New Rating System Trial
The current Badminton England National Rankings and Player Gradings are used together to measure and track players. They are essential to underpinning the competition pathway used for signposting players to the right level of competition, seeding and selection of players for competition, motivating and retaining players, and supporting players to progress.
The findings of a recent review showed there are limitations of the current ranking and grading systems outlined below. It does not inherently make them bad systems and whilst it would be possible to make changes to improve them, the recommendation of the review was to investigate the potential for a different and more robust system which can cater for players at all levels of the sport.
A significant amount of work has taken place to bring to life a full trial of this new system, called the ‘Rating System‘.
The key attributes of the new rating system:
- Award players based on the quality of match wins or losses, using as many results as possible to improve accuracy.
- Be a dynamic and responsive system that can be updated on a regular basis as required.
- Include all player and competition results in one system.
- Ability to easily expand to include results from more competitions, such as local tournament, local leagues, schools and university, to create a richer and more in-depth rating system for more players.
Competition is at its best when matches are close and the games are ’50/50′, and sport is also full of unexpected results when we are having a good day or a bad day. Therefore, we should not expect any system to be 100% accurate.
How Rating works – the basics
Players will have a ‘rating’ which is a number of rating points that signifies their skill level.
- High number = strong player
Your ‘rating’ changes based on your head-to-head results against other players/pairs. When you win you gain rating points, when you lose you lose rating points.
- Play a higher rated opponent = Gain lots for winning, lose little for losing.
- Play a lower rated opponent = Gain little for winning, lose lots for losing.
Two players play each other at singles:
| Simon Parker (441) | Chris Tustian (549) | 18-21 21-17 14-21 |
| -19 | +19 |
Simon has a 186 singles rating, and Chris has 195. Chris beat Simon – so Chris gains 19 rating and Simon loses 19 rating.
Chris is more highly rated than Simon, so we expected him to win. As a result Chris doesn’t gain very much, and Simon doesn’t lose very much.
They play each other again straight away:
| Simon Parker (427) | Chris Tustian (583) | 21-17 21-15 |
| +25 | -25 |
Simon won this time and because Chris is higher rated the rating shifts are bigger.
- Since the tier of tournament or league is not required for the rating calculation, any team and individual results can be included. Therefore the ratings can easily expand to include results from other competitions such as local tournaments, local leagues, schools and universities.
- International results are not currently included, we are reviewing the best way to include these.
- The ratings can be used to signpost players to the appropriate level of competition as well as being used for selection and seeding of players, in very much the same way the current ranking and grading system are used. For example;
How Rating works – in detail
The ratings are derived based on a number of settings, which can be changed to affect the resulting rating lists. The published lists represent just one version which will be fine tuned using the feedback received.
- Rating calculation uses a recognised system called the ELO system which was developed in competitive chess and then adopted by other sports and online gaming, including the new world tennis number and squash levels systems.
- A player’s ability is shown as a bell curve, which represents their typical playing range based on whether they are playing well or badly, and most of the time you would be expected to be in the middle of your curve. A stronger player’s curve will be the same but at a higher score range, but there will be some overlap where an unexpected result may occur.
- There are two formulas used to calculate players rating; 1) The first formula predicts the outcome of your game based on the ratings of the players, 2) The second formula updates your rating based on whether you won or lost.
- Each match yields a transaction of points won or lost between the players, so there is an equal transaction. If a player does better than expected they will increase, if they do worse than expected they will decrease, and if they do about as expected there will be little difference. So there is no benefit for strong players competing against weak players.
- Rating is therefore a measure of your ability relative to the population, so a rating of 800 this year may not be the same as a rating of 800 in 5 years. Even if your ability stays the same your rating can go up or down relative to everyone else. Therefore a rating rank order can be given to indicate your overall position.
- Players with fewer results are considered to have a less accurate rating (since we know less about them).
- This will be taken into account in the calculation, to avoid new/less known players having a disruptive impact on the rating of existing players.
- Points won or lost will be weighted based on the player confidence of your opponents known level. A player’s confidence factor will be based on the number of match results held over a recent time period on a sliding scale, for example:
No < 6        then   Player confidence factor = 25%
No 6-10           then     Player confidence factor = 50%
No 11-15      then     Player confidence factor = 75%
No > 15       then     Player confidence factor = 100%
- E.g. Player A (Rating 500) v Player B (Rating 200)
Player A has 100% confidence factor. Player B has 25% confidence factor.
Player A loses, however as Player B’s confidence factor is low, the number of points lost is reduced to 25%.
Player B wins, since Player A’s confidence factor is high, the number of points won is not affected.
- Rating Decay:
- After a period of inactivity a players points will decrease to reflect that they have not been playing any eligible tournaments. It will decay slowly over a period of time to a maximum reduction.
- This will allow players who are injured or take a break to restart at a relative level to their standard.
- Inactive players:
- Players who are inactive for a prolonged period will be considered as no longer competing. This is currently set at 365 days.
- They will be hidden such that they have no rating position, however their score will continue to decay in the background, but will reactivate once playing again.
- Players who are winning matches against known players far above their level, will be given a rating point boost to fast-track them to the right level.
- Whilst the system awards greater win points against players with a much larger score difference, there is a maximum that this will reach based on the system settings.
- This will be calculated by reviewing recent match results :
- Discarding expected losses below a certain threshold and discarding those with low opponent confidence factor.
- Of the remaining matches, the points won against much stronger opponents, will be multiplied to boost them upwards to a more relevant rating.
- No change is made to the scores received by opponents in any of these matches. (Their losses should be minimised already by the players low confidence factor.)
- Players in doubles events are listed individually on the rating list and not as a pairing.
- The player confidence factor is based on the average of all the other players in that match (opponents and partner). This means there can be no advantage/disadvantage from playing with a partner whose player confidence factor is low.
- Your rating is adjusted as an average of yours and your opponents rating versus the opposing pair’s average. Any points won/lost are equally shared.
Next Steps
- Webinars – sign up to one of the webinars to learn more.
- Feedback – give us feedback on the new system by 31st July.
- International results not currently included – we are reviewing the best way to include these.
- Review – if successful this could replace the current ranking and grading systems, but not earlier than January 2025.
