing the peak only after the first return would erase that first loss from the drawdown history.

Hold final wealth fixed and change the path

Sequence A has returns plus twenty percent, minus ten percent and minus ten percent. Wealth rises to 1.2, falls to 1.08, then ends at 0.972. Its largest decline from a running peak is nineteen percent. The final cumulative return is minus 2.8 percent, much smaller in magnitude than the path’s maximum drawdown.

Sequence B reorders the same returns as minus ten percent, plus twenty percent and minus ten percent. It also ends at 0.972, but its maximum drawdown is ten percent. Multiplication gives the same terminal wealth while the running peaks differ. This example explains why sorting by the correct period identity is necessary before calculating drawdown.

Separate recovery requirements from forecasts

After a twenty-percent drawdown, wealth is eighty percent of its previous peak. Returning to the peak requires a gain of one divided by 0.8 minus one, or twenty-five percent. That is a recovery identity, not a forecast of how quickly a strategy will recover or a recommendation to continue taking risk.

The engine accepts period returns strictly greater than minus one and no greater than ten. It therefore rejects a complete minus-one wipeout and does not define a subsequent recovery path from zero wealth. A report about instruments that can create liabilities beyond initial capital would require a different accounting model rather than stretching this normalized wealth convention.

Match the measurement frequency to the claim

The reference fixture uses monthly returns. A monthly maximum drawdown measures the worst decline visible at those period endpoints. It can miss a deeper intramonth decline followed by a recovery before month end. Describing it as a continuously observed or intraday maximum would overstate the information in the input.

NIST’s run-sequence guidance supports examining ordered observations rather than only aggregate summaries. Here, the ordered wealth path should remain available with the drawdown value. If a real review needs daily or transaction-level drawdown, obtain appropriate data and a consistent valuation policy; changing periods_per_year does not create finer observations.

Use a review boundary without inventing approval

The example policy can flag a drawdown at or below minus ten percent for review. Equality matters: exactly minus ten percent reaches that boundary. The resulting status says a chosen example condition occurred. It does not instruct a broker, implement a stop policy or establish that a real strategy violated an approved mandate.

A reviewer should connect the path to its window, model version, return definition and unresolved data issues. Missing aligned returns withhold the complete performance block because an omitted period could alter both peaks and troughs. Preserve the resulting evidence before considering any separate change or operational decision.

Keep in the decision record

  • Initial wealth one belongs in the peak history.
  • Equal final returns can hide different drawdown paths.
  • Monthly observations cannot establish intramonth maximum drawdown.

Source notes

  1. NIST/SEMATECH e-Handbook — Run-Sequence Plot
  2. CFA Institute — GIPS Standards Handbook for Firms, November 2020
  3. W3C — PROV-O, Recommendation, 30 April 2013

Sources inform the stated methods and governance context. All example monitoring observations are synthetic. No regulatory approval or production monitoring certification is implied.

Continue the analysis

Version 1 · Published 16 October 2026. Inspect the method · Open the monitoring report

Find a monitoring reference