United States
Hong KongHow to review a wrong answer so it stays fixed
Sort your answers into four piles: right and confident, right but guessed, wrong, and unanswered. Review the last three. Write one line per item naming what the wrong option was testing, not what the right answer was. Then re-test the same items a week later.
Most review is reading the explanation, nodding, and moving on. It feels like learning because comprehension feels like acquisition. It is not, and the proof arrives two weeks later when the same question catches you again in a mock.
What follows takes longer than nodding. It is also the only part of practice that reliably moves a score.
Sort into four piles first
Before you read a single explanation, split what you just did.
| Pile | What it means | What to do |
|---|---|---|
| Right and confident | You knew it and you knew why | Nothing. Do not re-read the explanation. |
| Right but guessed | You eliminated your way there or picked on a feeling | Review as if wrong. This pile is the most predictive of all four. |
| Wrong | You picked a distractor | Full review. Name the pattern. |
| Unanswered | You ran out of time or skipped | Review, then ask why it took so long. |
The second row is the one candidates refuse to take seriously. A question you guessed correctly counts as a mark in practice and will not count as one on the day, because the guess lands differently. If your score looks like a pass and half of it came from pile two, you are not where you think you are.
Mark uncertainty as you work, not afterwards. Retrospective honesty about whether you knew something is unreliable.
The one-line rule
For every item in piles two, three and four, write exactly one line. The line names what the wrong option was testing. It does not restate the correct answer.
| Useless note | Useful note |
|---|---|
| The answer is Part X | I picked the discipline Part because I was thinking about punishment when the question was about prevention |
| Client money goes into a segregated account within one business day | I picked the option describing what the firm did correctly. Partial compliance again. |
| The Fund Manager Code applies to Type 9 | I do not have the codes mapped to licence types. Build that table. |
| Enhanced due diligence for a politically exposed person | I read higher risk as a reason to refuse. Risk raises the standard, it does not bar the client. |
The left column is a fact and facts are cheap. The right column is a diagnosis, and a diagnosis transfers to questions you have never seen. That is the whole difference between review that works and review that feels productive.
If you cannot write the right-hand version, you have not finished reviewing. Go back to the source, find the rule, and try again.
The error log
One document, four columns, added to after every session. Keep it in one place and never split it by topic.
- Date. So you can see whether an error repeats after you thought you had fixed it.
- Topic. For the coverage view.
- Pattern. Which of the recurring wrong-option types caught you. The list is in the distractor patterns post.
- The line. Your one-sentence diagnosis.
After thirty or forty entries, count the pattern column. It will be lopsided. That lopsidedness is the single most useful piece of information you will generate during your whole preparation, and no provider can generate it for you, because it is about how you think rather than about the syllabus.
A candidate scoring poorly in Topics 4, 5 and 6 looks like they have three problems. If eleven of their fifteen errors are partial-compliance answers, they have one problem appearing in three places, and one week of the right drill fixes all three.
When to go back to the source
Not for every error. Going to the Ordinance for a question you simply misread wastes an evening.
Go back to the source when you cannot state the rule in your own words after reading the explanation, when two of your errors in the same session touched the same rule, or when the explanation and your memory of your notes disagree. In that third case one of them is wrong, and finding out which is worth the twenty minutes.
Everything else can be handled with a line in the log and a re-test.
Re-testing, which is the part everyone skips
A reviewed question is not a learned question. Put every item from piles two, three and four into a re-test set and do it again a week later, cold, without looking at your notes first. Anything you get wrong the second time goes into a third pass at two weeks.
This is spaced retrieval and it is the least glamorous thing in this entire cluster. It is also the thing that separates candidates whose mock scores climb from candidates whose mock scores plateau in the forties.
Here is the opinion. Review time should exceed practice time, roughly two to one, and almost nobody does this. An hour of questions deserves two hours of review across that week including the re-test. Candidates instead do three hours of questions and twenty minutes of reading explanations, then wonder why the volume is not converting into marks. Doing a hundred questions properly beats doing five hundred badly, and it is not close.
The concession: the two-to-one ratio is a heuristic from watching how candidates improve, not a measured finding, and we would not pretend otherwise. If your log shows patterns disappearing after a lighter review, run lighter. The ratio matters less than the fact that review is a scheduled activity rather than something you do while tired at the end of a session.
Reviewing a full mock
Same method, one addition. Leave it overnight. Reviewing ninety minutes of questions immediately after sitting them produces shallow notes, because you are tired and because you still remember what you were thinking, which feels like understanding. Come back the next day and the notes get sharper.
Then do the two tallies: one by pattern, one by topic. Pick the two weakest topics and the single most frequent pattern, and make those your next week. Everything else waits. The detailed protocol is in mock exam strategy.
Common questions
How long should I spend reviewing practice questions?
Roughly twice as long as you spent answering them, counted across the week and including a re-test. Ten questions in 15 minutes deserves around half an hour of review, plus the same items again a week later without notes.
Should I review questions I got right?
Only the ones you guessed or were unsure about. Mark uncertainty while you work, because retrospective judgement about whether you knew something is unreliable. Questions you answered confidently and correctly need no attention at all.
What should an error log contain?
Date, topic, the wrong-answer pattern that caught you, and one sentence diagnosing your own reasoning. Keep it in one document rather than splitting by topic, because the value comes from seeing the same pattern recur across different parts of the syllabus.
How do I stop making the same mistake twice?
Re-test. Put every wrong or guessed item into a set and redo it a week later, cold. Reading an explanation creates recognition, not recall, and recognition is exactly what fails you under time pressure in the exam room.
Should I go back to the Ordinance for every wrong answer?
No. Go back when you cannot state the rule in your own words after reading the explanation, when two errors in one session touch the same rule, or when the explanation contradicts your notes. Otherwise a log entry and a re-test are enough.