IPSC shooters rating

[Dmitry Sidorkin]

Интересная такая страничка: http://www.ipscrating.com/
Я даже себя там нашёл на… 14каком-то-месте (между 1400 и 1500… :biggrin: ). Есть возможность фильтрации по странам. Учёт по матчам выше 3 уровня, но, почему-то, не по всем…

[Dmitry Sidorkin]

«Of cause you want to know how this rating been calculated. It is not a simple process, but here the short explanation of algorithm:

— We initialize the rating with the results of the last known World Shoot as most trusted and representative source.
In the every subsequent match results we are looking for the set of «key shooters», whose results are already exists in the rating.
— We convert the particular match result for the each shooter to global rating percents. If shooter A has 80% from the current world leader in the global rating, and shooter C has 70% from shooter’s B result in the local match, then we predict his global rating as 0.8*0.7 = 56%. Such calculations been made to the each of the «key shooters» and the average value been used.
— Current rating of the particular shooter is calculated as an average value between previous and predicted percents (so called «moving average»).
— Every time we have a new rating leader we need to normalize all the data to the 0..100% basis.

The more match results processed for particular shooter (see the number in the last column of the rating table) — there is more trust for the rating result.

There can be some minor problems with the name recognition due to the fact, that if stats officer didn’t use any kind of electronic match registration system (like MakeReady, for example) he can easily make a mistype while entering the data for the match in the WinMSS software. We use very complex algorithm for eliminating doubles, but nothing is perfect. If you’ll find such doubled results — write us, please.

Some other points:
— Shooter’s result can be high enough in the particular match, but if he’s/she’s performance was not good enough this can lead to decrease of the global rating.
— Shooter A can beat shooter B in the particular match. But if shooter B beats A several times before that there no guarantee A will be higher than B in the global rating. This will depend on math, which eliminates sudden peaks of performance. The more consistently A will shoot better that B — the more higher results he/she will get.
— Shooter can have 100% result in the local match and be the leader of the world rating, but poor performance against other top shooters can lead to rating points loss, or at least to the negative value in the «Last Match Performance» cell.
— «Last Match Performance» column displays rating changes for the particular shooter before the normalization been taken place. It is not always equal the plain difference between the last rating value and the previous one.» — это описание принципа формирования рейтинга…

[Автор]

Сообщения