LumiRank

The LumiRank is an algorithmic power ranking organized by Barnard's Loop, EazyFreezie, Stuart98, and kenniky that ranks players in Super Smash Bros. Ultimate. It is the successor to the Panda Global Rankings Ultimate and also incorporates several elements from the OrionRank, which Barnard's Loop and EazyFreezie previously ran.

The power rankings were created in December 2022 under the brand UltRank weeks before the release of its first rankings, following the collapse of Panda, whose team's PGStats division ran the PGRU. The rankings changed their name to LumiRank after they announced their partnership with Luminosity Gaming on August 4th, 2023, days before the release of their first mid-year ranking; the UltRank team remains independent from Luminosity while having access to Luminosity's resources.

General methodology

Ineligible events

Similar to the PGR, not all Smash events are eligible for the rankings. The following events are excluded:

• Weeklies and bi-weeklies.
• Arcadians.
• Pre-major events, unless the event is tiered as a major and is on the same tier or one tier below the main event.
• Events with noncompetitive rulesets
• Events requested by the organizers to be excluded prior to the start of the event

Any event that does not fall under those events must also have a certain number of entrants or points in order to qualify for the TTS, with said numbers depending on the region's multiplier.

Multiplier	Minimum entrants	Minimum points
x1	64	250
x2	48	200
x3	32	200

Events that do not qualify for the rankings may still appear on the TTS's upcoming tournament section under the "U-tier". This tier is primarily to give untiered events exposure, and these events are removed from the TTS once they have concluded.

Score calculation

A tournament's score is calculated by adding the attendance at a tournament to the number of each attendee's player point value if the player has a value, all after factoring in DQs.

There are three different ways players are assigned point values; the highest of the three values is what's assigned:

• A player's placement on the most recent global ranking. Point values do not carry over to the following season, so if a player was ranked lower in the subsequent season, then their point value decreases as well.
• A player's placement in their region's power ranking. The top five players of a ranking are given points (50/40/30/20/10 points distribution). Stronger regions are considered "region seats" and grant 50 points to all of its top five players. Point adjustments are applied with each new power ranking released, however these adjustments have no effect on past tournament values.
• The "Hidden Boss" value, which is split into "Hidden Boss 1" and "Hidden Boss 2":
  • Hidden Boss 1 values are assigned to players who would have made the top 150 in the prior season but were unranked due to low activity. Hidden Boss 1 values were also applied to players who had historical notability, but this was discontinued after the LumiRank Mid-Year 2023.
  • Hidden Boss 2 values are assigned to players who placed highly at a tournament ranked C-tier and above, and past tournaments are retiered in accordance with the new value. The number of points given depends on a player's placements, and the number of players who are given points depend on the tournament's tier. These points are reset at the end of the subsequent season.

Additionally, some regions have either a x2 multiplier or a x3 multiplier: regions gain a x2 multiplier if they did not hold an A-tier event within the past two years or a x3 multiplier if they did not hold a B-tier event within the past two years. If a tournament held an event that crossed one of the thresholds even without the multiplier, then the multiplier would be removed or decreased; a few exceptions are granted to regions that have held large events in the past but otherwise seldom reached C-tier without the multiplier.

Algorithm

For every tournament a qualified player attends, they receive a score determined on a weighted average of the player's wins, losses, and outplacements at that event. All the player's tournament scores are then averaged and weighted to generate an overall score for the player, which is then adjusted for attendance penalties and scaled so #1 has a score of 100 and #50 has a score of 50.

A player's win score is determined by how much the losing players were worth, with an increased worth for players with higher scores. For example, a win on the number one player (with a score of 100) would be worth significantly more than a win on a player with a score of 50. In addition, two new changes were made starting from the LumiRank 2023 full-year ranking: wins are based on a player's overall head-to-head against another player, so a win on a player that they have a negative head-to-head on would be worth less than a win on a player that they have a positive head-to-head on, and players who consistently lost to players with lower scores than them would have reduced gains from wins. Finally, additional weights are given to wins at a player's best events and reduced weights are given at a player's worst event.

A player's loss score is most affected by a player's worst loss at the event, with this score having a higher weight for players with a higher overall score and lower weight for players with a lower overall score. There is a cap based on a player's loss score, so a player outside the top 100 will have no difference in score for losing to a top 20 player than losing to a top 5 player. Repeated losses to the same person at multiple events lowers the overall loss score at each subsequent event that has said loss. On the other hand, players who won an event would have a set value added to the loss score, with this value being determined by a tournament's difficulty and whether a player won the entire event without dropping a set. In addition, players with significantly better losses than wins at a tournament would have a lower loss score. Finally, the weight of losses are different between algorithms, with a full-year algorithm having less direct weight compared to a mid-year algorithm, however on the other hand on the full-year algorithm losses will have a more direct role in determining a tournament's overall weight.

In addition to the above, players who are more prone to upsets will be less valuable than a player with a similar score but is less prone to upsets. This "volatility" is determined based on the difference between the average score of all the player's best qualified wins (all the wins that are at and above the 90th percentile of a player's total wins), and the average score of all the player's worst losses (all the losses that are at and below the 10th percentile of a player's total losses).

Outplacements are determined by two different factors: a positive outplacement, or when the player outplaces another player, and a negative outplacement, or when other players outplace the player; the former receives a significantly greater weight. Both values are determined by the number of losers bracket rounds at the event, so a player who placed 7th will have a greater outplacement score for outplacing a person who placed 33rd than a player who placed 17th. In addition, these scores are also affected by a player's wins to get to their final placement, with a lower placement with more stronger wins creating a better outplacement score than a higher placement with less stronger wins, and a player's score may be lowered further for an "empty run", or a high placement that has significantly worse wins than the placement would suggest.

This article or section is incomplete. You can help SmashWiki by expanding or finishing it. The editor who added this tag elaborates: Weighting

After the player's tournament score is calculated, its weight in regards to the player's overall season score is then determined.

Contrary to popular belief, the algorithm penalizes players with low attendance, with players needing to meet an unspecified attendance threshold to avoid a penalty, and player with a higher score will have a greater attendance penalty. However, this adjusted score only affects a player's rankings, and it does not affect their head-to-head values. An example of this would be ShinyMark in the LumiRank 2024.1 season, who would have been ranked much higher than his final placement without the low attendance penalty. In addition, all players need to attend a certain number of tournaments and events A-tier and above in order to be ranked, with the number changing between season. Those who have attended more than one event but did not meet the basic attendance threshold were instead ranked as an "honorable mention".

External links

• X account