Drawdown Percentiles · The Psychology It Replaces
A definition of normal pain, signed before the pain.
Prohibitions
8
Decided
IN ADVANCE
Layer 01 — The artifact and its timestamp
Mid-drawdown, every instinct argues this time is different.
The psychological force of the bands comes almost entirely from when they were produced. They were generated in calm, from recorded evidence, before the current stretch began. In the middle of a losing run, the operator has no reliable access to their own prior judgement, and every available reading is filtered through the run in progress. The pre-committed definition is the one artifact in the room that provably was not produced by the current emotional weather.
Layer 02 — Two failures, one instrument
The bands prevent panic and denial with the same reading.
Without a defined range, an operator commits one of two errors and usually both across a year: de-risking inside ordinary variance because it felt severe, and holding through genuinely abnormal damage because the last few drawdowns also felt severe and recovered. A distribution answers both with the same number. This depth is common. This depth is not. Neither answer is available to intuition, because intuition has no access to the frequency of events it experienced once.
How MARS uses this
The pre-agreed definition of normal pain gets its force from this map: most simulated drawdowns cluster shallow and fast, and the frightening deep-slow events are visibly rare. When a live drawdown feels bad, the map answers with frequency — this territory is common, this recovery time is ordinary — before feelings get a vote.
How it benefits you
Drawdown stops being one scary number. A 7% dip reads as the common, quick-recovery event the map says it is, while the non-linear duration curve makes the case for gate compression viscerally: deep holes do not just cost more - they cost longer.
Normal pain, mapped: the shallow-fast cluster that dominates, and the rare deep-slow tail that feelings mistake for the norm.
Layer 03 — Streaks were shown in advance
The cluster you are in was modelled before you entered it.
Loss and win clustering, shock-week injection, worst cycle and worst week damage, and losing-streak distributions are all modelled explicitly rather than assumed away. That matters at exactly the moment it is hardest to believe: the fifth consecutive losing cycle feels like evidence the system has stopped working, and the streak distribution is the record showing how often paths that eventually completed passed through precisely that stretch.
- Streak damage and worst-cluster behaviour are simulated, not treated as tail noise.
- Regime shocks model hit-rate decay, volatility expansion, and drawdown cascades together.
- A losing stretch inside the bands is variance; a breach of the adverse bands outranks any green dashboard.
Layer 04 — The one rule with no hedge
Below benchmark with worse drawdown is a containment problem.
The standard permits nuance almost everywhere and refuses it here. When both readings are adverse, the answer is diagnosis, containment, and a model-to-live mismatch review, and it is never additional risk. Stated plainly and in advance, the rule is unremarkable. Its whole value is that it was written down at a moment when nobody needed it, so it is available at the moment when everybody would rather it were not.
Reference
The eight standing prohibitions, and the impulse each one blocks
| Prohibition | The impulse it blocks |
|---|---|
| No revenge or catch-up trading against the benchmark | Treating a band position as a debt to be repaid this week |
| No partial period compared to a full-period band unlabelled | Manufacturing an alarming gap out of an incomplete month |
| No judging the system by equity alone | Letting one flattering row overrule five unflattering ones |
| No treating median hit week as a deadline | Racing a statistic computed only from paths that survived |
| No overriding the throttle because live is below benchmark | Answering underperformance with exposure |
| No ignoring lock risk | Reading a high-performance model as a safe one |
| No changing benchmark assumptions without versioning | Quietly re-marking the ruler until this month looks acceptable |
| No promoting sandbox experiments into benchmark comparisons | Comparing live results against a model nobody has certified |
The governing idea
Connected inside MARS
This module doesn't work alone.
Go deeper
Operator briefs on this territory.
Deep dive — 01
What the adverse tail is telling you to prepare for.
The tail is what a healthy system does in its unluckiest futures. Size for it now or meet it unprepared.
Read the full brief →
Deep dive — 02
The bands' authority comes from when they were made.
The one artifact in the room that provably was not produced by the current emotional weather.
Read the full brief →
Deep dive — 03
Losing streaks have a distribution, and you were given it.
The worst streak in the distribution belongs to a healthy system. Meeting it is not information.
Read the full brief →
Every module ships in the complete MARS package.
One price. Eleven workbooks, three TradingView indicators, and the full manual library — $497.