How to read your certificate

Header and identification

The header shows the TRAVIDENCE mark with the descriptor Independent strategy validation, followed by the data of your validation:

STRATEGY — a system-generated identifier (asset class + evaluation code); not a name you choose.
ASSET CLASS — the market or instrument evaluated.
PERIOD COVERED — the date range of the validated trades.
CERTIFICATE ID — the unique ID, in the format TRV-XXXX-XXXX.
SCORED UNDER — the framework version used for the evaluation.
ISSUED — the issue date.

The Robustness Score and your band

The central block shows the ROBUSTNESS SCORE, from 0 to 100, and a band that interprets it. The band is the qualitative reading of the number:

Robust
Conditional Robustness
Limited Robustness
Not Robust

The Score measures robustness, not expected profitability. When a control caps the result, the certificate may show no number and instead indicate that the score was capped (see section 4). There is a fifth outcome outside these four Robustness bands: the Disclaimer of Opinion, when the sample is insufficient to evaluate (see section 5).

The verdict

The VERDICT section explains in words what the result means: what the strategy showed, what limits it if anything does, and what the next steps are. It's the part that translates the number and the band into an actionable conclusion.

Gate status

The certificate includes a GATE STATUS table with three columns: GATE, STATUS, and EFFECT. There are five possible states. PASSED means the gate cleared. FAILED means the gate did not clear and may limit the result. NOT EVALUATED appears when the outcome is a Disclaimer of Opinion: no gate ran, and NOT EVALUATED does not mean PASSED. The Component Isolation gate has two states of its own. NOT EVALUATED — no per-component data: the gate does apply to the strategy, but it was not evaluated because the platform did not receive per-component trade series — the case for every real evaluation today. N/A — single component: the gate does not apply because a single component was declared; today it appears only in the published sample certificates. In both cases that gate does not cap the result, and the other two show their actual state.

In the EFFECT column, a failed gate is marked as a binding cap — the control that determined the result — or as a non-binding cap. When two or more failed gates tie as determinant, the certificate reports as the binding cap the first one in a fixed order: regime, then subperiod, then component isolation. A PASSED gate shows no effect. A NOT EVALUATED or NOT EVALUATED — no per-component data gate shows not evaluated. An N/A — single component gate shows not applicable: in all three cases there was no cap, because the gate didn't run or doesn't apply. The Combined row summarizes the outcome: Final outcome: no gates triggered when everything passes, Final outcome: capped by … when a control limits the result, or Final outcome: gates not evaluated when the outcome is a Disclaimer of Opinion.

When a gate doesn't pass, the certificate doesn't show a number. Gates are must-pass requirements — you either clear them or you don't — not penalties that subtract points from a score. Showing a number next to a failed gate would imply a continuous scale that doesn't exist, and would invite optimizing against that number instead of addressing the actual cause. That's why the certificate reports the band, which gate didn't clear, and its effect — "binding cap" or "non-binding cap" — instead of a score.

The three gates evaluate, respectively, the strategy's consistency across volatility regimes, the stability of its performance over time (subperiods), and the contribution of each component when the strategy combines several. The table tells you what happened and which control was determinant; the thresholds that define each gate are part of the calibration and are not published.

In the GATE column they appear under these names: Regime Must-Pass, Subperiod Monotonic Decay and Component Isolation.

Page 2. The "Subperiod Monotonic Decay" block on your certificate looks for a sustained decline: performance getting worse from one subperiod to the next without recovering. A large drop in a single stretch, followed by recovery, doesn't count as deterioration. That's why the status can read "PASS" even when the "Largest drop between consecutive subperiods" row shows a high figure: it's printed for transparency, not because the test failed. A strategy that drops sharply one quarter and then recovers can still pass this test. Figures in this table can exceed 100% or be negative — both are mathematically valid, not an error — and when a pair of consecutive subperiods includes a Sharpe ≤ 0, that pair is not computed and the cell shows "n/a" instead of a number — the rule the certificate prints under the table: "Subperiod pairs with Sharpe ≤ 0 are not computed."

Disclaimer of Opinion: when the sample is insufficient

If your evidence isn't enough for an audit-grade evaluation, the outcome is a Disclaimer of Opinion. It is not a "failure" of your strategy: it means there isn't enough data to issue a supported verdict. With small samples it's hard to separate signal from noise, and we'd rather abstain than draw a conclusion the data can't support. The path forward is to generate more trades and resubmit; the credit is not consumed and remains available in your account.

For an opinion to be issued, TRAVIDENCE needs a minimum of 100 trades and at least 1 year of coverage, from the first entry to the last exit. Meeting this minimum does not guarantee the opinion: the evaluation also requires that your trades sufficiently cover the period's different volatility regimes, and that depends on how they're distributed over time — it isn't something you can calculate in advance.

In this case, the GATE STATUS table shows all three gates as NOT EVALUATED, with EFFECT not evaluated on each, and the Combined row reads Final outcome: gates not evaluated. The validator detail section doesn't appear either, because no validator ran. None of this is an unfavorable result: it's the absence of a result, until there's enough evidence.

The Reproducibility Hash

At the bottom you'll find the Reproducibility Hash: a cryptographic fingerprint of the scoring run's canonical inputs, the validators' results, and the calibration constants' fingerprint — not the PDF's bytes or its prose. Under the same engine, the same constants, and the same process, the same inputs produce the same verdict and the same hash. The public verifier, given the TRV- certificate ID, returns the band, issue date, and engine version of the original record; if you also paste your PDF's hash, it confirms that it matches that record — compare those values against your document.

The independence framing

The certificate is the product of an independent validation: we don't trade strategies, we don't sell funded accounts, and we don't earn based on the verdict. That's why the document asserts robustness against the evidence evaluated — it doesn't promise future profitability or eliminate the risk of loss. The second page collects the methodology references that support the evaluation, including López de Prado's Deflated Sharpe Ratio.

Notice. TRAVIDENCE is an independent validation service. It is not financial advice or an investment recommendation. Past performance does not guarantee future results; no validation eliminates the risk of loss.

Unsure about a term? Check the glossary →