Loading Revision Schedule Builder…
| Subject | Needs it |
|---|---|
40
20 days of revision at 2 a day
Sessions are shared out by the largest remainder method — Alexander Hamilton’s 1792 rule for apportioning seats between states by population. Each subject takes the whole part of its share and the sessions left over go one each to the subjects with the largest fractions, so the totals always add up exactly and no evening is invented or lost. They are then dealt out day by day to whichever subject has fallen furthest behind its own share, which is what stops a heavily weighted subject arriving as one solid week.
The exam day itself is not counted as revision time, because you are sitting the paper on it. Nothing here has any opinion on how long a session should be or how many a day you can sustain — those are not figures anyone has measured on your behalf, and a planner that supplied them would be dressing up a guess.
List the subjects, give each one a number saying how badly it needs the attention, name the day revision starts and the day the paper is sat, and say how many study slots a day you are prepared to fill. The builder works out how many slots exist between those two dates, shares them out in proportion to the numbers you gave, and then deals them across the calendar one day at a time so that no subject arrives as a single undivided lump.
Forty slots split between subjects weighted five, three, two and one come to 18.18, 10.91, 7.27 and 3.64 slots each, and nobody revises for 0.18 of an evening. Rounding each figure on its own overshoots or undershoots the total nearly every time, which means either inventing an evening that does not exist or losing one that does.
The fix is old and has a name. Each subject takes the whole part of its share, and the slots still unassigned go one apiece to the subjects with the largest fractions left over — the largest remainder method, written by Alexander Hamilton in 1792 for apportioning congressional seats between states by population, and still the standard answer to this exact shape of problem. It balances exactly by construction, so the plan can never gain or drop an evening in the rounding.
Knowing that chemistry deserves eighteen slots does not tell you when to sit them, and eighteen consecutive evenings of chemistry is the wrong answer for reasons everyone who has tried it already knows. Each slot on the calendar therefore goes to whichever subject has fallen furthest behind its own allocation, measured as the fraction of its own total still unplaced, with ties settled by the order you typed the subjects in. Two equally weighted subjects alternate, and a heavy one never swallows the opening week whole.
The result is deterministic: the same inputs produce the same calendar every time, so a plan you print on Monday is the plan you still have on Thursday. Nothing is shuffled, and there is no randomness anywhere in it.
Days are counted from the start date up to but not including the day of the paper, since that day is spent sitting it. The count is done on whole calendar days in a fixed reference frame, so a clock change in the spring cannot shorten a day to twenty-three hours and quietly lose it from the total — and February gets its extra day in a leap year without anything special being written.
What the builder refuses to supply is the part everyone wants: how long a slot should be, how many are sustainable, or which subject genuinely needs the most. Nobody has measured those for you, and a planner that filled them in would be presenting its own guess in the costume of a calculation. Every number this page divides up is a number you supplied.
Any scale, as long as it is consistent, because only the proportions matter. Weights of five and one divide the slots exactly as fifty and ten would. Most people find a one-to-five scale easiest to be honest on, since it forces a decision about which subject is genuinely the worst.
Because the leftover slots have to land somewhere. When the fractions are equal the tie is broken in favour of the subject listed first, which is arbitrary but deterministic — the alternative is a plan that reorders itself between one look and the next.
Set its weight to zero and it drops out of the division entirely while staying on the list, so you can see what you decided rather than deleting the row and forgetting it existed. Its slots redistribute among everything else immediately.
The calendar is capped at a hundred and twenty days and says so when it truncates. Beyond that a day-by-day schedule is a fiction: you do not yet know what your timetable looks like in March, and printing it as though you do only produces a plan you will abandon.