Rating-Budget Tournament System

[Mats Winther]

Wouldn't it be ideal if a complete pairing schedule could be generated before the tournament starts, regardless of the number of players and rounds? In fact, this is possible by introducing the concept of a rating budget:

• Rating budget = number of scheduled games × the average rating of the player's group.

Each player is assigned a personal rating group consisting of players with similar ratings in ranking order, covering roughly one-third of the field. The primary objective of the algorithm is to fulfil each player's rating budget as closely as possible by keeping the average rating of their opponents close to the average rating of their rating group.

The system aims to give each player at least half of their games against opponents from their own rating group, although this cannot be guaranteed in every individual case. The result is a fully predefined schedule for the entire tournament that closely emulates a Swiss tournament. Unlike the Berger system, this method scales well and can be used even in tournaments with well over a hundred participants. For blitz and rapid chess events, this means shorter waiting times between rounds and significantly less administrative work for tournament organizers.

Rating-Budget Tournament System - free online pairing system

[Christopher Kreuzer]

Er, why? How would an underdog win through to play the favourite in the final round in this sort of set up? If those on 4/4 in a 5-round event were in different rating pools, they would not end up meeting. The whole point of a Swiss is that you don't know who you will play as the tournament progresses. It is a feature, not a bug.

EDIT: I may have misunderstood, but to be fair it is not easy to understand what you are proposing here!

[Joey Stewart]

It sounds almost like once all the entries have been received then players are broken up into mini all play all sections of similar terms ratings with all the pairings immediately listed rather then a traditional Swiss where pairings are based on who won previous rounds.
Not sure what the "rating budget" part is all about but I'm sure that players would be relatively happy to be facing similarly graded opposition and no bottom feeders who aren't ever in contention for top places and are only there to try and nick some points

[Mats Winther]

Er, why? How would an underdog win through to play the favourite in the final round in this sort of set up? If those on 4/4 in a 5-round event were in different rating pools, they would not end up meeting. The whole point of a Swiss is that you don't know who you will play as the tournament progresses. It is a feature, not a bug.

EDIT: I may have misunderstood, but to be fair it is not easy to understand what you are proposing here!

(Thank you for acknowledging my contribution.) In that respect, it works much like a Berger system: two players on 4/4 are not necessarily paired against each other in the next round. The Swiss principle, whereby players with the same score are paired from the very start of the tournament, is essentially an algorithmic convention designed to simplify pairings. It is based on the same underlying idea as ladder systems or King of the Hill.

Today, however, there is little reason to adhere rigidly to this principle. As I have already demonstrated with my Berger-Swiss system, modern computing allows us to create tournament formats that are far more dynamic and engaging. Many players find the traditional Swiss "climbing the ladder" mechanism stressful. At a given rating level, a player will typically defeat a lower-rated opponent only to face a stronger opponent in the next round and lose, repeating this cycle from round to round.

In the Rating-Budget Tournament (RBT) System, the tournament leaders will typically be paired against one another regardless, because they are the strongest players. Their rating budgets are always higher than those of lower-rated players. To minimise the remaining budget for each player, the algorithm is naturally drawn toward such pairings. Moreover, because it seeks to schedule roughly half of a player's games within their own rating group, encounters between the leaders become even more likely.

Competing against players of comparable strength is far more engaging than the familiar Swiss pattern of defeating an opponent who is too weak in one round, only to lose to an opponent who is too strong in the next. Many players find this constant up-and-down movement through the field less satisfying than sustained competition among their peers.

Nevertheless, there is no requirement that the two tournament leaders meet. In that respect, RBT is no different from the Dubov Swiss, where the leading players are likewise not guaranteed to face one another.

The obvious advantage of RBT, especially with classical time controls, is predictability: players know both their opponent and their colour in advance. This is a highly valued feature that RBT shares with the Berger system.

That said, it is evident that the system is particularly well suited to rapid and blitz tournaments. It offers a much smoother playing experience, since there is no need to wait for all games in a round to finish, results entered and the next round paired. The entire schedule is already determined in advance, which greatly reduces both delays and administrative workload. As there are typically more rounds in rapid and blitz tournaments, the leaders are almost certain to meet.

[Mats Winther]

It sounds almost like once all the entries have been received then players are broken up into mini all play all sections of similar terms ratings with all the pairings immediately listed rather then a traditional Swiss where pairings are based on who won previous rounds.
Not sure what the "rating budget" part is all about but I'm sure that players would be relatively happy to be facing similarly graded opposition and no bottom feeders who aren't ever in contention for top places and are only there to try and nick some points

You can use my program to simulate a tournament with 50 pre-programmed sample players and verify that this is not how the system works. The players are not divided into separate mini-tournaments; rather, the entire field remains interconnected throughout the event. If a player has already spent a large portion of his rating budget on higher-rated opponents, he will typically be scheduled in advance to face a lower-rated player, since his remaining budget is insufficient to accommodate another expensive high-rated opponent.

That said, the system is unlikely to produce perfect pairings. But does such a thing as a perfect pairing system even exist? RBT is intentionally non-deterministic and can generate different pairings for exactly the same field. In a club setting, however, this is arguably an advantage rather than a drawback.

Note that it can also run locally, by downloading the page. The only downside is that one risks missing an update.

All my online pairing systems are described here: Chess Pairings: A Review of Monrad, Swiss, and Berger.

[Roger de Coverly]

In the Rating-Budget Tournament (RBT) System, the tournament leaders will typically be paired against one another regardless, because they are the strongest players.

I think too much is being expected of the rating system. It's not uncommon for ratings not to reflect current strength. A player with a low rating but comparable playing strength to other potential tournament winners would presumably get easier pairings.

[Mats Winther]

In the Rating-Budget Tournament (RBT) System, the tournament leaders will typically be paired against one another regardless, because they are the strongest players.

I think too much is being expected of the rating system. It's not uncommon for ratings not to reflect current strength. A player with a low rating but comparable playing strength to other potential tournament winners would presumably get easier pairings.

True, an underrated player would get easier pairings than he deserves. This is a weakness of the system. But I don't think it matters much for rapid and blitz tournaments, at least not at club level.

[Ian Thompson]

With a Swiss pairing system, prizes are usually awarded to the players who scored most points because they played better than players who scored fewer points. Obviously, it's not perfect because players who start poorly and finish strongly tend to play weaker opponents, on average, than those who start strongly and are amongst the leaders for the whole tournament.

With this pairing system, the number of points scored does not correspond in any way to standard of play; for example, in your sample tournament, the player seeded 10th (rated 2265) needs a TPR of 2490 to score 5.5/7, while the 40th seed (rated 1815) needs a TPR of 2040 to score 5.5/7. How would you suggest prizes are awarded with this pairing system?

[Mats Winther]

With a Swiss pairing system, prizes are usually awarded to the players who scored most points because they played better than players who scored fewer points. Obviously, it's not perfect because players who start poorly and finish strongly tend to play weaker opponents, on average, than those who start strongly and are amongst the leaders for the whole tournament.

With this pairing system, the number of points scored does not correspond in any way to standard of play; for example, in your sample tournament, the player seeded 10th (rated 2265) needs a TPR of 2490 to score 5.5/7, while the 40th seed (rated 1815) needs a TPR of 2040 to score 5.5/7. How would you suggest prizes are awarded with this pairing system?

I'm not sure I understand the concern. Lower-rated players will be paired within their own rating group in approximately half of the rounds and will only occasionally meet top players. The exact outcome depends, of course, on the rating distribution of the field. As a result, lower-rated players have a good chance of achieving a solid score. This is not fundamentally different from what happens in either Berger or Swiss tournaments.

It is easy to test the program by adding different sample participants and evaluating the resulting pairings. You can move back and forth between rounds to examine if the pairings make sense. The choice of the number of rounds has a big impact.

[Chris Goodall]

Why would a strong player agree to compete in a tournament where a weaker player can play worse than him and finish on a higher score? That's essentially a handicap tournament, isn't it?

If you want to know which of the two 2500 players is playing the best chess this week, having them play each other gives you vastly more information than having them both play a 1900 player because something something rating budget. Standard Swiss pairings extract the maximum information from every game. If you're leading the tournament, you get a tough draw, but you can be confident that your closest rival is getting a tough draw too.

[Mats Winther]

Why would a strong player agree to compete in a tournament where a weaker player can play worse than him and finish on a higher score? That's essentially a handicap tournament, isn't it?

If you want to know which of the two 2500 players is playing the best chess this week, having them play each other gives you vastly more information than having them both play a 1900 player because something something rating budget. Standard Swiss pairings extract the maximum information from every game. If you're leading the tournament, you get a tough draw, but you can be confident that your closest rival is getting a tough draw too.

It is the head-to-head contest that matters, much like in the gladiatorial games. Chess is not solely about determining who is the strongest player. If you want to generate interest, you have to appeal to people's competitive instincts.

In my opinion, a top player will normally win the tournament, even though pairings are largely made within relatively narrow rating bands. In rare cases, however, a lower-rated player may be paired against a significantly higher-rated opponent. This can occur when the lower-rated player has faced many low-rated opponents and therefore needs to spend more of his rating budget, while the higher-rated player has used too little of his.

The system would not work particularly well if lower-rated players frequently faced higher-rated players, as this would lead to unbalanced and unfair pairings. Simulations using realistic results suggest that the top-rated players usually finish near the top of the standings, often closely followed by a lower-rated player. It should be borne in mind, however, that lower-rated players tend to be more inconsistent in their performances and will occasionally drop points against opponents with significantly lower ratings.

A lower-rated player will be paired outside their own rating group in roughly half of the rounds and will then often lose to a higher-rated opponent. However, if the same player manages to win all of their games within their own rating group, which extends approximately one-third of the field above and below them in the rating order, they may still achieve a very high final placing.

The Swiss principle of pairing players on the same score from the very beginning of the tournament is, at its core, an algorithmic convention developed to simplify the pairing process. It is based on the same fundamental idea as ladder systems and King of the Hill.

There are now few reasons to adhere uncritically to the traditional Swiss principle. As I have already demonstrated with my Berger-Swiss system, modern computing technology makes it possible to create tournament formats that are considerably more interesting. In Berger-Swiss, the score leaders are not brought together from the outset. Instead, they may meet later in the tournament, during the Swiss phase, when such encounters carry greater sporting significance.

However, I am not entirely convinced that RBT is the ideal format for tournaments with classical time controls. For that reason, I have included the option of continuing the event with additional Swiss rounds, as in the Berger-Swiss system.

[Alex Holowczak]

Isn't this just a version of how UEFA handles the Champions League draw, but with a much greater emphasis on rating differences?

UEFA the 36 teams into 4 pots of 9 based on their rating, and each team plays two teams at random from each of the 4 pots.'

You've just generalised it a bit in terms of the number of rounds, number of participants, and tried to minimise the rating differences more than UEFA has.

Why does your greater degree of emphasis on rating differences have an advantage over the UEFA approach?

[Mats Winther]

Isn't this just a version of how UEFA handles the Champions League draw, but with a much greater emphasis on rating differences?

UEFA the 36 teams into 4 pots of 9 based on their rating, and each team plays two teams at random from each of the 4 pots.'

You've just generalised it a bit in terms of the number of rounds, number of participants, and tried to minimise the rating differences more than UEFA has.

Why does your greater degree of emphasis on rating differences have an advantage over the UEFA approach?

RBT is a much more advanced pairing system that aims to make the tournament an exciting experience for all participants by optimizing pairings so that players face many opponents of similar strength while still being challenged by stronger opposition.

Stop speculating and try it out. There are 50 sample players that can be added. See how it works.