Education Pillar IV · Vendor Credibility Risk of Ruin Calculations
Pillar IV · Vendor Credibility · False Assurance

Risk of ruin calculations.

The formula is correct. The question is whether the system being evaluated satisfies the conditions the formula requires. Four assumptions must hold. When they do not, a near-zero risk figure can mask structural exposure approaching 100%.

In this article
  • What risk of ruin calculates and its three required inputs.
  • Four assumptions the formula requires — and how each maps to a specific risk type.
  • The maximum distortion case: 0.001% reported probability vs. near-100% structural risk.
  • Three replacement questions that extract the information risk of ruin would provide.
  • How the Institute applies this as a false assurance mechanism within vendor credibility.

Risk of ruin is a legitimate quantitative tool. The standard formula calculates the probability of losing a specified percentage of capital given a defined win rate, average win size, and average loss size. When its underlying assumptions hold, the output provides genuine insight into sustainability. The issue is not with the mathematics. It is with the conditions under which the mathematics is being applied.

§ 01

What risk of ruin calculates.

D
Risk of ruin — the standard model
The standard model takes three inputs: the probability of a winning trade, the average size of a win, and the average size of a loss. From these, it calculates the probability that a series of trades will produce a cumulative loss exceeding a defined threshold — typically some percentage of total capital. The calculation assumes independent events drawn from a stable distribution, bounded losses, and representative historical data.

When these conditions are satisfied, the formula provides a mathematically rigorous answer to the question: given this statistical profile, what is the probability that normal trading activity will reduce the account below a survivable level? The formula is correct. The question for investors is whether the system being evaluated satisfies the conditions the formula requires.

The 80% violation rate
Across the Institute's vendor credibility assessment, approximately 80% or more of risk of ruin figures presented in retail algorithmic trading marketing are calculated on systems where at least one assumption is violated. The result is not a small discrepancy. In specific architectural configurations, a formula may report 0.001% probability of ruin while the actual structural risk approaches 100%.
§ 02

The four assumptions and how each breaks.

Each assumption maps to a specific risk type in the Institute's framework. Understanding which assumption each risk type violates explains why the formula's output becomes unreliable for systems exhibiting those structural characteristics.

ASSUMPTION 01
Independent outcomes
Each trade outcome must be statistically independent. The result of trade five must be unrelated to trade four. Violated by systems where each new position is opened because the previous is losing — the architecture makes them a single escalating exposure.
VIOLATED BY: Warehoused risk
ASSUMPTION 02
Stable distributions
Statistical properties must remain constant over time. The win rate observed historically must represent ongoing system characteristics. Violated when Phase 1 data is presented as the full lifecycle — a snapshot of a temporary condition treated as permanent.
VIOLATED BY: Latent risk
ASSUMPTION 03
Bounded losses
Each individual loss must be bounded to a known maximum. The most consequential assumption failure. Systems holding losing positions with no defined exit have unbounded losses — the formula's maximum loss assumption and the system's reality are disconnected.
VIOLATED BY: Warehoused risk
ASSUMPTION 04
Reliable historical data
Historical data must be representative and sufficient. Overfit backtests produce win rates that do not represent future performance. Insufficient sample sizes produce artifacts of a specific market period. The formula cannot assess data reliability.
VIOLATED BY: Inherited risk
1. Independent outcomes 2. Stable distributions 3. Bounded losses 4. Reliable data WAREHOUSED RISK Violates A1 + A3 LATENT RISK Violates A2 INHERITED RISK Violates A4 FORMULA ASSUMPTIONS → RISK TYPE VIOLATIONS
Fig. 01
Assumption-violation mapping. Each of the formula's four assumptions maps to a specific risk type in the Institute's framework. Warehoused risk violates two assumptions simultaneously (independent outcomes and bounded losses), making it the most consequential violation category. When any single assumption fails, the formula's output shifts from informative to misleading.
Maximum distortion
The distortion reaches its maximum when losses are unbounded: a formula designed to quantify the probability of account destruction produces a near-zero output while the system holds open positions representing a substantial fraction of total capital. The number creates confidence in precisely the situation where caution is most warranted.
§ 03

Three better questions.

When the four assumptions cannot be confirmed, the risk of ruin formula's output cannot be trusted. In those cases, three direct questions extract the information that the formula would provide if its conditions were met.

1
What is the maximum position size the system takes?
Addresses Assumption 3 directly. If the system can hold positions representing 20%+ of account capital, no formula assuming bounded 2% losses is applicable.
2
What is the largest unrealized loss the system has held?
Reveals actual downside exposure as opposed to the closed-trade drawdown the balance curve reports. A system that has held an unrealized loss of 35% has demonstrated that capacity regardless of what the formula calculates.
3
What is the worst-case single loss, realized or unrealized?
Establishes the empirical bound on individual trade losses. If the worst single loss is 28% of capital, the formula's 2% assumption is not conservative — it is a different system.
§ 04

How the Institute applies this.

Risk of ruin appears in the Institute's vendor credibility assessment as one of several metrics within the broader category of false assurance mechanisms. A false assurance mechanism is a metric that is technically correct in its calculation but structurally incomplete in its analytical scope, producing confidence that the underlying evidence does not support.

The Institute does not dismiss risk of ruin. When a system's architecture satisfies all four assumptions, the metric is included as a legitimate data point. The analytical question is whether the assumptions hold for the specific system being assessed. When a vendor presents risk of ruin as evidence of system safety without addressing these questions, the metric functions as a credibility mechanism rather than a risk assessment.

§ 05

What this means for investors.

Risk of ruin is a conditional tool. Its value depends entirely on whether the system it measures satisfies the mathematical conditions the formula requires. An investor who can confirm independent, bounded trades from a stable distribution with sufficient data has a useful instrument. An investor who cannot confirm those conditions has a number that may be worse than no number at all, because it creates false confidence in a system whose actual risk profile is unknown.

The three replacement questions provide a practical entry point. They do not require the investor to evaluate the mathematical formula. They require the vendor to provide the empirical information that determines whether the formula is applicable. A vendor who can answer all three with specific figures is providing transparency that supports further analysis. A vendor who cannot is presenting a formula without the foundation it requires.

This connects to the broader Evaluation Framework: every metric requires structural context before it carries evidentiary weight. A risk of ruin figure, like a win rate, a Sharpe ratio, or a verified track record, becomes analytically meaningful only when the conditions under which it was generated are understood.

Without structural context, precision in the calculation creates an illusion of precision in the assessment.
§ 06

Frequently asked.

QWhat is risk of ruin in algorithmic trading?

Risk of ruin is a mathematical formula that calculates the probability of losing a specified percentage of trading capital given a system's win rate, average win, and average loss. When its four underlying assumptions are met — independent outcomes, stable distributions, bounded losses, and reliable historical data — the formula provides genuine quantitative insight into sustainability.

QWhy do risk of ruin calculations often understate actual risk?

The formula requires four assumptions frequently violated by retail algorithmic systems. Warehoused risk violates the assumptions of independent outcomes and bounded losses. Latent risk violates stable distributions. Inherited risk violates reliable data. The most consequential failure occurs when losses are unbounded: the formula may report near-zero probability while the system holds positions representing a substantial fraction of capital.

QWhat questions should replace risk of ruin when evaluating algo systems?

Three questions extract the information risk of ruin would provide if conditions were met. First, what is the maximum position size? Second, what is the largest unrealized loss the system has held? Third, what is the worst-case single loss, realized or unrealized? These reveal actual exposure without relying on a formula whose preconditions may not apply.

Cite this article
The Algo Institute. (2026). Risk of ruin calculations — when the math breaks down. The Institute's Evaluation Framework, Pillar IV, False Assurance. FILE VC-043-26. Methodology v3.1.