Methodology

How Prop Firm Secret calculates live firm scores

Last updated: 23 April 2026

What the score actually measures

The Prop Firm Secret AI Score is a funded-stage survivability score. It does not reward the most attractive headline offer. It rewards firms that show real execution evidence, meaningful public trust, verifiable company behavior, and fewer structural red flags.

  • All main score displays use funded-stage values, not challenge-phase marketing values.
  • The live rankings page and firm detail pages now use the same server-side scoring formula.
  • The score is recalculated from structured analysis data, not copied from a stale stored field.
  • The algorithm is intentionally biased toward trader survivability, transparency, and real-world trust signals.

Live scoring formula

The current live formula is a 4-factor weighted score:

Final AI Score = Broker/Liquidity × 0.40 + People's Trust × 0.30 + Firm Behaviors × 0.20 + Red Flags × 0.10

Each factor is normalized to 0-100 before weighting. The final score is rounded and clamped to a 1-100 range.

Factor 1 — Broker / Liquidity (40%)

This is the heaviest factor because named and credible execution infrastructure is the strongest public trust signal in the prop-firm space.

  • 100: Broker-backed and clear A-Book execution.
  • 75: Broker-backed, but hybrid or virtual environment.
  • 50: Named liquidity provider or broker relationship exists, but broker-backed status is not strong enough.
  • 20: No broker or liquidity provider is named, or the environment appears virtual only.

Signals used include broker-backed flags, named liquidity providers, named broker relationship entities, A-Book vs Mixed vs B-Book classification, and supporting source quotes.

Factor 2 — People's Trust / Trustpilot (30%)

This factor only becomes eligible when review volume is large enough to be meaningful. Small review counts are stored, but do not earn trust points.

  • Minimum threshold: 500 reviews.
  • If the profile is flagged, the factor score is 0.
  • If review count is below 500, the factor score is 0.
  • If eligible: rating × 14, capped at 70 points.
  • Volume bonus: +10 for 500+, +20 for 5,000+, +30 for 10,000+.

Example: a 4.6 rating with 12,000 reviews scores 64 rating points + 30 volume bonus = 94 factor points before weighting.

Factor 3 — Firm Behaviors (20%)

This factor measures whether the business behaves like a verifiable operating company rather than a pure landing-page shell.

It is built from four subscores, each worth up to 25 points:

  • Registration evidence: full points only when legal name, registration number, and a register/evidence path can be verified.
  • Years in market: 10y = 25, 5y = 20, 3y = 15, 2y = 10, 1y = 5.
  • Registered country quality: high-trust = 25, mid-trust = 15, other named country = 5, unknown = 0.
  • Managing-country quality: based on Facebook admin countries or equivalent public managing-country evidence, scored on the same country-trust scale.

Important: years in market are hard-capped by known domain registration age where domain evidence exists, so inflated founding claims do not receive full credit.

Factor 4 — Red Flags (10%)

This factor starts at 100 and loses 15 points per detected flag. It bottoms at 0.

The live algorithm currently checks these six red-flag classes:

  • Support or documentation quality issue.
  • Open-loss or floating-loss rule exists.
  • Minimum trading days required.
  • No trading rules publicly disclosed.
  • No prohibited trading styles explicitly listed.
  • No restricted countries section.

Funded-stage only rule

The public site deliberately treats funded-stage values as the main truth for trader-facing scoring and display.

  • Displayed drawdown, daily loss, max overall loss, drawdown type, profit split, and minimum funded trading-day pressure all reflect funded-stage conditions.
  • Challenge-phase targets and evaluation structure are stored separately for reference, but they do not override funded-stage survivability data.
  • If a visible number is based on challenge rules instead of funded rules, that is treated as a data error, not a design choice.

Evidence hierarchy

Not all sources are treated equally. When marketing copy and legal text conflict, legal text wins.

  1. Terms, policy pages, and legal disclosures.
  2. FAQ, help center, and rule pages.
  3. Registration sources, broker pages, and Trustpilot evidence.
  4. Homepage and marketing pages.

This is why hidden counterparty clauses, payout discretion, floating-loss rules, and buried consistency traps can override positive marketing claims.

Hidden rules and toxic-rule extraction

Prop Firm Secret does not mix hidden rules with ordinary strategy restrictions.

  • Hidden rules: buried fine-print clauses such as open-loss layers, payout discretion, best-day pressure, dual-company structure, and consistency traps.
  • Trading restrictions: public bans or limits such as Martingale, copy trading, EA restrictions, weekend holds, news trading, and country blocks.

Hidden rules are classified by severity and are used both in narrative review sections and in structured downstream scoring logic.

Ranking order and tie-breakers

The rankings feed is sorted in a deterministic order:

  1. Higher AI Score first.
  2. If scores tie, fewer published red flags rank higher.
  3. If still tied, the more recently updated firm ranks higher.

The public machine-readable feed is available at /api/rankings.json.

Score labels

The numeric score is also mapped into trust bands for display:

  • 90+ = Prime Trust
  • 80-89 = Strong Trust
  • 70-79 = Stable Trust
  • 60-69 = Balanced Trust
  • 50-59 = Mixed Trust
  • 40-49 = Watchlist Trust
  • 30-39 = Developing Trust
  • 20-29 = Early Trust
  • 10-19 = Building Trust
  • 0-9 = Initial Trust

Structured data and public outputs

Each review page publishes review-style structured data tied to the firm being audited. The main site publishes a live ItemList and a machine-readable rankings feed. The methodology page exists so users, search engines, and LLM agents can inspect the current scoring logic instead of guessing from page labels alone.

New firm update checklist

  1. Verify canonical domain, slug, site identity, and logo/favicon source.
  2. Extract funded-stage plans, drawdown design, payout structure, and restriction set.
  3. Audit hidden rules, especially floating-loss layers, consistency rules, best-day limits, and payout discretion.
  4. Confirm liquidity or broker evidence, registration details, longevity, jurisdiction, and Trustpilot data.
  5. Recompute the live AI Score and refresh rankings, schema, and supporting summaries.