Boards›Esports

Quick Notes: Breaking Ties

Riot·3/6/2018, 9:00:03 PM·1 votes·16,092 views

Chris “Riot Commish” Greeley here—I am the Commissioner of the NA LCS. I’m here to continue our Quick Notes series where we deep dive into topics the community wants to know more about, discuss work in progress, and break down how we make decisions.

So far in Quick Notes, we’ve covered where NA LCS Spring Finals will be, our start time move on Sundays, and updates to the NA LCS Lounge. 

For this post, I wanted to explain what the League Ops team does, then dive into how and why we made changes to tiebreakers in the NA LCS. If you have a burning question for us or a League Ops topic that you want to know more about, let us know in the comments.

League Ops very happily works in the background, but that background work makes the workings of the League less transparent than we would like. In searching for a way to pull the curtain back and provide some visibility, we are going to be trying some stuff out. This is the first shot at that.

What is League Operations?

The NA LCS League Ops team manages the day-to-day back-office operations and maintains competitive integrity of our professional League of Legends leagues including:

  • Reviews, approves, and manages rosters and NA’s portion of the Global Contract Database
  • Writes rulesets and policies
  • Investigates potential rule violations and issues penalties as appropriate
  • Game day tasks such as overseeing the referees and the instances in which Chronobreak can and should be used

At its heart, League Ops is about solving problems. When you are firing on all cylinders, the job is about anticipating the problems and creating the rules and policies to make sure those problems don’t come up. When the world has gone to shit, teams and pro players expect that you will figure out a resolution to an issue that is fair for everyone involved.

Breaking Ties

With a new season comes a new rulebook. As we were drafting our rulebooks, NA and EU decided to go in different directions on some aspects of tiebreaker rules. I wanted to talk a little bit about why and hopefully get some feedback on the two systems as well as get folks thinking about what other League topics they have questions about.

When we deal with tiebreakers, our first question is always “what kind of limitations are we dealing with?” For example, if we are at an international event there are often caps on the amount of games we can play on a day because of restrictions at the venue or from local police or fire departments. At the NA LCS studio, we worry about burning out pros or having a broadcast day that runs into the wee hours of the morning. Those limitations, coupled with our overarching need to do something that upholds competitive integrity, help frame our thinking around how we should deal with ties.  

For two-way ties, the tiebreaker is always pretty simple—if one team won all of their games against the other, they win the tiebreaker. If they split their games, then the two teams play a best of one with the winner advancing.  

Three-way ties have always been the hard case to deal with. In the past, we’ve shied away from playing a round-robin between those three teams because the result wouldn’t necessarily be conclusive (i.e. they all go 1-1 against each other). When you run the possible outcomes, 75% of the time you will get one team that is 2-0, one that is 1-1 and one that is 0-2. Easy and clean. But in 25% of the cases you get three 1-1 teams, and when we don’t have the ability to play more than 3 games, that leaves us in an impossible situation (and in some cases, even 3 games can be too many).

To combat that, we’ve searched for an in-game metric that we can use to rank the teams so that we can have the two lowest teams play each other, with the winner of that game playing the third team. In other words, we give one team a “bye” in the first round. We’ve looked at a lot of different ways to fairly assign this bye, including gold, gold differential, number of kills and kill differential. None of these felt good to us because they can be farmed and would likely have a big effect on the way teams approach games. We have focused instead on how fast one team wins the game (the game victory time).

Because we don’t love the solution we have, we’ve spent a lot of time looking for fixes. This year, we are trying some new things for the NA LCS. We’ve incorporated a strength of schedule component to our tiebreakers to determine who gets the “bye” in three-way ties, as well as for assigning side-selection where we would otherwise have used a coin flip. To figure out strength of schedule, we look at every team’s wins at the end of the season—if you beat the first place team, that win is worth 5 points. If you beat the last place team, that win is worth a half point (and it scales in between). This system provides teams a benefit for winning the tough games. The strength of schedule system also gives us a better way to decide side selection in tiebreaker games (rather than game victory time or coin flips).

Understanding that people tend to compare NA and EU, it is important to note that the EU LCS League Ops team went in a different direction. Their rulebook says that in a three-team tie, they are going to play all the additional games. If the teams are all tied after the round robin, the three teams are drawn into a single-elimination bracket where the team with the lowest game victory time from the single round-robin tiebreaker games has a bye into the finals.

We’re not sure which tiebreaker system is going to feel like the right decision at the end of the season, but we intend to spend a lot of time talking to pros and fans about whether the system in NA feels fair and meaningful.

What do you think? I’d love to hear from fans what feels good and bad about these tiebreakers systems or about any other NA LCS League Ops topics that you’d like to hear about. Let me know in the comments below and I’ll also be monitoring over on reddit.

Chris “Riot Commish” Greeley is the Commissioner of the NA LCS and ran League Operations for MSI and Worlds in 2017. He is a recovering lawyer from New York City who can’t write a rulebook in less than 50 pages. He channeled years of raiding on his Resto Druid into life in the bottom lane dropping wards and shooting glitterlances. You can find him on Twitter and Reddit as @IAmGrza.

 

 

32 Comments

Okshama3/7/2018, 12:35:01 AM7 votes

You call it Strength of Schedule, but a more accurate term would be Strength of Victory. Because all teams play the same schedule (2 matchups against everyone else) if 2 (or more) teams end up tied, they all have an identical Strength of Schedule. Strength of Schedule makes more sense in leagues where it is not a balanced schedule, or when you have too many teams to have head to head games among them all (NCAA Basketball).

TBM Wardless3/6/2018, 9:23:29 PM6 votes

The 2017 Mid Season invitational had the absolute worst tie rule. Win time hardly matters, just ask Team Liquid. It definitely shouldn't matter enough to move from tied between first and third place and getting a lock in first. Both of these solutions are better than 2017 Mid Season invitational rules by far, and both rules are better than 2017 Worlds rule, where we had a team given a bye based on game time and matchups. However, the only real solution is the one that EU LCS is going to do, anything besides letting them play it out is just trying to get it over with. Sure that may make a lot of people's jobs easier but from fan perspective a feeling of true closure is important.

chosenchamp3/7/2018, 6:53:26 AM5 votes

I'm glad to see you're looking into these tie-breaking policies. I've always thought that game time is a bad metric(as well as the bad metrics listed in the article like gold, kills, etc.). To me the only good metric in the game to go by is objectives. I think that objectives tell the overall story of the game. But you can't go by how many objectives the winning team got, or that could skew the game(meaning once they have the win in hand, they just go around taking all the objectives they can before they actually end). So you have to go by how many objectives the losing team secured. And objectives would be worth various points. So all 6 outer turrets would count as 1 point. Dragons could be 1 point. Barons could be 2 points. Inhibitor turrets maybe 3 points. Each inhib taken would be 3 points. Nexus turrets could count for 5 points. So let's show how this would work in a 3-way tie. Say that Cloud 9 beat Echo Fox. And in that game Echo Fox only took the 3 outer turrets(which would be 3 points). Then Echo Fox beat CLG, but CLG took all the 6 outer turrets, 1 dragon, 1 baron, 2 inhib turrets, and 1 inhib(18 points). And lastly CLG beat C9, but C9 took the 6 outer turrets, and inhib turret and the same inhib twice(15 points). Cloud 9 had the most dominating win, only giving up 3 outer turrets. CLG had the "best" loss, having taken objectives worth 18 points. So now you take the points accrued by each team in their loss, and subtract how many objective points they gave up in their win. The higher the points, the better.
For C9: 15 - 3 = 12 points For Echo Fox: 3 - 18 = -15 points For CLG: 18 - 15 = 3 points So C9 would be at the top of the tiebreaker, then CLG, and then Echo Fox at the bottom.

I think this is a lot better metric than the way spelled out in your article. Because I would ask the question, which is better... winning against the 1st rank team in the league, or losing to the 10th rank team. Your article says, "This system provides teams a benefit for winning the tough games." But what about losing the easy games? For example, if C9 loses to ALL the low rank teams, and beats all the high rank teams...is that really better than if we say that CLG loses to all the high rank teams, but wins against every low rank team. I personally think that both are equally bad. But your system is rewarding good wins, but not punishing bad losses. And has no way possible to do that.
Thanks for reading.

Darth Ted3/7/2018, 5:36:34 PM4 votes

Just a note...Having wins against stronger teams, aka the team that has the most amount of points in the tie breaker, would mean that they also lose against easier teams, so are we saying a win vs the first place team is stronger than a loss vs the last place team?

Spoo3/7/2018, 12:28:26 AM2 votes

My ideal would be something like Strength of schedule, then Strength of schedule within teams concerned in the tie.. (for instance a 4-0 team in a three way tie should outrank a 3-1...etc.) I do like the idea of going away from game stats as that favors certain teams over others within metas whereas league is a dynamic sport where the most optimum route may not always be to have the shortest games. Why should we punish a team if they are playing for late game victories if they know their good at closing them out.

I'm really curious how you calculate strength schedule if their are multiple ties? Not likely but possible. For instance a three-way tie for 3rd and three way tie for 6th occurs. I believe we've seen a 4 way tie before way back. Are all six teams SoSchedule calculated immediately or will one of the ties required to be broken for the other to be seeded? Does each 3rd place team just get the (6th place equivalent points for all three 6th rank teams they've beaten) ..and so on. or are they arbitrarily ranked then points allotted.

Freekosuave3/9/2018, 12:00:59 PM2 votes

Head to head should always be the first tiebreaker. If one team goes 2-0 against the other teams involved in a multi-team tiebreaker that should mean they win the tiebreaker. Keep it simple in this regard.

The second tiebreaker is always tricky when you have more than two teams. Since all teams play each other twice, common opponents is not a viable way to look at breaking a tie. There is no way to isolate any particular team and go, ok Team A went 2-0 against some team while Team B only went 1-1 therefore Team A wins the tiebreaker on that merit. You would have to go totally off the deep end and say that you compare against the first place team in the split. In the case of a tie being for first place, you would then select the highest ranked team that is not involved in the tiebreaker and compare the records of each team against that common opponent.

I personally would prefer the "play it out" tiebreaker. This way there is a definitive resolution that does not rely upon some random thing that happened 6 weeks ago because Team A won a game in 45 minutes and Team B won in 48. Sure this means that the tiebreaker may need a tiebreaker within a tiebreaker, but at the end of the day this provides what I as a fan of the LCS want to see. I want to see teams "playing it out" as it were. Lets go ahead and build a nightmare situation, shall we?

EF, C9, CG, and TL take the top 4 spots with the problem being 100T, TSM and CLG end up tied at 9-7 and are in 5th 6th and 7th slots. This leaves OPT, GGS, FLY at the bottom of the table in no particular order. I did say this was a nightmare situation, and for one of the teams tied at 9-7 this means that they end up in 7th and are not even considered for the playoffs at the end of the split. Like I said before, head to head should always be the first tiebreaker. Did one team go 2-0 against all other teams involved in the tiebreaker? Alternatively, did one team go 0-2 against each of the other teams? Lets look at the facts.

Wk 1 - 100T W - L CLG Wk 3 - 100T L - W TSM Wk 4 - TSM W - L CLG Wk 5 - 100T W - L CLG Wk 3 - 100T W - L TSM Wk 9 - CLG ? - ? TSM

So far, 100T is 2-0 against CLG and 1-1 vs TSM. TSM is 1-0 against CLG pending their matchup in week 9. Since CLG needs to win all their games to have this scenario play out in the first place, that would leave them 1-1 vs each other as well. The way I see it in isolation, 100T is 3-1, TSM 2-2, and CLG 1-3. Why would this not be the best way to break the tie at the first level? Comparing the record against common opponents involved in the tiebreaker in this situation provided a clean method to say that Team A is better than Team B who is better than Team C.

Let's rewrite history a little and have 100T beat TSM in week 3. Now you have one team that is 2-0 against the other teams involved and should be in 5th place. This removes them from the three way tiebreaker comparison and now the tiebreaker only involves TSM and CLG which means a single game playoff between the two would take place. That seems fairly straight forward as well. Wait, I said this was a nightmare situation so I think it is time to rewrite history one more time and instead have CLG defeat 100T in either week 1 or week 5. Now all teams are 2-2 overall and 1-1 against each other individually. Now it is time to get ugly. This is a true tie with all teams involved.

Now you have to look at how each team did against every other team outside the tiebreaker and assign the points to each team as well per this tiebreaker rule. I will list the wins for each team so far in the split.

100T - OPT, TL, CG, C9, FLY +2 wins inside the tiebreaker means they split their remaining schedule of GGS, OPT, EFX and CG to be 9-7 TSM - OPT, GGS, EFX, OPT, FLY +2 wins inside the tiebreaker mean that they win two of the three games remaining against TL, C9, and GGS (they need to lose vs CLG) to be 9-7 CLG - GGS, EFX, FLY, TL, C9, +2 wins inside the tiebreaker mean they need to win one of the remaining games against FLY, EFX, and OPT as well as beating TSM to be 9-7

Assuming anything with the standings is always a bad thing, but both TSM And CLG have beaten Echo Fox so far. CLG has the edge so far on strength of schedule as they have beaten both the first and second place team as well as the provisional 4th place team in TL. If CLG can 2-0 EFX that would give them a stranglehold on the strength of schedule tiebreaker as it is written in the rules presently. With one team clearly winning or losing the three way tiebreaker, does this then remove them from the equation and put the two remaining teams in a one game playoff? Would resorting to the three way tiebreaker of strength of schedule decide the final ranking of all teams involved within the tiebreaker by declaring Team A to have a better Strength Score than Team B who rated higher than Team C?

XpertSkills3/12/2018, 10:45:03 PM2 votes

2-way tied system is fine but for 3-way tied, playing all the games is the fair system. Each games is different, and teams are not improving at the same rates all the time. it's unfair to judge the teams by their past games during the seasons, If they all tied they all deserved equal chances. I know it's hard with all the limitations, but it's not all the time that we have a 3-way tied. I am not too concern about burning out pros, if the win is the fruit of hard labors, there is nothing we can do, may the team with the most endurance win. Therefore Riot should accommodate the time for tied- breaking in their schedules, and offer all teams equally chances on that day, not bye them because of recent strength again other teams. It happens all the times in sports that team A counter team B, while time A is weak again team C, and team B crushes team C easily all the time.

bobbybigwheels3/12/2018, 6:31:24 PM1 votes

I agree with the NA style. Endless round robins end up into stamina vs strength of play. Also it rewards play throughout the season, which this playoff is for. Hot teams should be able to win the first match and overcome this disadvantage, and they could of played better all season. 2 games and get on with playoffs!

feedme56833/13/2018, 1:23:44 AM1 votes

Why must you give me another reason to believe EU is better than NA? This to me is you saying money is all you care about. Tiebreaker games don't make you more money because they aren't highly viewed and advertising is probably hard to sell. And you have to pay the casters and production staff, but seriously LET THE BOYS PLAY. Nothing else makes sense.

iFairy3/13/2018, 9:11:42 AM1 votes

So this makes sense in only a very small number of situations. First what is the end of the season? The last 4 games, the last 6, the second half of the season? This punishes teams that have an easier final few match ups based on a schedule they have no control over; they have the same win record as their tied partners but because they won "easier" games at the end of the split they are considered weaker? But lets say you just judge by the entire second half of the split - all teams then have the same strength of schedule and then strength of wins (?) can be justifiably used because you've limited yourself down to one variable instead of punishing teams who just happen to have an easier end of split schedule.

BUT then if two teams have the same record at the end of the season and played the same second half of the split you have a couple possibilities. First a team (1) wins a ton of matches at the beginning of the season and only a few during the second half while another team (2) surges at the end of the split giving them a better record for your system. This makes sense to me - you reward a team (2) on the rise while a team (1) that surged early but lost steam is probably a weaker team at this exact moment so they are seeded lower. Great! But what if they have the same win/loss record for the second half of the split? What if both lost half their matches on the last half of the season - one team (1) beat every opponent on the top of the rankings while the other team (2) beat every team at the bottom? In your world team 1 gets the better placement while team 2 gets the lower seed. But what are you rewarding? Team 1 beat better teams but choked against easy teams they should defeat. Sure team 2 lost to better opponents but played much more consistently to beat those at the bottom. Should we reward clutch play and punish consistency?

At the end of the day I think this system probably works. The best team of the three will beat the others and win out. Doesnt matter if they have to play 1 or 2 matches. The best team should win out on the day. BUT if your making rules and regulations for teams to determine who is best out of a tied group of teams you should consider what you're rewarding. Currently you're rewarding either 1) teams that happen to have a harder end of split schedule or 2) teams that surge at the end of the split or 3) teams that play more clutch against better opponents but choke against weaker. If this is what you want then awesome - #1 is downright terrible but #2 makes sense to me and #3 works well enough because the playoffs are between the best teams and seeding a king killer higher in the bunch makes it more competitive. But if this isnt exactly what you want then probably best to keep looking.

Fuedaddy3/13/2018, 5:13:01 PM1 votes

Playing it out is the only reasonable way to handle it.

This game is so volatile, the items change, the meta shifts. It is not only important to see who played and beat who, but also, WHEN they beat them. Some teams kill it in certain meta's but the meta also shifts several times throughout the split. Case in point is the current iteration of fox. First place, but how many games have they lost just in this last 3 weeks? I think the only way to gauge which team is better, especially since the tie breakers will be held at the end of a split in that specific meta, is to play out the round robin.

CallMeTex3/14/2018, 1:16:06 AM1 votes

I have looked at the tie breakers for several College Conferences for their sports and the first tie breaker with more than 2 teams tied is a mini round-robin of already played games to do some initial eliminations (see the Big 12 conference for example). Before the final weekend of games in this season there is a reasonable chance we could have a 4 way tie for 3rd - 6th places. If you use the round-robin on already played games the following happens.

4 Way Tie scenario: 100T beats CG and loses to Echo Fox (11-7). CG loses to 100T and beats GGS (11-7). TSM beats GGS and beats CLG (11-7). TL beats C9 and beats OPT (11-7).

This results in a 4 way tie, so use the mini Round-Robin results on already played games for the season: 100T (4-2), TL (4-2), CG (2-4), TSM (2-4). 100T and TL end up tied for 3rd place and because they split the season series must play a one game playoff to determine 3rd and 4th place. CG and TSM end up tied for 5th place and because CG won both games in the season series CG takes 5th and TSM takes 6th.

3 Way Tie scenario: Same scenario as above except TL loses to C9 (or OPT), so a 3 way tie for 3rd - 5th (TL (10-8) gets 6th place). Mini Round-Robin results on already played games for the season: 100T (3-1), CG (2-2), TSM (1-3)

So 100T is 3rd, CG is 4th, and TSM is 5th. No playoff games needed.

I think this "Head to Head" type system on only the teams involved in the tie is much fairer for the 1st level. Of course if still tied after this (25% of the time in 3 way ties as was stated), then look at the "Strength of Victories" and have playoffs.

So you didn't define what happens with the current rules on a 4 way tie. Is it the 2 worst Strength teams play, winner plays 2nd best, then winner plays best? OR 1 plays 4 and 2 plays 3, then winners play and losers play to get all 4 positions?

Chasmweaver3/21/2018, 5:02:41 AM1 votes

How about they treat the game like soccer/football or hockey? When the teams are tied up they have to prepare for and play a different aspect and style of game. Like a shoot out round. Everybody likes fights and this particular one would be held on Aram for either a 1v1 or possibly team matchup. This would solve the length of game problem of tie breakers (much like it works in soccer) and also force teams to train for more scenarios than just the rift. It would also help determine best sup, best add, best etc on a 1v1 level (unless the team option is preferable and personally I think that would be more fun. Again, it would and should only be brought out if there is a three way unreasonable tie. Pretty sure no one will agree with me on this since it is so far from the rift's style of play but so are shoot outs and they are widely used for two way tie breakers. The only other option is for them to create a new map with three nexus .... mind blown! lol

Morticianjohn3/6/2018, 9:24:01 PM1 votes

Strength of schedule is nice. It seems better to me than playing round robin even though I am sure the round robin would be more entertaining.

Eggr3/7/2018, 5:46:17 AM1 votes

Regarding 3 way ties: Using in-game metrics as the tiebreak for inconclusive round robins is a reasonable way to gaurentee a winner after the 3 games; it allows each of the teams a fair chance to win outright, while preventing infinite rounds.

It might affect the meta game during the tie break games, but as long as you only count metrics recorded during the tie-break games it won't have any impact on the regular season.

III BAKURYU III 3/7/2018, 9:02:29 AM1 votes

A year ago the LCS had a "Best of 3" for a whole season but when it comes to tie-breakers the "Best of 3" is " 3 games can be too many" - K Also, we're talking about pros who play 20+ games of solo Q a day but can't play 4-5 more games to determine a clear winner and to wrap up their season? But hey, throw in best of 3's/ 2's, rift rivals, some crappy Katowice tournaments, IEM things, MSI, Seasons, playoffs, Finals and finally, Worlds but stamina is too much of a factor for the pros?

While I get the "Bye" part in which 3 or more teams are tied with the avg game time but I personally would choose the "Who beat whom" more so than the average game time mainly because some teams are a bit more strategic and take their time in winning games while others are too fast for their own good.

2017 NA Spring Split Cloud 9 had the most wins and were in the longest average game time. 2017 EU Spring G2 Esports had the most wins and were in the 2nd longest average game time. 2017 NA Summer Team Solomid had the most wins and were in the 3rd longest average game time. 2017 EU Summer Fnatic had the most wins and were in the longest average game time.

Hell, back in 2014 Team Complexity would have constant 50+ min games in which I think like 3-4 of their games are in the top 20 longest games in pro history and yet, they would consistently beat the top teams.

EU LCS " they are going to play all the additional games". Wow, NA needs to stop "Copy and pasting" what Korea does and first start off with EU, especially since EU tends to actually get further than NA at Worlds each and every year. While EU did have some of the weirdest scheduling in the past(2 divisions, best of 2) the fact remains that at least the changes they made actually had some weight to such/results.

Why aren't articles like this released near the start of the season so fans, experts, and theory crafters can actually take more than 3 WEEKS to figure out? "Going to the moon on X date but I'm sending out blueprints on how to build the spaceship" type of feel going on right now.

Virtual Prophet3/7/2018, 9:04:11 AM1 votes

maybe have them do 1v1's until someone wins like a shootout in hockey or soccer. it's not a perfect fix but it would address the time constraint and burning out pros issue because of the lower time frame of an outcome. 5 1v1's, best out of 5 wins simple and easy and everyone gets to compete. top vs top, jung vs jung, mid vs mid, etc..

Ravens69693/7/2018, 6:06:10 PM1 votes

This is very interesting. One thing that is fairly disconcerting is the location of the finals this year. For every year, for the past 5 years. league has been played at a venue that seated 20k......this venue seats 2.5k. This worries me about the future of the sport.

Séance3/7/2018, 9:09:23 PM1 votes

" When you run the possible outcomes, 75% of the time you will get one team that is 2-0, one that is 1-1 and one that is 0-2. Easy and clean. But in 25% of the cases you get three 1-1 teams, and when we don’t have the ability to play more than 3 games, that leaves us in an impossible situation (and in some cases, even 3 games can be too many)."

This operates under the assumption that the teams are evenly matched and all are 50% likely to win, which seems unlikely even if they are tied at the end of the season.