TCF Canada simulations: how to build one that actually predicts your score

Ask most candidates what a simulation is and they describe a pile of questions. That definition is the reason so many simulations predict nothing. The questions are the smallest part of it. A simulation is a set of conditions, and it is the conditions that decide whether the number at the end has any relationship to the number you will get on the day.

Get the conditions wrong and you will produce a score roughly a band above what you can actually deliver, feel reassured by it, and book a test date on the strength of a measurement that was never valid.

The four conditions that do the work

Everything that makes a simulation predictive comes down to four rules, and each one is boring individually.

• One sitting. All four abilities in a row, in the order the real exam uses, with only the breaks it allows.
• No pausing. The clock does not stop for any reason, including good ones.
• Nothing to lean on. No dictionary, no notes, no phone, no conjugation table, nothing open in a second window.
• First attempt only. Whatever you produce is the answer. No going back once a section is finished, and no polishing a text the next morning and counting the polished version.

Break any one of them and you are still practising, which is fine, but you have lost the right to draw a conclusion from the result. Keep all four and even a rough paper made of mismatched material will tell you something real.

Building the paper

Assemble it in advance and then do not look at it until you sit down. A simulation you built this morning and sit this afternoon is not a simulation, because you have already seen the reading texts.

For compréhension orale and compréhension écrite, collect enough items to fill the real section length and use audio you have not heard. For expression écrite and expression orale, you need prompts of the right type, and the types matter more than the topics. The TCF Canada asks for particular kinds of production rather than free composition, and practising the wrong kind of task carefully is worse than practising the right kind roughly.

Take the current section structure and timings from the France Éducation International site rather than from a forum post or an old PDF, and rebuild your template whenever the format is revised. Français Prep is an independent preparation service and is not affiliated with France Éducation International, so the official page is the authority on the shape of the exam and this page is only the authority on method.

Marking it, and where the conversion is honest

The receptive half converts honestly. Mark it strictly, convert each ability separately using the TCF Canada score chart, and let the score converter handle the arithmetic. Two abilities, two levels, recorded separately.

The productive half does not convert honestly, and you should stop pretending otherwise. Whatever you write down for your own expression orale is a guess shaped by how you felt during the task. What you can record instead is evidence: the recording itself, the text exactly as produced, whether you finished, and how long the silences were.

Then take your four levels and look only at the lowest one, because each ability converts on its own and the weakest decides where you land against NCLC 7. A simulation reported as a single average number has destroyed the only piece of information it was carrying.

Reading a plateau correctly

Run the same simulation format three times across six weeks and look at the trend rather than any single result. Three outcomes are possible and they mean different things.

Rising means the method is working, and the correct response is to change nothing. Falling almost never means your French has got worse; it usually means the conditions tightened, or you were tired, or the material happened to be harder. Check the conditions before you conclude anything.

Flat across three sittings is the one that matters. A plateau does not mean you have reached your ceiling. It means you have reached the ceiling of what your current practice can teach you, which is almost always because the practice is dominated by things you already do well. Doing more listening when listening is already your strongest ability feels productive and moves nothing.

A four-week cycle that uses simulations properly

Simulate at the start of a cycle, not the end. A simulation run as a final check before a real sitting arrives too late for its findings to be acted on, which is the most common way candidates waste one.

• Week one: full simulation under the four conditions, marked and converted the same day.
• Week two and three: work aimed only at the weakest ability, with shorter timed sections rather than full papers.
• Week four: a second full simulation to see whether the weakest ability has moved, and to check the others have not slipped.

The study planner will lay that out against your target date, and the exam timer handles the section splits so you are not watching a clock and answering questions at the same time.

One honest limit to all of this. A self-built simulation can measure your receptive abilities well and your productive abilities barely at all, and for most candidates the productive abilities are the ones sitting lowest. That gap is not something more simulations will close. If you want to know what a marker would actually do with your writing and speaking before you pay for a real sitting, that is what a 15-minute diagnostic call is for.