About The Ratings

TruVolley is a rating system that uses a modified Glicko algorithm to rate players. The following is a high level overview of how this works.

A player’s rating has 2 main parts:

  1. Rating

    This is the numeric value that encapsulates a player’s “skill”. In our system, this number is on the scale from 0-14, with most players falling in the 2-8 range. Highly skilled players will be above 8, pros above 10, and the best players of all time achieving a rating of 13+. There is technically no cap on how high a rating can go, but it is extremely hard to keep climbing when you are at the top.

  2. Confidence

    Confidence in this system indicates how confident the system is about a player’s rating. 100% confidence means the player is active and has played enough games for their rating to settle in to a place where we can confidently rank them amongst their peers. When a player’s confidence is 0%, wins and losses will change their rating at a much faster rate, as the algorithm is trying to sort them based on their match results. When confidence hits 100%, a player’s rating is still able to change quickly by winning matches against higher rated players or losing to lower rated players.

Next, we get to the match by match updating that the algorithm is based upon. The following are the main points considered in each match that happens:

  1. Expected Result

    Each match that happens has an expected result based on player ratings in the match. When the team’s average ratings are the same, that implies a 50% win probability for each team. If a team has an average rating of 1 point higher than the other team (and all confidence is 100%), that implies a 75% win probability for that team, and 25% for the other team. A team rated 2 points higher (with all confidence at 100%) has a 95% win probability.

    When there are players involved who have a confidence of less than 100%, it gets more complicated, but a general rule is as follows:

    Winning or losing against players with low confidence will affect your rating less than if all players have full confidence.

  2. Actual Match Result

    Next, the actual result of the match is considered. A tie rewards 50% win credit to each team.

    Matches that are played to one set factor in the score of that set. Winning by 2 points (i.e. 21-19) awards 75% win credit to the winning team and 25% win credit to the losing team. Winning by 11 points or more (i.e. 21-10) awards 100% win credit to the winning team and 0% win credit to the losing team. There is a linear scale between an 11 point margin and a 2 point margin.

    Matches that are played to best of 3 sets but won in 2 sets also factor in the score of that set. Winning by 4 points (i.e. 21-19, 21-19) awards 75% win credit to the winning team and 25% win credit to the losing team. Winning by 22 points or more (i.e. 21-10, 21-10) awards 100% win credit to the winning team and 0% win credit to the losing team. There is a linear scale between a 22 point margin and a 4 point margin.

    Any match that goes to 3 sets rewards 66.67% win credit to the winning team, and 33.33% win credit to the losing team.

  3. Match Value

    Match Value is the difference in the expected result and the actual result of the match.

    For example, if a team has a 55% win probability going in to the match, and wins the match 21-19, 21-19, the match result is 75%. The match value for this team is 20%.

    Any match value that is greater than 0 will influence your rating positively, whereas a negative match value will influence your rating negatively.

    It is possible to win a match and lose rating points. If the expected result is 95%, and the actual result is 75%, the match value for the winning team will be -20% because they were expected to do better, and their rating will drop slightly.

This all happens with every single match played. Over the course of time, and hundreds of thousands of matches later, these match by match movements create a pool of highly accurate player ratings that can be used to compare players against one another.

Another important point is that all matches are evaluated the same. The ratings and confidences of the players involved are what influence rating changes. However, there are a few things that can be controlled by us, which we have spent hundreds of hours testing to provide the most fair and accurate rating system for everyone involved. Those include

  1. Maximum Rating Deviation - this is the initial value we set when a player’s confidence is 0%. A higher value here means we start with less confidence and it will take longer for a player’s confidence to become 100%. This value is the same for every player.

  2. Minimum Rating Deviation - this is the value we set for when a player’s confidence is 100%. A higher value here means more volatility in ratings. A lower value results in what is known as “elo-hell”, when it starts to feel impossible to improve your rating. We have chosen this value based on a middle ground where teams that get hot move up quickly, but it is not too penal to lose a few matches as well. This value is the same for every player.

  3. Initial Rating - because we have all players on the same scale from juniors to the top professionals, it doesn’t make sense to initialize everyone at the same point. Players who appear in pro level matches as their first match in the system start higher, 12U players start much lower, and everything in between. While this is important at a mass scale, it’s not actually important where one individual starts. When confidence is 0%, it is possible to make huge jumps up or down the ladder, as the algorithm is intended to do this. Trying to game the system by starting in a tournament above your skill will not help you, you will end up exactly in the same spot as you otherwise would have.

  4. Rating Decay - unfortunately, inactivity results in a decay in rating. This is done for fairness and applied the same across the board for all players. 9 months of inactivity starts the process, and ratings decay at a rate of 0.5% per month until confidence hits 0%. After that, the next match played will reinitialize the player as if they are new.

Keep in mind when you see these ratings, there will be upsets often. That is necessary and expected, as a team rated one point higher than another is only expected to win 75/100 of the matches they play together.

 

FAQ

  • TruVolley is the largest Elo-based beach volleyball rating system in the world. Ratings are based on who you play and how much you beat them by, not by overall tournament finish.

    Every match that you play goes toward your rating, and playing players of different skill will effect your rating accordingly.

    For example, a 12.00 rated player that beats an 8.00 rated player will only gain a few points.

    But if the 8.00 rated player wins they will make a big jump up, and the 12.00 rated player will take a hit.

    After only about 10 matches played, players start to get sorted very accurately based on their performance. The more matches a player plays, the more accurate this rating becomes.

    This is similar to the system currently used in games like online chess. Some adjustments were made to the algorithm to make it work for beach volleyball.

    No preference is given to anyone. Every match is analyzed on an individual basis, so there are no weights or multipliers added ever. Beating players rated higher than you is the best way to grow your rating.

  • We currently have data agreements with bvbinfo.com, volleyballlife.com, and collegebeachvb.com. Any match played that shows up on their platform is automatically included in the dataset.

  • The above ratings are active players only. Active players have a confidence score of 85% or above, meaning the algorithm has a high level of confidence about their rating.

  • TruVolley ratings are based on head-to-head matches only, not tournament results. These ratings automatically account for strength of tournament, whereas FIVB/AVP do not and apply points based on tournament prize money. Neither system is perfect, but TruVolley aims to be a more universal and easy to compare rating.

    On average since 2002, TruVolley is about 4% better in prediction accuracy than AVP at the pro level and 5% better than the FIVB. At the amateur level, TruVolley is about 9% more accurate than the AVP points system.

  • Individual ratings help with the high frequency of partner switches in beach volleyball. While the very top teams typically stay together long term, lower rated players switch partners frequently, which is easily accounted for in these ratings.

    Teams that play together for years will see their ratings start to converge, which is why many partners on the all time ratings list have the same rating as their partner.

  • Because they weren’t always partners! These ratings account for a player’s entire professional career so even one match with a different partner can make ratings very different. As partners play more and more with each other, their ratings will converge, but this can take a very long time in some cases.

 

Questions?

Come to discord and ask away!

 

Individual Beach Volleyball ratings based on performance.