A two-week CCAR-F study plan for the flattest blueprint in the programme

A two-week CCAR-F study plan: one cold timed form on day one, eleven study days allocated to the published domain weights, and two fresh timed forms used to identify topics to revisit. The allocation comes out almost even: three days on agentic architecture and two on each of the other four. Every form is built to the five published domains at their weights, and no result on one is a pass prediction.

The flattest paper in the programme

Every other post on this site argues the same thing: weight your study time to the blueprint rather than to your interests. Here that argument produces an almost even answer. Agentic architecture is 27% and context management is 15%, so eleven study days split by weight comes out at 3.0, 2.2, 2.2, 2.0 and 1.6, which rounds to three days and then two, two, two, two.

That is a result, not a failure to prioritise. The comparison worth holding is the developer plan, where one domain takes three of eleven days and two others share a single afternoon. Full weights for all four tracks are in the domain weights post.

The shape of the fortnight

One cold practice paper, eleven study days, two fresh forms, and one review. The cold paper gives a starting snapshot by domain at the published weights, which on this paper mostly tells you which of five nearly equal domains you know least, not which one you can skip. The free 20-item CCAR-F practice paper needs no account. It is shorter than a form, but it is timed at the official pace and still scores by domain.

Use fresh forms late in the plan so the results are less affected by recalling a specific paper. Each of the five forms is laid out in scenario blocks: it presents four of the six published scenarios, with its items grouped under each. Treat every result as study context, not as an official score conversion, outcome prediction, or recommendation about exam scheduling.

The day-by-day plan

Three days on agentic architecture is about breadth rather than depth: it holds seven objectives, more than any other domain on the paper, at roughly two items each. Covering all seven beats mastering three, which is the argument in the post on that domain.

The risk moves to omission

On an uneven blueprint the failure mode is mis-allocation: spending a week on something worth two items. On a flat one there is no such trap, and the failure mode becomes omission: finishing the fortnight having never opened one of the five.

The two most commonly skipped here are the ones with the least distinctive vocabulary. Tool design looks like a subset of agents until you notice it is 18% on its own and turns on contracts rather than orchestration; and context management looks like a footnote until you notice every failure in it is silent. Between them that is a third of the paper.

What one study day looks like

Ninety minutes in three parts. Twenty reading the objectives for that domain and writing each as a one-line decision rule: a trigger and a response, in your own words. Fifty on practice items from that domain only, closed-book. Twenty on the misses, naming which rule would have caught each.

A miss that maps to no rule you have written is the useful one: it means a rule is missing rather than forgotten. The twelve recurring decision rules are a reasonable starting library, and reading the ask before the options is the habit that makes the practice half go faster.

The review on day 14

No guide publishes a raw-to-scaled conversion for the 720 standard, so a practice percentage cannot predict the official result. What we read instead is Cred Farmer's readiness score, which runs on its own 100 to 1,000 practice scale and counts only your first sitting of each full form. Drills, re-sits and the free 20-item paper never count.

Exam-ready (90) is out of reach until the average of those first sittings is 800 or more: 47 of 60 correct (78%) on a CCAR-F form, against 42 of 60 (70%) for the 720 cut. Nobody has checked that scale against real exam results, so treat it as a signal, not a promise. The real exam does not judge domains one by one, but on a flat paper the weakest domain still costs more than it would elsewhere, because nothing else on the paper is heavy enough to carry it: that domain takes the last study block. For context on the official scaled score, read the passing score post. The repair loop is in the study guide.

Key takeaways

• The flattest paper in the programme. 27% down to 15%, a spread of 1.8 to 1 against the developer exam’s 12.7 to 1.
• Weighting produces a near-even plan. Three days on agents, two on each of the other four. That is the blueprint, not laziness.
• The risk is omission, not mis-allocation. No domain is small enough to skip when the lightest is still nine items.
• Compress, do not cut. If the fortnight shortens, shorten every day rather than deleting a domain.
• Agents get breadth, not depth. Seven objectives at about two items each; cover all seven.
• A form result is study context. The five forms are built to the published weights, and none of them provides an official score, prediction, or scheduling recommendation.