We’ve all heard jokes at tournaments about players who lost their first match and found themselves at the bottom of the event ladder, only to win their next three games and have a shot at the podium leaving a trail of decisive victories against less skilled opponents. Unfortunately these aren’t always jokes: submarining can be a serious problem in ranked gaming, especially when a ranking system rewards tournament placings and other similar metrics. Thankfully, the Elo system that underpins 8Realms is highly resistant to submarining tactics. As the header image clearly shows, Arpad Elo crafted an algorithm that takes a torpedo to malicious submarining, and prevents player ratings being unduly benefitted by this phenomenon.
Understanding Submarining
Submarining occurs when a player loses an early match, sometimes deliberately, and face weaker opponents in the following rounds. As they climb back up the ladder by defeating these lower-ranked players, they ultimately secure a strong placement in the tournament. Often, observers – generally other players – note that the player has performed much better than they would have if they had won their first round pairing and faced more challenging opponents.
Nobody can prevent submarining completely, but 8Realms helps mitigate its effects in several ways:
- Penalising players who lose against a challenging opponent and then beat a series of less skilled opponents
- Using performance ratings to reward players who performed better than their skill level would suggest, even when playing against submariners
- Seeding first round pairings properly, encouraging challenging play right from the start of an event
- Incentivising players to fight every match rather than trying to ‘game’ a single tournament
The Role of 8Realms Ratings in Seeding
Using 8Realms Ratings to seed the first round pairings can significantly reduce the impact of submarining. The system, originally developed for chess but now used in many sports and competitive games, assigns a numerical rating to players based on their performance against opponents of known ratings. By pairing players based on their 8Realms ratings in the first round, tournaments can ensure that competitors face opponents of similar skill levels from the outset, reducing the incentive to submarine. This initial seeding based on ratings minimises the likelihood of early mismatches, promoting fair competition and accurate rankings. Most importantly for TOs, players are more likely to compete fairly if they know that their rating will be considered in seeding, as deliberately losing would harm their rating and subsequent pairings.
The Impact of Submarining on Elo Ratings
To show just how much submarining can negatively impact a player’s score, take a look at this example which lifts the hood on the maths behind 8Realms.
Suppose Alarielle has an Elo rating of 1600, and Belakor has an Elo rating of 1400. Using expected performance calculation, the system determines how likely Alarielle is to win this match:
EA = 1/(1 + 10(1400 – 1600) / 400) ≈ 0.76
According to 8Realms, Alarielle is expected to win this match 76% of the time. If she wins (sparing you the equations), Alarielle gets an 8 point increase to her 8Realms rating. Good for her, but not a huge increase as 8Realms judged that she had a higher skill level so was expected to win the match.
But now, let’s say that she tries to submarine, deliberately throwing a match against a slightly lower skilled player. According to 8Realms, she only has a 24% chance of losing fairly, so this drops her rating significantly:
ΔRA = 32(0−0.76) ≈ -24
Alarielle’s rating has dropped by 24 points! Elo ratings are time-sensitive, meaning that round 1 results affect round 2 scores. The 8Realms system is built to take account of this, so her 8Realms rating has now dropped significantly. But can her submarining strategy pay off? Let’s assume that Alarielle now faces a series of opponents with ratings around 1200, given that she is now lower down in the field:
EA = 1/(1 + 10(1200 – 1600) / 400) ≈ 0.91
This means that Alarielle is expected to win these matches 91% of the time, so a victory will have minimal effect on her 8Realms rating. Each victory would only affect her rating by around 2.88 points. If she goes on to win all four of those, she will ultimately only gain around 12 points in total from this strategy.
So, Alarielle goes 4-1 in this tournament, but at the cost of her 8Realms rating. Playing fairly would have gained her 8 points, but submarining will go on to lose her around 12 points! This example shows how 8Realms overcomes the effect of submarining with scores that truly reflect a player’s skill level.
TOs can dig deeper into this, and look at whether Alarielle did as well as expected in this event. Playing fairly against slightly weaker opponents would give a performance rating of 1500, a solid finish. Meanwhile, submarining would give her a performance rating of only 1360 even though she won four of her games. This is mathematical evidence that she did have a much easier time than other players. More importantly, performance ratings – which can be used for finding event MVPs and selecting for teams – are resilient to attempts to submarine.
Conclusion
Submarining can harm an 8Realms rating more than playing fairly, as the system penalizes unexpected losses more than it rewards expected wins against weaker opponents. By using 8Realms ratings to seed the first round pairings, tournament organizers can reduce the incentive for submarining, promote fair competition, and maintain the integrity of the tournament. Players are encouraged to compete at their best, knowing that their performance will be accurately reflected in both their standings and ratings.