Start with one set of records, one data feed, or one automated action.
01 / EVIDENCE
Show where a finding came from
Keep the original public material, when it was collected, and how it was used. Like a receipt folder, the record lets another person follow the work. Clients can use Matrix Scroll to sign records and check them offline.
AGREED OUTPUTSA list of sources, collection notes, and the files needed to review the finding.
02 / SIGNAL QUALITY
Find the alerts worth attention
Compare a feed with known examples and real operating conditions. Count false alarms and missed events separately. Check whether a change in the source or the software reading it has affected the results.
AGREED OUTPUTSA starting measure of feed quality, test results, and recommendations for improving the checks.
03 / AUTHORIZATION
Check what machines can do
Review who can request an action, how signing keys are protected, and how old or repeated requests are handled. For SIGNET USB and Nerve Center, check the software and hardware together and record which controls were tested on each.
AGREED OUTPUTSA clear permissions map, a record of the setup tested, and a list of checks still needed.
THE EVALUATION BASIS
A track record, with room for doubt.
We ask two different questions: how well has this source performed, and how much evidence backs that judgment? A new feed may be hard to judge. A consistently poor feed may be easy to judge. Keeping those questions separate helps your team decide what to investigate.
The math behind a source rating
In the Glicko-2 reference, μ is the rating, φ is its rating deviation on the converted scale, and σ is volatility. Rating deviation describes uncertainty in the rating estimate. Volatility describes expected rating fluctuation. The rating interval uses rating deviation, not volatility.
A rating is not a percentage probability that the next signal is correct. A telemetry assessment needs defined outcomes, representative data, and an update policy tested for that use. More evidence can reduce uncertainty even when it confirms poor performance.
Here, p is the predicted probability, y is the observed outcome (0 or 1), and N is the number of evaluated predictions.
Worked example: a probability of 0.8 for an event that does not occur contributes 0.64 to the error. This is a mathematical illustration, not an SSX360 benchmark.
Lower error on one dataset does not establish performance elsewhere or guarantee the next prediction. Dataset choice, sample size, and operating conditions remain part of the review.
Tell us what you want checked and which records we can use. We put the methods, schedule, report, and price in writing.
02 / ASSESS
Connect findings to evidence.
Record the setup and what happened during each test. Make clear what was observed, what is assumed, and what still needs checking.
03 / HAND OVER
Make review practical.
Receive a report, its supporting records, and a prioritized action list. If agreed, we include signed files and instructions for checking them offline.
A signature helps you check who signed a record and whether it changed. The report explains what the evidence actually supports.
A short description of your system, the decision or review you need to support, and your preferred schedule. We will agree a suitable transfer process before requesting confidential designs, hardware, or raw records.
Can we begin with one feed or workflow?
Yes. Scope can begin with one source collection, feed, or machine workflow. The available inputs, reference data, and review questions determine the methods and deliverables.
How is public-source material handled?
We agree the permitted sources and uses before collection. Captures record available source references and retrieval dates. The review documents gaps, transformations, and relevant access or licensing limits.
Is Matrix Scroll publicly available?
No. Matrix Scroll is licensed to SSX360 clients. Access, documentation, integration, and support are arranged directly as part of a client engagement.
Does an assessment provide certification?
No. These are scoped assessments and engineering reviews. Any required formal evaluation, validation, or certification is a separate process with its own responsible authority.
TELL US WHAT NEEDS REVIEW
Start the conversation.
Tell us what is difficult to check today. We’ll help you choose a practical place to start.