Concours Banque PT common mistakes
Concours Banque PT · Understand the traps and how to avoid them.
In maths, the candidate writes a final result or a limit without ever stating the domain of validity, the hypotheses of the theorem used, or the continuity/derivability conditions, so a correct computation is scored as unjustified.
Under time pressure the candidate treats the calculation as the goal and mentally skips the preamble, because in class the hypotheses were already given by the teacher. At the concours the grader is checking whether the candidate can produce the conditions themselves, and an unjustified invocation of a theorem (interversion limite/intégrale, convergence dominée, théorème des accroissements finis) is worth almost nothing.
What to do instead
Train a fixed written ritual: before each theorem application, write one line 'hypothèses: ... vérifiées car ...' and one line giving the interval or domain. Do timed exercises where the first 30 seconds are spent listing hypotheses on scrap paper, then copy them into the answer. Keep a one-page checklist of the theorems used most (continuité, dérivabilité, intégration, séries) with their exact conditions and check it after every mock paper.
In physics, the candidate plugs numbers into a formula before doing the dimensional and limiting-case analysis, and never checks the homogeneity of the result or its behaviour at the boundaries.
The habit comes from terminale-style exercises where the numeric answer is the deliverable. At the concours, a large share of the marks sit in the analysis of the expression (dimension, sign, asymptotes, limiting regimes), and a wrong unit or a non-physical limit reveals an algebraic error that the candidate could have caught alone.
What to do instead
Impose a three-step routine on every literal result: (1) unit check written explicitly, (2) two limiting cases named and interpreted physically, (3) sign and order of magnitude. Practise by re-deriving the same result and predicting the limit before computing. In mock papers, reserve the last five minutes for a 'contrôle d'homogénéité' pass over every boxed formula.
In sciences industrielles, the candidate starts writing equations of motion or equilibrium without first drawing a clean isolated-system diagram (bilan des actions mécaniques extérieures) with the chosen frame, the point of reduction and the torsor notation, then mixes reference frames mid-calculation.
The candidate believes the algebra is the hard part and rushes the schema, often because the diagram feels like decoration. Without a clear isolation, forces are forgotten (poids, frottement, effort d'un vérin), and changing frame without saying so silently corrupts every projection that follows.
What to do instead
Adopt a rigid template for each mécanique question: name the système isolé, list every extérieur in a table with its point of application and torsor, state the repère and the point of réduction, only then write the theorem (résultante or moment). Draw the diagram large and label it. During revision, redo corrigés by covering the solution and rebuilding only the isolation step for ten different mechanisms until it takes under two minutes.
In automatique, the candidate answers a stability or performance question by intuition on the step response alone, without reading the poles, the margins on the Bode plot, or the characteristic equation, and confuses gain margin with phase margin.
The Bode and Nyquist plots look familiar, so the candidate skims them and gives a qualitative sentence. The vocabulary (marge de gain, marge de phase, degré de stabilité, précision) is often memorised as labels rather than as measured quantities, so the numbers read off the graph are attached to the wrong criterion.
What to do instead
Rebuild the link between each plot and each criterion on paper: for a given transfer function, compute poles, then read gain and phase margins with the crossing frequencies explicitly marked, and state what each margin protects against. Do a drill of ten transfer functions where you predict stability before plotting, then verify. Keep a glossary card linking each performance spec (précision, rapidité, stabilité, dépassement) to the quantity that measures it and the plot where it is read.
In français-philosophie, the candidate treats the dissertation as a catalogue of authors and quotations strung together, or as a personal opinion piece, instead of building a problem (problématique) from the terms of the subject and arguing a progression of theses.
The candidate has memorised the works on the programme and wants to prove it, so quotations become the skeleton. But the concours jury grades the construction of a problem and the conceptual work on the notions in the subject, and a quotation dropped without being analysed and tied to the argument counts for little or is penalised.
What to do instead
Practise the first twenty minutes of the dissertation only: decompose the subject word by word, identify the tension or paradox, and formulate a problématique with two or three possible theses. For every quotation in your notes, write one sentence of what it proves and one of its limit. In mock essays, cap yourself at one quotation per paragraph and require each to be followed by an explicit analysis linking it to the thesis.
In LVE, the candidate prepares answers as written texts learned by heart and delivers them as a monologue, then fails to react to the examiner's follow-up questions or to reformulate when a word is missing.
Written preparation feels safe and the oral is feared, so the candidate over-scripts. At the oral, a recited text sounds flat, breaks down at the first unexpected question, and prevents the interaction the jury is actually testing: comprehension, reactivity and the ability to sustain an exchange.
What to do instead
Prepare bullet-point notes only, never full sentences, and rehearse aloud with a timer and a partner who asks unpredictable follow-ups. Practise paraphrase strategies ('what I mean is', 'in other words') and circumlocution for missing vocabulary. Record yourself once a week and check fluency, pronunciation of technical terms and whether you actually answered the question asked.
In TP and experimental technique, the candidate manipulates before writing the protocol, forgets to record the uncertainties and the conditions of measurement, and produces a single measurement presented as a definitive value.
Laboratory time is short and the candidate wants results, so the notebook is filled in afterwards from memory. The concours evaluates the experimental method: whether the measurement is repeatable, what its uncertainty is, and whether the conclusion is supported by the data. A result without uncertainty or without stated conditions cannot be defended at the oral.
What to do instead
Impose a written protocol before touching the equipment: what is measured, with what instrument, at what resolution, how many repetitions, what is held constant. Record raw values live in a dated table with units and instrument precision. Compute the uncertainty explicitly and state the final result as value ± uncertainty with the method used. Rehearse a few standard measurements (calibration, repeatability, linear fit) so the routine is automatic.
At the oral entretien, the candidate presents only grades and generic qualities, cannot explain why a particular chapter or project interested them, and has no specific answer when asked about their motivation for the school or the programme.
The candidate prepares the written concours and treats the entretien as a formality, assuming good results speak for themselves. The jury is testing self-knowledge, coherence of the project and knowledge of the formation, and vague answers signal that the candidate has not thought about why they are there.
What to do instead
Build a one-page personal file: two or three concrete academic or project experiences with what you did, what you learned and what it changed in your orientation; one difficulty and how you handled it; and a precise reason for each school applied to, drawn from its actual programme and laboratories. Rehearse aloud with a teacher or peer playing a sceptical examiner, and prepare three questions of your own about the formation.
Across all written papers, the candidate spends disproportionate time on one hard question, leaves entire pages blank or unfinished, and does not read the whole subject before starting.
The reflex of finishing what was started is strong, and a difficult first question can consume half the allotted time. Concours subjects are long and graded across many independent questions, so abandoning easy marks later to chase one hard question is a structural error, not a detail.
What to do instead
In the first five minutes, read the whole subject and mark each question as easy, medium or hard. Answer all easy questions first, then medium, then return to hard ones with the remaining time. Set a per-question time budget and move on when it is exceeded, leaving a marker to come back. Rehearse this strategy in full-length timed mock papers so the triage becomes a habit.
The candidate presents calculations with no structure: no labelled results, mixed units, intermediate values without rounding control, and a final answer buried in the text or missing its unit.
Working on scrap paper and copying fragments into the copy leaves the grader to reconstruct the reasoning. The candidate underestimates how much presentation affects the score, especially when the final numerical value is wrong but the method was sound, or right but impossible to locate.
What to do instead
Adopt a fixed layout: state the question number, write the formula literally, substitute with units, give the result boxed with its unit and a sensible number of significant figures. Keep a separate clean draft for calculations and transfer only the final structured lines. In mock papers, ask a peer to find each final answer in under ten seconds; if they cannot, the presentation needs rework.