The CFTC’s 2016 testing rules required independent testers and a written decision to fix or accept findings

Two CFTC rules published September 19, 2016, named five kinds of cyber testing for exchanges, swap facilities, data repositories and clearinghouses. They set who could test, minimum frequencies for covered entities, and a documented decision to remediate or accept each risk found.

Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.

Two-minute summaryThe dek and the first sentence of each section, quoted from the article

Two CFTC rules published September 19, 2016, named five kinds of cyber testing for exchanges, swap facilities, data repositories and clearinghouses. They set who could test, minimum frequencies for covered entities, and a documented decision to remediate or accept each risk found.

  1. The week The first rule covered designated contract markets, swap execution facilities and swap data repositories. [2] Those are futures exchanges, swap trading platforms and the repositories that hold swap data.
  2. The tests: the five types as the rules defined them The Commission presented much of the rule as clarification.
  3. The testers: who may run each test The rules used one formula across most test types.
  4. The calendar: minimum frequencies and who they applied to For every entity, each test type had to occur at a frequency determined by an appropriate risk analysis. [2] [3] That was the baseline for swap execution facilities and for exchanges that were not covered. [2] It meant the entity’s own analysis set the schedule, and the Commission could ask to see it.
  5. The findings: reporting, remediation and records That duty changes the audience.
  6. How the rules land on different entities That flexibility puts weight on the written analysis.
  7. Objections, in order A floor can become a target.
  8. With hindsight The 2016 text has lasted.

The week

On September 19, 2016, the Federal Register published two final rules from the Commodity Futures Trading Commission on system safeguards testing. [2][3] Both took effect that day. [2][3] The Commission had approved them at an open meeting on September 8, 2016, by unanimous vote. [1]

The first rule covered designated contract markets, swap execution facilities and swap data repositories. [2] Those are futures exchanges, swap trading platforms and the repositories that hold swap data. The second rule covered derivatives clearing organizations, the clearinghouses that stand between buyers and sellers. [3]

The rules grew out of proposals the Commission issued on December 15, 2015, and published on December 23, 2015. [2] The final rule for exchanges, platforms and repositories explained the need. It said cyber threats to the financial sector continue to expand, increasing the need for enhanced cybersecurity testing. [2]

The same rule said testing should focus on an entity’s ability to detect, contain, respond to and recover from cyber attacks. [2] It called compromise of data integrity perhaps the greatest threat for financial sector data. [2] It cited guidance on cyber resilience for financial market infrastructures from international standard setters, published in June 2016. [2]

The rule’s summary named five types of testing. They were vulnerability testing, penetration testing, controls testing, security incident response plan testing and enterprise technology risk assessment. [2] The Commission said it was clarifying its existing rules by specifying and defining those types. [2]

The rule also listed categories of risk analysis and oversight that each entity’s program had to address. For exchanges, the categories ran from enterprise risk management and governance through information security, business continuity, capacity, systems operations, systems development and physical security. [2]

New provisions applied to a narrower group. The rule set minimum testing frequencies and required certain tests by independent contractors for “covered” designated contract markets and all swap data repositories. [2] A covered market was one whose annual trading volume in 2015 or later was 5 percent or more of the combined volume of all designated contract markets. [2]

Swap execution facilities were left out of those minimums. The Commission had asked for comment on a possible “covered SEF” definition. [2] It said it agreed with comments calling for further consideration and consultation before proposing such requirements. [2]

The clearinghouse rule went further on frequency. It required every derivatives clearing organization to run vulnerability tests at least quarterly. [3] It required external and internal penetration tests at least annually, with the required annual external test performed by independent contractors. [3] It required key controls to be tested at least every three years, also by independent contractors. [3]

Both rules required internal reporting. Senior management and the board had to receive and review reports of the testing results. [2][3] Both also required each entity to document the vulnerabilities and deficiencies found and decide, with a documented analysis, whether to remediate them or accept the risk. [2][3]

Compliance came in stages. For exchanges, platforms and repositories, the rule gave 180 days after the effective date for vulnerability testing and incident response plan testing. [2] It gave one year for penetration testing, controls testing and the enterprise technology risk assessment. [2] The clearinghouse rule set calendar dates instead: March 20, 2017, for vulnerability testing and incident response plan testing, and September 19, 2017, for penetration testing, controls testing and the enterprise technology risk assessment. [3]

The September 2016 CFTC rules are easy to read as a list of tests. The more durable content is about people and paper. The rules said who could run each test and what had to happen to the findings. [2][3]

Our argument is that the rules turned security testing from a technical activity into a governance record. They required testers to be independent of the systems they tested, for most test types. [2][3] They required results to reach senior management and the board. [2][3] And they required a documented decision on whether to fix or accept the risks found. [2][3]

We take the rules in four parts:

  1. The tests: what each of the five test types is, as the rules defined them.
  2. The testers: who may run each test, and how independence was defined.
  3. The calendar: the minimum frequencies, and which entities they applied to.
  4. The findings: reporting to management and the board, remediation decisions and records for the Commission.

The first part sets what is done, the second who does it, the third how often, and the fourth what follows. Together they cover a testing program from design to consequence. After them we look at how the rules land on different entities, take the strongest objections in order, and end with checks a CISO or chief risk officer could run.

The tests: the five types as the rules defined them

Clarification, not invention

The Commission presented much of the rule as clarification. It said it was clarifying its existing system safeguards rules by specifying and defining the types of testing essential to meeting existing testing obligations. [2] It said the system safeguards provisions of the Commodity Exchange Act and its regulations had become effective in August 2012. [2]

It also said generally accepted best practices had called for each type of testing well before that date. [2] In our reading, that framing mattered for enforcement. An entity could not argue that the five test types were new duties it had no reason to expect. The new parts were the frequencies and contractor duties for covered entities. [2]

Vulnerability testing

The rules defined vulnerability testing as testing of automated systems to find what information may be discoverable through reconnaissance and what vulnerabilities may be present. [2][3] It had to include automated vulnerability scanning that follows generally accepted best practices. [2][3]

That definition starts with reconnaissance. It asks what an outsider could learn before attacking. That framing ties vulnerability testing to an attacker’s first steps as well as to a patch list.

External and internal penetration testing

External penetration testing meant attempts to penetrate automated systems from outside their boundaries to identify and exploit vulnerabilities. [2] Internal penetration testing meant the same from inside the boundaries. [2] Methods included circumventing the security features of a system. [2]

The two definitions matter because they assume a breach can start in either place. An internal test asks what an insider or an intruder who is already in could do. The rules required both. [2][3]

Controls testing

Controls testing meant assessing controls to determine whether they are implemented correctly, operating as intended and enabling the entity to meet the rule’s requirements. [2] Controls were the safeguards protecting the reliability, security or capacity of systems, or the confidentiality, integrity and availability of data. [2]

The rules introduced “key controls.” These were controls a risk analysis found to be critically important, or aimed at risks that change often enough to need more frequent review. [2] Key controls carried the strongest testing duties. [2][3]

Security incident response plan testing

The rules defined a security incident as a cyber or physical security event that jeopardizes, or is significantly likely to jeopardize, system operation or data. [2] A security incident response plan documented the policies, controls, procedures and resources for identifying, responding to, mitigating and recovering from incidents. [2] It also documented the roles of management, staff and independent contractors. [2]

The plan had to include the entity’s definition and classification of incidents. [2] It had to include its policies for reporting incidents and for internal and external communication and information sharing. [2] It had to set out the hand-off and escalation points in its response process. [2]

Testing could use checklists, walk-throughs, tabletop exercises, simulations or comprehensive exercises. [2] The rules let entities coordinate this testing with other testing or with business continuity and crisis management tests. [2]

Enterprise technology risk assessment

The fifth type was a written assessment analyzing threats and vulnerabilities in the context of mitigating controls. [2] It had to identify, estimate and prioritize risks to the entity’s operations or assets, or to market participants and others. [2] Those were risks from impairment of data or systems. [2]

An entity that had done a compliant assessment could update it in later years rather than start fresh. [2][3] In our reading, that makes the assessment a living register. It also means a weak first assessment can carry its weaknesses forward if updates only add to it.

The scope requirement

Every test type had to be of a scope sufficient to meet a common standard. [2][3] The scope had to cover the systems and controls that the risk program and current cyber threat analysis indicated were necessary to identify certain risks. [2] Those were risks that could let an intruder, an unauthorized user or an insider interfere with operations or impair systems. [2]

They also included risks that could let such a person add to, delete, modify, exfiltrate or compromise the integrity of regulated data, or take other unauthorized actions. [2] The scope clause tied every test to a threat analysis. In our reading, that was the rules’ check against testing only what is easy to test.

Data integrity as a test target

The rule singled out data integrity. It said testing should address detection, containment and recovery from compromise of data integrity, which it called perhaps the greatest threat for financial sector data. [2] It set that beside compromise of availability or confidentiality. [2]

Integrity is hard to test. An outage announces itself; altered data may not. For an exchange or a repository, a changed trade record or price can spread to clearing, reporting and surveillance before anyone notices. The scope clause named adding, deleting or modifying regulated data as a risk tests must cover. [2]

In our reading, a test program that took the integrity point seriously would include scenarios where an attacker changes data rather than steals it. It would ask whether reconciliation, audit trails and monitoring would catch the change, and how fast. The rule did not script those scenarios. It named the threat they address. [2]

The risk categories behind the tests

The tests sat inside a broader program. For exchanges, the program had to cover seven categories of risk analysis and oversight. [2] Enterprise risk management and governance came first, including board and management oversight of technology and security and remediation of deficiencies. [2]

The rule described six categories that existed before 2016. [2] The final list placed enterprise risk management and governance alongside them. [2] In our reading, adding governance as a named category was a signal that testing results were meant to reach decision-makers.

The testers: who may run each test

The independence standard

The rules used one formula across most test types. A test could be run by independent contractors or by employees who are not responsible for development or operation of the systems or capabilities being tested. [2][3]

That formula applied to vulnerability testing, internal penetration testing and enterprise technology risk assessments. [2][3] It applied to controls testing beyond the key controls that had to go to outside contractors. [2][3] It applied to external penetration testing beyond the one annual test that had to go to outside contractors. [2][3]

Where outside contractors were required

For covered exchanges and all swap data repositories, two tests had to go to independent contractors. [2] The required annual external penetration test was one. [2] Testing of key controls at least every three years was the other. [2]

Clearinghouses carried the same two requirements. [3] The clearinghouse proposal had also sought independent contractors for two of the four quarterly vulnerability tests each year. [3] The final rule let clearinghouses use either independent contractors or employees not responsible for the systems tested. [3]

The exception for incident response testing

Incident response plan testing was different. The rules let entities run it with independent contractors or with their own employees, without the restriction on who those employees could be. [2][3]

That exception makes sense. An incident response test exercises the people who would respond. Excluding the operations team from it would defeat the purpose. The independence rule follows the logic of each test.

The incident response plan also had to cover external communication and information sharing. [2] In our reading, a realistic test would include the people outside the security team who handle those duties: legal, communications, the business lines and, where relevant, clearing members or other market participants. A test that stops at the security team tests only part of the plan the rule described. [2]

Why independence matters for a security test

The Commission explained the independent testing of key controls by reference to best practices. It said NIST standards call for controls testing to include assessment by independent assessors, free from actual or perceived conflicts of interest. [2] It said that validates the completeness, accuracy, integrity and reliability of results. [2]

A team that built a system knows where it is strong. It may also know where it is weak and test around it, without intending to. A team that did not build the system has no reason to avoid the weak spots. The rules required that second view for most tests. [2][3]

Current responsibility, not history

The independence test looks at responsibility for development or operation of the systems being tested. [2][3] The text speaks in the present. In our reading, an employee who once built a system and has moved to another role is not barred by the words of the rule.

That reading leaves a judgment call. A former developer may carry the same blind spots as the current team. An entity could adopt a stricter internal rule, such as a cooling-off period, without any requirement to do so. The rule set a floor for independence, not a ceiling.

What independence does not guarantee

Independence is about relationship, not skill. An employee from another department may be independent of a system and still lack the expertise to test it. The rules did not set qualifications for testers. [2][3]

In our reading, that left two questions to each entity. Is the tester independent? Is the tester competent? A program that answers only the first meets the rule’s text and may still produce weak tests.

Building an internal testing team

The employee route invites a structural choice. An entity can build a security testing team that reports outside technology operations. That team then qualifies as independent for most tests across many systems, because it is not responsible for building or running them. [2]

The choice has costs. A small exchange may lack the scale for a separate team. Outside contractors then become the practical route even where the rules allow employees. The rules left that choice to each entity, except for the tests where they required contractors. [2][3]

Rotation and familiarity

The rules did not require rotating contractors or internal testers. [2][3] Independence was measured against responsibility for the system, not against familiarity with it. A contractor that tests the same systems every year stays independent under the text. [2]

Familiarity cuts both ways. A returning tester knows the environment and can go deeper. A returning tester may also repeat the same approach and miss what changed. In our reading, the rules’ requirement to evaluate the effectiveness of testing protocols is the place to weigh that trade-off. [2][3]

The contractor relationship

Using a contractor raises its own issues. The contractor needs access to systems and data during testing. Its reports become records the Commission can request. [2] Its findings go into the remediation process. [2]

In our reading, an entity would want the engagement terms to cover scope, rules of engagement, data handling, report ownership and retention. The rules did not prescribe those terms. They followed from the duties the rules did set.

The calendar: minimum frequencies and who they applied to

The default: frequency by risk analysis

For every entity, each test type had to occur at a frequency determined by an appropriate risk analysis. [2][3] That was the baseline for swap execution facilities and for exchanges that were not covered. [2] It meant the entity’s own analysis set the schedule, and the Commission could ask to see it.

The minimums for covered exchanges and repositories

For covered exchanges and all swap data repositories, the rules set floors. [2] Vulnerability testing had to occur at least quarterly. [2] External and internal penetration tests had to occur at least annually. [2] Incident response plan testing and the enterprise technology risk assessment had to occur at least annually. [2]

Key controls had to be tested at least every three years, and the rules allowed that testing on a rolling basis over the period. [2] A risk analysis could set a shorter period, and then the shorter period governed. [2]

The minimums for clearinghouses

The clearinghouse rule applied its minimums to every derivatives clearing organization. [3] Vulnerability testing was at least quarterly. [3] Penetration testing, external and internal, was at least annual. [3] Incident response plan testing and the enterprise technology risk assessment were at least annual. [3] Key controls were tested at least every three years. [3]

The Commission said its final rule required key controls testing at least every three years rather than every two, as proposed. [3] It reiterated that if a clearinghouse’s risk analysis showed a key control needed more frequent testing, the clearinghouse had to test more often. [3]

Planning a rolling key-controls cycle

The rolling option for key controls asks for a plan. [2][3] An entity first has to decide which controls are key, through a risk analysis. [2] It then spreads testing over the period, so that every key control is tested within it. [2]

The definition of key controls includes controls aimed at risks that change often. [2] Those may need testing more than once in the cycle. An entity that lists its key controls once and never revisits the list could miss new ones. In our reading, the key-control list belongs in the enterprise technology risk assessment, which is updated each year for covered entities. [2]

The outside-contractor duty adds a constraint. For covered exchanges, repositories and clearinghouses, independent contractors had to test and assess the key controls at least every three years. [2][3] An entity can phase that work, but the contractor’s coverage over the cycle has to reach every key control. [2][3]

Who counted as covered

The covered exchange definition turned on market share. A designated contract market with 5 percent or more of combined annual trading volume across all designated contract markets in 2015 or a later year was covered. [2] A market that stayed below 5 percent for three consecutive years ceased to be covered as of March 1 of the following year. [2]

The rule built in the data flow. Each market had to report its annual trading volume to the Commission, and the Commission would tell each one its share. [2] That made coverage a fact the Commission computed, not one each market decided for itself.

Why swap facilities were treated differently

Swap execution facilities stayed on risk-based frequency. The Commission had asked about a “covered SEF” concept in an advance notice. [2] Commenters raised issues including the early stage of the swap facility market and the chance that a facility could be covered one year and not the next. [2] The Commission agreed further consideration and consultation were needed. [2]

In our reading, that left a gap between venues trading similar risk. A swap facility could test on its own schedule while a covered futures exchange followed floors. The gap was a policy choice the Commission said it would revisit. [2]

Business continuity alongside testing

The rules also tightened business continuity. An exchange’s continuity plan generally should enable resumption of trading and clearing during the next business day after a disruption. [2] Markets the Commission determined to be critical financial markets faced more stringent requirements under a separate rule. [2] Plans had to be updated at least annually. [2]

The clearinghouse rule required its continuity plan and emergency procedures to be updated at least annually as well. [3] In our reading, testing and continuity planning met in the incident response plan, which could be a section of the continuity plan. [2]

Staged compliance

The rules gave time to comply. For exchanges, platforms and repositories, vulnerability testing and incident response plan testing were due 180 days after the effective date. [2] Penetration testing, controls testing and the enterprise technology risk assessment were due one year after. [2] Covered entities had three years from the effective date for independent contractor testing of all key controls. [2]

The Commission said it had considered comments that cited the SEC’s approach under Regulation SCI, which used a later compliance date. [2] It said it agreed with those comments and set compliance dates to reflect them. [2]

The findings: reporting, remediation and records

Reporting up

Both rules required that senior management and the board receive and review reports of testing and assessment results. [2][3] Both required procedures for remediating issues identified in that review and for evaluating whether testing protocols worked. [2][3]

That duty changes the audience. A penetration test report written for engineers can be dense and technical. A report a board must review needs a summary a director can act on. In our reading, the rule pushed security teams to write for two audiences.

The remediation decision

The remediation clause had three steps. The entity had to identify and document the vulnerabilities and deficiencies that testing revealed. [2][3] It had to conduct and document an analysis of the risks they presented. [2][3] And it had to determine and document whether to remediate or accept the risk. [2][3]

The clearinghouse rule put it per finding. Its analysis covered the risks presented by each vulnerability or deficiency. [3] When an entity chose to remediate, it had to do so in a timely manner given the nature and magnitude of the risk. [2][3]

Why “accept” is the important word

The rules allowed risk acceptance. [2][3] They did not require every finding to be fixed. In our reading, that reflects a trade-off: some findings may cost more to fix than the risk they pose.

What the rules added was a record. An accepted risk had to rest on a documented analysis. [2][3] In our reading, that turns acceptance from an unrecorded deferral into a decision someone made, with reasons, that a regulator can read.

What “timely” means

The rules did not set remediation deadlines. They required remediation in a timely manner given the nature and magnitude of the associated risk. [2][3] That standard scales with severity but leaves the numbers to the entity.

In our reading, an entity would do well to write its own targets by severity, so that “timely” becomes a date someone can miss. Without them, an examiner has to judge timeliness after the fact, and the entity has to defend each delay one at a time.

Who signs the acceptance

The rules did not name who decides. They required the board and senior management to review the reports. [2][3] In our reading, a risk acceptance that never reaches the people who reviewed the report is hard to square with that design.

An entity can set thresholds. Low risks might be accepted by a security manager. High risks might need a senior executive or a board committee. The rules left the structure open but made the decision traceable.

Linking findings to the risk assessment

An accepted risk does not disappear. It stays in the environment. The enterprise technology risk assessment analyzes threats and vulnerabilities in the context of mitigating controls and prioritizes the resulting risks. [2] In our reading, that assessment is the natural home for accepted findings.

Placing them there has a practical effect. Each year’s update would revisit what was accepted and ask whether the threat or the control has changed. An acceptance made when a system was internal may not hold once it is exposed to the internet. The annual assessment for covered entities gives a fixed point to ask. [2]

What a board report could contain

The rules required the board to receive and review reports of test results. [2][3] They did not prescribe a format. A useful report would show what was tested, by whom, and against what scope. It would list significant findings with the decision on each: fix, with a date, or accept, with a reason.

It would also show what remained open from earlier tests. A board that sees the same finding accepted year after year can ask whether the acceptance still makes sense. In our reading, that continuity is what the review duty is for.

Records for the Commission

The exchange rule required production, promptly on request, of current continuity plans, assessments of operational risks or controls, and testing reports. [2] It covered reports whether performed by independent contractors or by employees. [2] It also covered other records Commission staff requested for oversight of system safeguards. [2]

The clearinghouse rule had a parallel list. [3] That made test reports regulatory records. An entity could not treat a contractor’s findings as private advice to management once the Commission asked for them.

Evaluating the testing itself

Both rules required procedures to evaluate the effectiveness of testing and assessment protocols. [2][3] That is testing of the tests. It asks whether the program finds what it should.

An entity might evaluate by comparing findings over time, by checking whether incidents occurred in areas the tests rated clean, or by rotating contractors. The rules did not prescribe a method. [2][3] They required the evaluation.

How the rules land on different entities

A large futures exchange

A covered exchange carried the full set: floors on frequency, outside contractors for two tests, board review and documented remediation. [2] Its risk analysis could still require more frequent testing than the floors. [2]

A smaller exchange or a swap execution facility

A non-covered exchange or a swap facility set frequency by its own risk analysis. [2] It still had to run all five test types and meet the independence standard. [2] It could use employees not responsible for the systems tested, or contractors, for every test. [2]

That flexibility puts weight on the written analysis. An examiner reviewing a smaller venue would ask why its chosen frequency fits its threats and systems. In our reading, a venue that tests less often than the covered floors should be able to explain why, in terms of its own exposure, rather than by pointing to its size.

For a small entity, the employee route may be thin. A firm with a handful of technologists may have nobody who is both qualified and uninvolved in running the systems. The analysis that sets frequency becomes the key document an examiner reads.

A swap data repository

Every swap data repository carried the covered-entity minimums. [2] Repositories hold transaction data that regulators and the public rely on. The rule’s emphasis on data integrity fits them closely. [2]

For a repository, the scope clause’s reference to modifying or compromising the integrity of regulated data reads as the central risk. [2] A repository whose data is altered passes the error to every regulator and market participant that relies on it. In our reading, its testing program would weight integrity scenarios, reconciliation checks and audit-trail reviews more heavily than an exchange’s might.

A clearinghouse

Every clearinghouse carried the minimums and the contractor requirements. [3] The Commission had determined that clearinghouses are not small entities for regulatory flexibility purposes. [3]

A clearinghouse sits between members and depends on their connections, its banks and its own margin systems. Its business continuity plan had to be updated at least annually. [3] In our reading, its incident response testing would need to include members and settlement banks, because a clearinghouse incident can quickly become theirs. The rule allowed coordinating incident response tests with other continuity and crisis tests, which suits that kind of joint exercise. [3]

A technology vendor to a market

Vendors that build or host systems for these entities were not regulated by the rules. The entities were. The program’s systems development category included outsourcing and vendor management. [2] The scope clause reached the systems and controls the risk program found necessary to test. [2]

In our reading, an exchange or clearinghouse that runs critical functions on a vendor’s platform would need test rights in its contract. Without them, it could not meet the scope requirement for those systems. The vendor would feel the rules through its customers.

A bank-affiliated futures commission merchant or swap dealer

The 2016 rules did not reach futures commission merchants or swap dealers. [2][3] Those firms had other CFTC rules and, when part of banks, bank supervisors. In our reading, the 2016 testing model was a reference for them, not a requirement. A later CFTC proposal aimed at them is covered in the hindsight section.

A bank or credit union outside derivatives markets

Banks and credit unions were not subject to these rules. Their testing expectations came from their own supervisors. The Commission said some financial regulators, such as the FFIEC agencies, are themselves sources of generally accepted best practices. [2] The independence formula and the fix-or-accept record are design patterns any financial institution could borrow.

Objections, in order

1. Prescribed frequencies turn testing into a calendar exercise

A floor can become a target. The rules addressed that by making risk analysis the primary driver and requiring more frequent testing where analysis called for it. [2][3] The floors applied to a subset of entities. [2] The risk remains, but the text pushes against it.

2. Outside contractors add cost without adding insight

For clearinghouses, the final rule moved key controls testing from every two years, as proposed, to every three. [3] Outside contractors were required for two tests only. [2][3] The rest could use independent employees. [2][3] The cost burden was narrower than the objection suggests.

3. Allowing risk acceptance weakens the rule

It could, if acceptance were undocumented. The rules required a documented analysis behind each decision and review of results by senior management and the board. [2][3] In our reading, a documented acceptance is a stronger control than an unwritten deferral, which the rules replaced.

4. Leaving swap facilities out created uneven protection

That is a fair criticism. The Commission chose to consult further rather than define covered swap facilities in 2016. [2] All swap facilities still had to run the five test types at risk-based frequencies. [2] The unevenness was in floors and contractors, not in the duty to test. A swap facility whose own risk analysis supported quarterly scans and annual penetration tests could match the covered entities’ schedule without being told to. [2] The record would then show that choice, which is what an examiner would look for.

5. Testing standards drawn from best practices are too vague to enforce

The Commission tied each test type to generally accepted standards and best practices, and gave definitions. [2] Definitions, scope rules and record duties give examiners something concrete to check. Vagueness remains at the edges, such as who counts as competent to test. [2]

6. Testing finds problems but does not fix them

Testing alone does not reduce risk. The rules recognized that by attaching duties to the results. [2][3] Findings had to be documented, analyzed and either remediated in a timely manner or accepted with reasons. [2][3] The objection describes a testing program without those duties, which is what the rules were written to prevent.

Checks a CISO or chief risk officer could run

The tests

  • List the five test types and the last date each was performed, by system in scope.
  • Confirm each test’s scope traces to the current threat analysis rather than to what is easy to test.
  • Check that the enterprise technology risk assessment was revised where risks changed, rather than appended to.

The testers

  • For each test, record who performed it and confirm they are not responsible for building or running the system.
  • Check qualifications as well as independence.
  • Review contractor terms for scope, data handling, report ownership and retention.

The calendar

  • Find the written risk analysis behind each test frequency.
  • Compare frequencies with any regulatory floor and note where the analysis requires more.

The board report

  • Show what was tested, by whom and against what scope.
  • List significant findings with the decision on each and its owner.
  • Carry forward open and accepted findings from earlier tests until they close.

The findings

  • Confirm test results reach senior management and the board, with a summary they can act on.
  • For each significant finding, check for a documented decision to fix or accept, with reasons and an owner.
  • Track remediation dates against the stated urgency of each finding.
  • Confirm testing reports, including employees’ reports, can be produced to a regulator promptly.

With hindsight

The 2016 text has lasted. As checked on November 16, 2026, the eCFR text of 17 CFR 38.1051 still defined a covered designated contract market by the 5 percent trading volume test. [4] It still required documented analysis to decide whether to remediate or accept risks found in testing. [4] The section’s source note listed the September 19, 2016 amendments and a September 29, 2017 amendment. [4]

The clearinghouse rule stood as well. As checked on November 16, 2026, 17 CFR 39.18 still required independent contractors to perform the required annual external penetration test. [5] It still required an analysis of the risks presented by each vulnerability or deficiency, and a documented decision to remediate or accept the risk. [5] Its source note cited the September 19, 2016 rule. [5]

The Commission later tried to extend a similar model to other firms. On January 24, 2024, it published a proposed rule on an operational resilience framework for futures commission merchants, swap dealers and major swap participants. [6] The framework would have included an information and technology security program, a third-party relationship program and a business continuity and disaster recovery plan. [6]

The proposal carried governance, training, testing and recordkeeping requirements. [6] It would have required testing of the information and technology security program to include vulnerability assessments and penetration testing. [6] It also proposed notifying the Commission of certain incidents no later than 24 hours after detection. [6]

That proposal did not become a rule. On September 11, 2025, the Commission published a notice withdrawing it as of September 8, 2025. [7] The Commission said it did not intend to issue final rules on the proposal. [7] It said it would issue new proposed rules if it decided to pursue future action in the area. [7]

The Commission gave a reason. It said it was withdrawing the proposal to reconsider how the proposed rules would work with other regulators’ operational resilience rules, domestic and foreign. [7] It said many swap dealers and futures commission merchants are part of banking organizations that already run enterprise-wide operational resilience programs under other regulations. [7]

The proposal itself had cited earlier incidents to explain its purpose. It described a 2016 case in which a hacker accessed customer records on a futures commission merchant’s backup storage device after a default configuration left it open to the internet. [6] It described a 2018 phishing attack that led to a fraudulent wire of $1 million from a customer’s account. [6] Those are the Commission’s own descriptions in the proposal.

What held up from 2016 is the core design. The test definitions, the independence formula, the board reporting duty and the fix-or-accept record remain in force for exchanges and clearinghouses, as checked on November 16, 2026. [4][5]

What did not hold up is the expansion. As checked on November 16, 2026, the Commission had withdrawn its proposal to apply a comparable framework to intermediaries. Its withdrawal said any future action would come through new proposed rules or other issuance consistent with the Administrative Procedure Act. [7] In our reading, testing at firms outside market infrastructure therefore rests on other CFTC rules, other regulators and the firms’ own programs.

For a CISO in any financial firm, the 2016 rules remain a usable template. They answer three questions examiners ask: who tested, who saw the results, and who decided what to do about them. [2][3]

Spotted an error? Tell us and we will check it and correct the article where needed.

  • CFTC
  • penetration testing
  • controls testing
  • independent testing
  • remediation
  • market infrastructure