The overall score
One number for a season, made of four. Here is what goes into it and what each part is worth.
Four parts, one number
A rating says how well you do per problem and nothing about how many you have done. A solve count says the opposite. Neither is the whole picture of a season, and a board that ranks on one of them alone rewards a particular way of playing. The overall score puts four things together instead.
| Part | Weight | What it rewards |
|---|---|---|
| Rating | 35% | Solving hard problems, and solving them right first time |
| Problems solved | 35% | Turning up and working through the archive |
| First solves | 20% | Getting there before anybody else does |
| Solving time | 10% | Getting there quickly |
Every part is a z-score
The four figures are measured in different units: a rating is in the hundreds, first solves are in the tens, a solving time is in minutes. Adding them as they are would let whichever has the biggest numbers decide the board on its own.
So each one is turned into a z-score first. A z-score says how far above or below the field you are in that part, measured in standard deviations, which is a unit all four share. Zero is the average solver of the season. Positive is above it, negative below. The four z-scores are then combined in the proportions above: 35, 35, 20 and 10.
Two things follow from that, and both are deliberate:
- Your score moves when other people solve. It is a standing in a field, not a total you accumulate. A quiet week for you while everyone else is working will move you down without you doing anything wrong.
- The gaps are real. Somebody twice as far ahead of the field scores twice as far ahead. Ranking would flatten that, and a season's best run deserves to look like one.
What counts, and when
- The season only. Everything is counted from the day the season opened. Nothing carries over from the last one, and nothing you did before it counts here.
- Rated problems only. Problems you gave up on do not count as solved, here or anywhere else.
- Solving time needs a handful of timed solves before it means anything, so it is left out for anybody who does not have them yet. Their remaining three parts are scaled back up to a full score, so nobody is penalised for a part they have not earned a reading on.
- Solving something nobody has solved yet is worth the 20%. It is rarer than the other three parts, so it is held to the same range as them rather than being allowed to decide the board by itself.
- Accounts under review are left off the board entirely while they are looked at. Their profiles stay readable.
Reading a score
Because the middle of the field is zero, the number reads the same way every season whatever the field is doing:
| Score | Where that puts you |
|---|---|
| 0 | The middle of the season's field |
| 1 | Clearly above it, top of the board's shoulder |
| 2 | Among the best handful of the season |
| 3 | A season somebody will remember |
A negative score is not a mark against you: half of any field is below its own average by definition, and a new account with one solve starts near the bottom of a board it has barely played on. The way up is the same for everybody, and it is the four parts above.
While it is in beta
The score is on the leaderboard now so that it is familiar by the time it decides anything. It becomes the ranking that counts when Season 2 opens on 1 October. Until then nothing on the site is decided by it.