Explore

SRS
SRS

--IVRL Board

--IVRL Board

The Problem

When IVRL was founded, we spent a chunk of time experimenting and developing the SRS formula. SRS is the Seasonal Ranking System, the values behind IVRL’s matchmaking and ranking system. As season 1 started, we kept a close eye on weekly results to make sure it was performing as expected. On a per-match basis it was performing fine, however over the break when looking at overall tendencies, a problem was noticed.

Let's outline the different things that should happen

1) Winners receive SRS, Losers lose SRS

2) Stronger victories should provide more SRS

3) Underdogs should receive more SRS for a victory than the favorite (given the same match outcome)

4) SRS gains should be proportional to SRS difference for Underdogs winning

5) SRS gains should be inversely proportional to SRS difference for Favorites winning

6) Ties should provide SRS to the underdog, bringing them closer to their opponent’s SRS without passing them

7) Ties should not provide more SRS than any victory would

1-4 were all performing as expected. However #5 was not occurring. We noticed that favorites were receiving more points for victory based on the SRS gap between them and opponent. Effectively, favorites winning received more points than they should be.

At a per match level this wasn't caught, however when looking at more data comparatively this became easier to spot. Additionally, the effects of this only presented themselves with decent enough SRS separation to begin with. Early in the season when teams were closely clustered together the effects weren't noticeable.

The SRS change around ties have also been adjusted. The first part of #6 was correct, but in some cases the teams would result in equal SRS after a tie which could change the order in the standings. This has been corrected in the adjusted SRS formula. Ties were also providing a lot more points than they realistically should. They were accurate, but a tie should never provide more points than a victory would. This has also been corrected.

While it's season one and bumps were expected along the way, our intention was to run all of season 1 on one ranking system and improve it for season 2. When competitive integrity is at the core of IVRL, maintaining a stable ranking system is key. However, with a fundamental problem in the ranking, we couldn't continue using it. Fixing it swiftly is the better choice. The long season 1 will allow for any small irregularities from this change to smooth out by the end of season for a fair and stable ranking ladder.

Now that we've established the problem and the why, where do we go from here? Favorites, very often won their matchups. A little less than 50% of the 143 matches to date fall into this category.

Using the adjusted SRS formula we have recalculated all of the matches retroactively from week 1. You can view the simplified data here to see how it affects your team. Upon our review, the general order of the standings remained the same with some teams fluctuating 2-3 positions each week versus the previous standings. This is a result of less SRS being given when favored teams win their matchup which takes into account the strength of their opponent. Favored teams still won their matchup, however they do not receive more SRS than they should. For example: a 30 point favorite winning 10-2 is a relatively expected result. They will still be rewarded for their victory, but not to the point of jumping several ranks.

What’s changed

You can see this simplified spreadsheet below to see the week to week standings changes. Most standings are still intact even if you notice the raw numbers changing.

[Click this below]

⁠

Loading docs.google.com⁠

⁠

Let’s read about how the new SRS formula works!

⁠

Magic behind the Math :SRS

SRS (Seasonal Ranking System) is IVRL's matchmaking and ranking system. In this post we hope to explain how it works and what to expect. There’s a few fundamental truths to SRS that have been the case since day 1.

Fundamentals

SRS is a zero-sum balanced system. What the winner gains is what the loser loses

SRS is relationally relevant. A leaderboard is created with teams, but raw SRS loses meaning if not compared with peers in the same region

3 weeks of TBD SRS to gain stability

Factors

SRS changes from the match results are directly related to 4 things.

1) round win - smallest point gains

2) map win - median point gain

3) match win - largest point gain

4) opponent SRS - adjustment relative to each team’s incoming SRS

⁠

Round win points are the smallest potential SRS gain, but when combined they have the potential to make up the majority of the points gained a match SRS change. A decisive 10-0 victory over a team will provide a relatively large SRS gain over a 10-8 win with the winning team taking both maps and the match. Total points from round wins are adjusted based on the incoming teams’ SRS differential.

Map win points are awarded to the team that wins the map. Match win points are awarded to the team that wins the most rounds. Any rounds or maps that your opponent wins will remove points from the potential SRS gain that your team can receive. This leads to every round and map being a major factor to the final SRS change from the match results.

Opponent SRS is also a consideration to prevent inflation of SRS results based on potentially unfavorable matchups. An underdog win over a favored team will trade more SRS points compared to a favored team’s win over an underdog with the same match scores and incoming SRS.

⁠

Ties

There are 2 cases for matches. The first case is the most common: the match has an underdog and a favorite based on ranking/SRS. The second case is when the two teams are very close to each other to the point of being considered equals.

When 2 teams tie, the underdog receives SRS points for outperforming expectation. They will inch closer to their opponents SRS but NEVER surpass it. When the 2 teams that tied, were considered even, no points will be awarded as that was the expected result. The points awarded on a tie are handled uniquely. Taking the case of a 1 point underdog win, removing the match win bonus, yields the points awarded on a tie.

In the case of a tie due to an official points adjustment (ie 4 to 4), the underdog will receive be awarded the map points based on their differential from the favored team. If the team SRS differential is minimal / tied, the no points will be awarded for match ties. When two tied teams tie in their matchup, no one has won!

⁠

How many points will I get?!?

To get the most points, you will want to win against strong opponents, in strongest fashion. If you are to lose, try to win as many rounds as possible, and try to win a map! SRS is tuned to provide good enough gains, but not too many to drastically change standings on a week to week basis. Stability is important to the ladder just as much as mobility is. It’s also key to keep the spread within matchmaking ranges.

Let’s look at a 16 week sample season for a mythical team that has been made up for this exercise. You can name them whatever you want. The numbers have been rounded for reader convenience. We hope that seeing this will help understand SRS a bit better and know what to expect in your matches.

Note: Numbers are floored to nearest 5

⁠