Rating methodology

The CSR Evidence Rating, and the rules that make it worth reading.

Most crypto signal ratings are a number someone chose and a rubric written afterwards to justify it. This one runs the other way: the rubric is published first, the gates are non-negotiable, and the score is whatever the evidence produces — including zero, and including no rating at all.

1Pass the gates

Four conditions before a provider is eligible for any score.

2Score five dimensions

Each worth 0-2 points, against published descriptors.

3Publish the breakdown

The total is meaningless without the per-dimension reasoning.

What the rating measures

The CSR Evidence Rating scores the quality of a provider's evidence and the discipline of its observable trading process. It is not a prediction of future returns, not a profitability estimate, not a safety guarantee, and not financial advice. A 10/10 provider can still lose your money; the rating says its record is complete and its process is disciplined, not that its next call will work.

Eligibility

Four gates before any score exists.

These are pass/fail. A provider that fails any one of them is Not rated — which is not a bad score, it is the absence of a score, and it is the default state of every provider on this site.

Source archive checked. CSR has inspected the original signal archive — not a summary, not a PDF built from it — and sampled every year and every outcome type against it. Until an archive is inspected, no number about it means anything.
Operator identified. A named contracting entity, a working correction route, and a dispute path. An anonymous operator cannot be held to a score.
Loss-inclusive period. The reviewed window includes losing trades, losing months, edits, deletions, and cancelled calls, with an exclusion log CSR has audited. A record with the losses removed is not a record.
No disqualifying connection. CSR does not rate any provider it shares ownership with, is paid a revenue share by, or has a financial stake in. Such providers are recused: they can hold a published evidence record, but never a CSR score. See the common-ownership boundary.

The rubric

Five dimensions, 0-2 points each, 10 points total.

Every dimension is scored against the descriptors below and published with a one-line reason. Dimension 4 is the professional trading assessment: a reviewer with trading experience judges the observable process, not the outcome.

1. Record completeness

Does the reviewed period contain everything that happened, or only what survived?

2 ptsFull continuous period. Wins, losses, breakevens, cancelled orders, edited and deleted calls all present. Gaps are declared and explained.

1 ptContinuous period with minor unexplained gaps, or a complete record for a subset of products only.

0 ptsSelected periods, missing losses, or gaps the provider cannot account for.

2. Source verifiability

Can a third party get from the published number back to the original message that produced it?

2 ptsEvery row traces to an immutable source captured independently of the provider — live bot observation, or an export with hashes CSR took itself — and the hashes are published.

1 ptEvery row traces to a source reference, but the capture was retrospective or provider-run, so pre-scan edits and deletions are invisible.

0 ptsScreenshots, typed summaries, or numbers with no path back to a source.

3. Method disclosure

Is enough published for a competent trader to reproduce the result, or only to admire it?

2 ptsEntries, exits, stops, targets, leverage, position sizing, fees, funding and slippage handling are all disclosed, and the ROI method states which of them it includes.

1 ptTrade mechanics disclosed but costs, leverage comparability or sizing policy are missing, so account-level outcomes cannot be derived.

0 ptsHeadline percentages with no method behind them.

4. Risk discipline Professional trading assessment

Judged by a reviewer with trading experience, against the record: does the desk actually run a risk process, or does it just talk about one?

2 ptsA stated risk framework that the record demonstrably follows: position sizing consistent with the stated rule, stops present and honoured, losses cut at planned invalidation rather than held, exposure and drawdown controlled across correlated positions, and no leverage escalation after losing streaks.

1 ptA risk framework is stated and mostly visible, but with material inconsistencies: stops widened after entry, sizing drifting with conviction, or concentration the stated rules do not permit.

0 ptsNo discernible process: no stops, martingale or revenge sizing, leverage escalation into drawdown, or a "risk-managed" claim the record contradicts.

5. Claim integrity

Does the marketing survive contact with the provider's own record?

2 ptsEvery public performance claim reconciles to the reviewed record, headline metrics are labelled for what they actually measure, and losses are as visible as wins.

1 ptClaims reconcile arithmetically but are presented misleadingly — a non-loss rate sold as a win rate, or leveraged signal sums presented as account returns.

0 ptsPublic claims the record cannot support, or performance figures with no published basis at all.

Reading the score

What each band actually means.

Bands describe evidence and process quality. None of them describes expected profit.

Score	Band	What it means for a reader
9-10	Fully evidenced	Complete, independently captured, fully disclosed, disciplined, and honestly marketed. Rare by construction: it requires independent capture, which most providers will not permit.
7-8	Well evidenced	A serious, checkable record with one or two real gaps, each named on the provider's file.
5-6	Partly evidenced	Enough to review, not enough to rely on. Read the dimension breakdown before paying anything.
3-4	Weakly evidenced	Material gaps in record, method, or discipline. The marketing is ahead of the proof.
0-2	Not evidenced	Claims without a record, or a record that contradicts the claims.
—	Not rated	A gate was not passed, or the provider is recused. Not a criticism; simply no score. Most providers on this site are here.

Standing rules

The rules that keep the number honest

Published before applied. This rubric is public and dated. A score is never issued before the rubric it is scored against is published.

The breakdown ships with the score. A total without its five dimensions and their reasons is marketing, not a rating. Both always appear together.

Ratings expire. Twelve months, or immediately on material change: a new operator, a new product, a claim the record no longer supports.

Money cannot move a dimension. Paid evidence production is a service; the score is an editorial output. A provider may pay CSR to build a record and still be scored 2/10, or refused a score entirely.

Falsifiable by design. Providers and readers can dispute any dimension with evidence. A successful dispute changes the score and the change is logged on the file.

Deliberate limits

What this rating will never be

Not a profit forecast. Evidence quality is backward-looking. A disciplined, fully evidenced desk can still lose money in the next regime; an unrated one can make money. The rating measures what can be checked, not what will happen.

Not a safety guarantee. Scoring the process does not audit custody, solvency, or intent.

Not for sale, and not for our own. No sponsorship, placement, or commercial relationship touches a score, and any provider CSR is financially connected to is recused entirely — the strongest record on the site still gets no stars if we are on the wrong side of that line.

Not applied to what we cannot see. Products a provider keeps private stay unrated and are named as excluded scopes, rather than folded into a flattering average.

Current state

Today, no provider on this site holds a CSR Evidence Rating.

The rubric is live; the scores are not, because gate 1 — an inspected source archive — has not yet been passed by any provider, including the ones with the most complete published records. That is the honest state, and it is published rather than hidden behind a placeholder number. Providers who want to be scored can start at Signal Passport; the first archive CSR inspects will be the first score CSR issues, whatever that score turns out to be.

Risk language

A rating is due-diligence material, not instruction.

Crypto trading is speculative and can result in losses. The CSR Evidence Rating is an editorial assessment of published evidence and observable process. It does not recommend trades, guarantee outcomes, provide personalized financial advice, or replace your own risk management. Past results, however well evidenced, do not predict future performance.