About us


Looking for an accurate rating, not just any rating? Look no further than Peterman Ratings!

At Peterman Ratings, we utilise an impartial, data-driven approach to generate just and precise ratings for an array of head-to-head win-loss sports, including tennis, badminton, squash, table tennis, padel, and pickleball. Our advanced statistical analysis caters to both singles and doubles matches.

Whether you're a player of one or more of these sports, our ratings system can provide valuable insight into your skill level relative to other competitors at the local, state, national, or even international level. Even if you don't win trophies or prize money, you can earn Peterman points by competing!

Club managers for any of these sports can also benefit from utilising Peterman Ratings, as our easy-to-use platform allows for accurate ratings and assessments of player performance and progress based on event results entered into the system.

Competitors can be added to the system for one or more sports, and historical match results, player performance, seeding, and event eligibility are all meticulously tracked for convenience. Contact us to learn more about how Peterman Ratings can positively impact your sports community!

At Peterman, we aspire to provide ratings for players of all levels, not just the elite. By comparing a player's standing to others who participate in competitive fixtures, we help individuals understand their place in the larger sports community. Our gradual development of each player's rating points profile facilitates objective-based selection of teams for challenges, rather than subjective opinions. Additionally, a rating-based system can attract other clubs to compete and create more legitimate challenge matches. As more recreational players appreciate the nuances of their sport, it becomes more engaging and entertaining. Increased popularity may also attract sponsors, thus promoting the sport further. It is crucial to note that ratings serve not only to boost individual egos but also to increase engagement, socialization, and publicity for recreational play.

Great clubs need a great rating system!

For a record of historical match results, player performance, seeding, and event eligibility, rely on Peterman Ratings.

What is it


Peterman Ratings is an automated rating system that provides sports clubs with accurate player ratings. Accessible from anywhere, it is easy to use and requires no additional software. It also serves as a content management system, providing information on events, matches, and results that are backed up daily.

Unlike many other rating systems, Peterman Ratings is designed specifically for 1:1 Win-Loss sports such as singles and doubles. Traditional rating systems for these sports are often subjective and lack a statistical basis. Peterman Ratings, on the other hand, provides accurate ratings by tracking a player's progress as they play matches in pennants and tournaments. Every match entered is processed to compute the most accurate rating for the player or team.

Peterman Ratings uses a Bayesian model that considers a player's playing history, recent events, and the strength of the opponents they faced to calculate the most accurate rating for each player. This approach makes the ratings distribution for top players more symmetrical and normal than those produced by other systems, which tend to be skewed.

The system also gives points for sets won and the strength of the opponent faced, even if the player lost the match. It caters to both singles and doubles matches in all listed sports, and reports all team matchups involving the player on their individual profile page under "Player Teams". The system lists them in ascending order of rating uncertainty.

Although Peterman Ratings is currently intended for grassroots-level individual sports, the system could be used for higher levels. Unlike the ATP/WTA ranking system, it does not award points based on the stage of the tournament or its prestige. The creators of the system believe that the use of a Bayesian model, rather than the Elo rating algorithm used by other rating systems, allows for more accurate ratings. In fact, the probability distributions of the three systems for the world's top men's tennis players in December 2022 show that Peterman Ratings produces almost symmetrical ratings, while the ATP and UTR systems have positive skews.

Below is a ranking of the 30 highest-ranked male tennis players according to the PMR method, which has been applied to ATP tournaments since October 2019. Please note that this ranking was last updated in December 2022.



Below is a ranking of the 28 highest-ranked male badminton players according to the PMR method, which has been applied to BWF tournaments since January 2023. Please note that this ranking was last updated in Feb 2024.



User Categories


The functionality of Peterman Ratings is divided into four user groups: General User, Contest Manager (CM), Administrator, and Site Owner. The General User group has basic access and can freely browse the site's content. The functions within this group apply to all user groups.
Click here to read the General User Manual

To join Peterman Ratings, click on the Sign In button and create an account for your club. Once you are approved as a CM, you can create a club and add members. You will receive a link to the User Manual - Contest Manager.pdf. If you do not receive a welcome message in your primary email, be sure to check your junk mail.

Every club must nominate at least one CM who will be responsible for updating their players' wins and losses for tournaments or pennants. This job is made easy with Peterman Ratings' intuitive graphic user interface. For further information, please refer to the "Notes for Contest Managers" section below.

To help you get started, Mr. Peterman has provided two explanation-by-example videos:
Click here to see how to Create Account
Click here to see how to Create a Club

What it Costs


Peterman Ratings offers free access to player ratings and past competition outcomes, and this will remain unchanged in the future. Additionally, club administrators can upload match results to the website for free during a one-month trial period. However, after the trial period, a nominal annual fee will be applied. Don't wait any longer, take advantage of the free access and trial period now!

Methodology


The rating system used by Peterman Ratings estimates a player's strength by analyzing their match results and constructing a probability distribution that describes their playing ability. This process is based on Bayesian Inference, which determines the degree of uncertainty in a given fact. As a player competes and gains more wins and losses against opponents of varying strengths, their rating becomes more accurate.

A player's rating is a somewhat arbitrary number that varies based on their performance in matches. Ideally, players with equal strengths should win or lose at the same rate when playing against each other. However, a player's actual playing ability can differ significantly from their rating, and this difference is minimised with more matches played. The rating scale ranges from 1 to 4000, and the uncertainty of a rating reflects the confidence held in its accuracy.

At Peterman Ratings, the uncertainty of a player's rating is represented by a second number next to the mean rating, separated by a +/- sign. If the uncertainty is greater than 100, little confidence should be placed in the rating. A visual cue is used to indicate the level of confidence in a player's rating, with ratings marked in red for low-level confidence (uncertainty > 80), amber for medium-level confidence (uncertainty between 40-80), and green for high confidence (uncertainty < 40).




The uncertainty value can be seen as the factor that determines the stability of a player's rating. The higher the uncertainty, the more susceptible the rating is to sudden fluctuations. Players with high uncertainties are typically new to the system or those who have returned to competition after a long break.

The system assumes that a player's ability is normally distributed around a mean, forming a bell curve with the highest likelihood at the center. In this context, the mean, along with the standard deviation, represents the estimated player strength, with the standard deviation acting as the uncertainty.

For instance, if a player has a mean rating of 1000 and an uncertainty of 50, there is a 68% chance that their actual playing strength falls between 950 and 1050 (within 1 uncertainty of the mean). There is a 95% chance that the actual playing strength falls between 900 and 1100 (within 2 uncertainties of the mean), and a 99.7% chance that the actual playing strength falls between 850 and 1150 (within 3 uncertainties of the mean).



As players participate in events, their mean rating is expected to increase or decrease depending on their performance. However, it's impossible to know for certain whether a stronger player will defeat a weaker one. When a weaker player beats a stronger opponent, it's called an "upset." The greater the difference in playing strengths, the more likely the weaker player is to lose. If two players have the same mean rating, the chance of an upset is 50%. But when the difference in rating is 600 points or more, the chance of an upset is nearly zero.

For most clubs, "Novice" players will start with a mean of 700, "Intermediate" 1100 or "Advanced" 1500. Every new player mean will have a high uncertainty to indicate the rating is inaccurate. As they play more events the rating will converge (to their correct rating).
The uncertainty provides important information about a players rating before the match. Aside from the player being new to the ratings system, a high uncertainty (indicated by a red rating) tells me they might have had a lay-off period, because the uncertainty increases when the player stops playing. He will probably be weaker than his mean suggests or he could have attended a training boot camp and gotten very good. The plus/minus before the uncertainty means both contingencies are taken care of at Peterman Ratings.

If a player has a significant layoff period, their uncertainty will increase at a nominal rate per year (their mean will increase too, albeit at a smaller rate). This is why the uncertainty at the top of the players/team page will be higher than the Final Rating uncertainty (after their most recent event).






According to statistical theory, we are 95% confident the player is within the range: Rating ± 2xUncertainty. With every match played, information is gained and the uncertainty decreases or remains the same. When the player is having a break from competitions, his uncertainty increases at a set rate per year. For players who play less frequently to quickly find their place on this scale, the point change is higher for players with a larger uncertainty.The point change depends on the mean difference and the set score, but also significantly on their uncertainties.

In some systems, the same amount of rating points will be deducted from the losers rating, regardless of how close the match was. Here at Peterman Ratings, we believe its fairer when the score line is taken into account. By way of example, if a superior player defeats an opponent two sets to one when the system believes they should have won straight, the winner won't receive as many rating points as they would have, and the weaker player won't lose as many points (because they took a set off a better player). If the strategy of a stronger player is to sacrifice a game early to test a strategy or game plan, a reduction in his rating would be in order. It is statistically valid and means stronger players will be disincentivised to toy with weaker players; they will fight for each point and this will result in time savings. In the same way, a player who defeats an opponent in straight sets will get more ratings points compared to if they dropped a set.

Based on statistical theory, the Peterman Ratings system expresses 95% confidence that a player's skill level falls within the range of Rating ± 2*Uncertainty. This uncertainty value decreases or remains the same with every match played. However, during a break from competitions, a player's uncertainty increases at a fixed rate per year. Players who play less frequently experience a higher point change if they have larger uncertainty. The point change is influenced by the mean difference, set score, and the player's uncertainties.

It is at the granular level Peterman Ratings comes into its own. Not only does it meticulously process each match, but it also considers the set score a crucial factor in determining the change in ratings points for a match. Unlike other systems that deduct the same number of rating points from the loser's rating regardless of match closeness, Peterman Ratings takes the scoreline into careful consideration. For example, if a stronger player wins a match 2-1 against an opponent they were expected to defeat in straight sets, the winner will receive fewer rating points than usual, and the weaker player will retain more points. This approach disincentivises stronger players from experimenting with weaker players and encourages them to fight for every point.

In the rating system, it's important to note that it's not a zero-sum game. This means that the points gained by the winner may not necessarily match the points lost by the loser. The exception to this rule is when both players have identical levels of uncertainty.

By clicking on an event hyperlink under "Events," you can access a detailed match-by-match report of a specific tournament or pennant fixture. The page includes the event description, hosting club, type of match (singles or doubles), date, and location. The winner is displayed on the left and the defeated on the right, following the order of data entry. The report shows the Previous Rating, Rating Change, New Rating, and Set Score. Matches should be listed in chronological order, with the oldest ones appearing first, and the most recent ones at the end. If the event was a tournament, the round of 32 matches would appear at the top, with the winner and runner-up at the bottom. The website is hyperlinked, allowing you to click on a player or team name to view their profile page, which summarises their performance in each event. Additionally, if a player competes against the same player more than once in an event, you can track their ratings trail as they play subsequent matches:



In the two-match event described above, Max and Charlie's initial means were almost identical, but Max had higher uncertainty compared to Charlie, which was nearly twice as much. This meant that we were more certain about Charlie's rating, leading to Max losing 73 points in the first match, while Charlie gained 29 points. In the second match, Max's uncertainty remained high, and he gained 117 points, while Charlie lost 25 points, technically making it an upset.

To encourage regular play, the system favors players who play frequently, as their increase in uncertainty will be small. Regular players are less likely to lose more points in the event of an upset than players who have been dormant for an extended period before the event. This aspect of the model is designed to incentivize players to compete regularly rather than staying out to protect their ratings.

Furthermore, to encourage dormant players to become active again, their mean will increase by a small amount each year, but they will not receive these points until they play again. This mechanism, combined with the increase in uncertainty outlined above, ensures that points are injected into the system, preventing deflation of ratings. Deflation in ratings systems can lead to players being stronger than their calculated ratings suggest.

Notes for Contest Managers


To ensure effective management of clubs, each club must nominate at least one Contest Manager. If a club is large, it can have more than one Contest Manager to handle all responsibilities. Contest Managers are responsible for maintaining players' details and uploading events such as tournaments and pennants for their club. They can also add or remove another Contest Manager for the same club and edit player details and events only for their club.

Administrators have the authority to manage all clubs, including adding or deleting another Administrator or Contest Manager. They can update players' wins and losses and correct false information. Additionally, Administrators can update events and players of all available clubs in the system. Presently, the Site Owner is the only Administrator.

Before entering players into their club, Contest Managers should check if the player already exists on the system by navigating to Club Administration > Club Members > "Add Existing Players". They should search for the player's family name and, if found, add them to the club. If not found, they can click the "Add New Player" button and fill out the form. It is the Contest Manager's responsibility to verify the identity of new club members who express interest in playing a tournament or pennant using photo ID. This is to prevent the same player from having multiple ratings on the site due to different names or misspelled names. The "One Name - One Rating" policy applies.

Contest Managers must choose an initial prior for each of their club players, based on whether they are Novice, Intermediate, or Advanced. A custom rating can be chosen, but it must be within the range of 650-2000 and paired with an appropriate rating uncertainty. For new clubs using Peterman Ratings, all Div 3 players can be given an initial rating of "Novice," all Div 2 can be given an initial rating of "Intermediate," and all Div 1 can be given an initial rating of "Advanced." It is good practice to play unknown players against a known player in the green zone to estimate their ability and assign an appropriate rating. Once the system has processed some singles events, new players can be slotted in according to their estimated abilities.

When a club adopts PMR, individuals, whether they are longstanding players or new, will be given a rating that best describes their ability according to the Contest Manager. High-uncertainty players will need to play several matches to get an accurate rating (colored green: uncertainty<40). As the group plays more matches and most players' ratings settle into the amber and green zones, new entrants' ratings will become accurate more quickly.

Contest Managers should upload events to Peterman Ratings as soon as possible after they are complete. If a player is a member of another club using the same system, there could be discrepancies in rating and date ordering.

When submitting events, they must have some sort of definite structure, such as round-robin, single or double elimination, team match, league match, ladder, or winner stays. Events must be open to some segment of the public, and publicity is required, such as mailing or emailing entry forms, posting a notice on a club website, handing out flyers, or posting a notice in the club. Two players playing against each other multiple times at the club is not an acceptable event. All matches that are part of the event must be submitted, not just some of them.

Once a player's details have been added for one type of sport (like tennis), there is no need to add their details again if they play another sport listed on the system

Data Input


To input the matches in the correct order, there are two methods available. The first is to enter them "on the fly" as they are being played, while the second involves using comma-separated values (CSV) data for easy updating of player ratings.

To use the first method, select "Upload Event" under Administration and upload frequently during the session to avoid losing information. You can then edit the event when a new match is entered. For the second method, wait until all matches have been played and select "Upload CSV" under Administration. A link is provided to obtain a table of club members and their PMR ids, which is updated automatically when a new player is added to PMR. Copy-paste this table into a spreadsheet worksheet, convert it to CSV, and then paste it into the designated field. Fill in all relevant fields and click upload. If you have any questions or need assistance, please contact PMR at the email address below.

After the Contest Manager uploads the new event data and the matches have been processed, all players who participated in the event will receive an email with their win/loss status for each match played and their updated rating. The email, titled "Your Rating Has Been Updated," will appear to come from the website, but the "reply to" field will show the Contest Manager's email address. It is important to inform your players about this website and to check their junk mail if they are not receiving messages. If the messages are in the junk mail, they can be labeled as "not junk." Players can also choose to unsubscribe if they do not wish to receive any further communication.

Finally, if you are new to this process, Peterman has created a series of short explanation-by-example videos to help get you started:

Click here to see how to Add New Players to your club and add existing ones
Click here to see how to Add Existing Players to your club who havn't yet got a rating for a new sport
Click here to see how to Add a New Player to your club at the Upload Event stage
Click here to see how to Add a New Player with a long or hyphenated name to your club
Click here to see how to Upload a Singles Event
Click here to see how to Upload a Doubles Event
Click here to see how to Edit An Event
Click here to see how to Add another Contest Manager for your club
Click here to see how to Add another Contest Manager for your club (Alternative method)
Click here to see how to Reset Your Password
Click here to see how to add CSV data (and add another player)

Drawn matches, Did Not Finish, Retirement due to injury, Forfeits and Walkovers


In the Peterman Ratings system, match outcomes are limited to either a win or a loss. Draws are not recognized, which is consistent with the nature of win-loss sports. If a match is abandoned due to time constraints, injury retirements, or bad weather (in tennis, for instance), it cannot be included in the Peterman Ratings. Likewise, forfeits and walkovers should not be recorded in the Peterman Ratings.

Acknowledgments


I would like to express my gratitude to my mentor Barrie for his invaluable statistical contribution. I am also grateful to Tom Leslie, Carl Love, and "Acer" (who prefers to maintain anonymity) for their assistance in coding Maple, which was of great help. As the project needed to be deployed on the World Wide Web, it was presented to third-year IT students at UTAS for consideration. I am thankful to the professor in charge for providing this fantastic opportunity. The website was constructed from scratch using a blend of web technologies, and the math engine behind it is fueled by Maple™.
Maple is a high-end software, and only the finest quality is acceptable. I would like to express my gratitude to the exceptional team members who worked on this project. Marcus served as the Project Manager, while James and Grant were the Co-Lead Programmers. Harinder led the Documentation efforts, Adib was the Client Liaison, and Yusuf and Mingxin were the Co-Lead Designers. Once the project was completed in October, James, one of the developers, was hired to refine the "rough diamond" that was handed to me. I extend my heartfelt appreciation to all of these individuals for making my vision a reality.

For my mum




4th April 1934 - 3rd May 2020




Mr Peterman BAppSc

info@petermanratings.com
+

Account Information

Only for club manager, if you are looking for your rating, event results or club, please visit the Home page

By clicking below and creating an account, you agree to our account

Terms and Conditions
+

Forgot Password

Enter your email address and we'll send you a password reset email

Email sent!
+

Notification

Notification placeholder text here to be changed in JS.