Response to “ SMITE: Why SmiteGuru Score Broke”
For those of you that don’t know DukeSloth is a 25 year old Smite community member from Germany. He recently released a video called “SMITE: Why SmiteGuru Score Broke” which outlines why he thinks the recent changes made to SmiteGuru have made our score system useless, however there seems to be some confusion that I can hopefully break down and clear up without boring you to death.
With our recent update to the site we added a lot more stats, that the community was requesting, which is where I think some of the confusion comes from. So let’s break some of them down.
Elo Plus (Elo+) — This is our Elo calculation for only 1 queue. Before the update this was only displayed in ranked games and labeled “Elo”. After the update it is now shown for each casual mode and ranked. The season 3 and season 4 Elo formula are the same.
SmiteGuru Score (SGS) — This is a combined casual queue score. Taking into account any casual games played on this account. Before the update this was the only thing displayed for casual gamemodes. The season 3 and season 4 formula are the same.
Queue Scores — These are brand new scores that rank you from 1 to 10, 1 being the worst, 5 being average and 10 being the best. These are based on your average stats for each gamemode or champion and are gathered from players with at least 10 played matches with the god or gamemode.
Queue Rankings — This is also a brand new feature that lets you see a exactly where you place based on your Elo+ or SGS. Players with only 10 games in each gamemode are ranked.
Previous seasons SGS was never really accurate it did give you a decent representation of players in ranked… HiRez confirmed this.
This is because scores for Ranked were supposed to be more serious where casual was supposed just show who’s a better overall casual player not who’s the better Conquest player or Arena player, etc. We’ve changed this so now you can see both.
There was a lot of players that would four-man queue into less populated gamemodes like 3v3 or Siege. You would often be matched against single players with no coordination because not enough players…
This is not as big of an issue as you might think, almost all the other gamemodes are more popular than Conquest and teams are more often matched against eachother than not. With Elo+ if you have a high chance to win a game you gain less points and lose less. Also remember, this was in ranked too, looking at you three-man pro queues.
There was a decision that was made that made these scores worthless — what should have happened is factoring in previous seasons with win rates and different modes were there. All the ratings were set to 1,500 and because of that there’s no value for almost all player groups.
So as you say later in the video win loss is not directly tied to the skill level of the players, which is correct, but that’s the goal of the system, so we can’t score you based on your win loss.
Having a SGS / Elo+ reset puts you at the same skill level as every other player just for that first match, after that every game that you play makes it more and more accurate. We also have additional qualifier systems to make it more accurate faster (Check out the FAQ on the site).
Because you already had a different rating from previous seasons and casual mode didn't see a reset — there is now no way to find out what your actual score is. If you were already at 3,000 Elo last season in casual conquest then you will now stand at 1,500 Elo and obviously you wont get that many wins at 3,000 Elo because you will be matched against people of high Elo and the system is supposed to place you with equal opponents so its going to be hard to get a ridiculous win rate that will allow you to grind much further than that.
Again the point here is not to match your MMR. It is to compare players and the more you play the more your score will go up or down depending on your skill level — and not only as a player but as a community as a whole. So that the players that are supposed to rise up will and those that are supposed to stay near the bottom will. It’s the same way when MMR is reset — those masters players eventually get into masters, even though they started at qualifiers like everyone else. This isn’t the Season 3 TP promotion system. You don’t need win streaks. If you have 10 wins against low ranking players it can be equal to 1 win against normal ranking players. It all depends on your team’s ranking.
In the stats of the SmiteGuru Score you now still stand at 1,500 Elo a little more or less depending on your win/loss but will not be close to your actual Elo at all and this is true for almost every player. This is for 2 non level 30 or you had 1.5k Elo in that mode before and this is not something that you would know. There’s no way to know your SmiteGuru Score for previous seasons.
THAT. IS. NOT. THE. GOAL. We aren’t trying to match your MMR — we are showing you your skill level against all the other players in the game.
You can see your score from last season by clicking the button on your profile. We said many times that this would be available a week or so after the update, there was a lot of data to transfer, but at the time of this video it is available for everyone already. Your season 3 stats can be found here: http://smite.guru/profile/pc/Blackpage/casual?season=3
That also means if you are decent at the game and you create a new account now that you haven't played on before and play that account it will get a higher Elo than your old account… Maybe that account will get to X Elo and your old one will get less. This means that any pre-level 30 player will end up with much higher rankings compared to players that were playing in last season and already had a standing and just started again from that reset standing.
So the concern here is pretty valid, but it’s something that with our new framework and back-end system we are looking at. It’s a difficult one because the Smite matchmaker seems to deal with pre-level 30 players a bit different and on some type of weighted scale where they attempt to match them against people their own level, but don’t always. As anyone who has an alt can tell you — you start getting matched with level 30s way before you reach that level. Then there’s also parties involved. What if a level 10 is partied with a level 30, etc? So it is something we are looking at, but I don’t see it being solved super quickly.
A good rule of thumb for this is to look at a pre-level 30 and compare them with only pre-level 30s around the same range.
This even applies to the main score which we have where you would be able to see the combined for all modes but because this also got reset this value will not tell us anything anymore.
They tell you overall casual performance…just like the old SGS did… what it does hasn’t changed…
The only scores at this point that have value is ranked. However these rankings won’t remain reliable either simply because of the next season split we will see a soft reset of Elo and I don’t think there will be any way for SmiteGuru to find out where the soft reset will place players. So in the end it’s useless. So there is no accurate way to tell where a player should roughly based on any SmiteGuru stats anymore.
We’ve been through a few soft resets in the past and it always depends on what and how they change. We get information and if we feel the need we change accordingly. This won’t be the first soft-reset and both the community and you seem happy with how it went considering the video opening statements. All players are judged by the same system which shows their ranking against each other not against the match making system.
Other Profile Stats
And there are these rankings the percentages and overall rank that you are given that you can look at but those don’t really make sense as you can see I have played in neither of the modes…
This was actually a bug that wasn’t reported. It has been fixed.
…What I think is that the players get randomly ordered in these ranking and that will skew the curve for anyone because really the good players or the players that win the game-mode a few times should be in the top 5% all the sudden where the players that loose will be in the low percentage very very quickly because there is a massive amount of players with 1,500 who actually just don’t play the mode.
What actually happened is that it was not respecting the season change, so your season 3 scores were showing up. You have to have played 10 games to be included in the rankings.
Likewise there are queue scores which look interesting at first but once again they factor in players that don’t play the mode at all — once again I am in the top 0.01 for wins which makes no sense and the same for mins. So anyone who has never played conquest is just factored in there and that pushes anyone who plays the mode a little bit into the top percentages right away and all the other players are in there for no reason. But once again there is no information on how this is calculated and no further details on the FAQ and its really weird.
Again a bug — simply with the way that win percentages were calculated because they are a percentage unlike the rest of the numbers. You also have to have played 10 games to be included in those community averages. (You can see where you would rank before playing 10 games.) Also this bug was not reported until now.
A few things to take away
We found that most people like resets, I would assume HiRez came to the same conclusion, since they are adding them to ranked.
It allows people to start fresh without having to level up another account. So that if they have improved they can see it in the numbers quickly without having to grind it out. We like giving them the ability to do that over the seasons which is why on this update we focused on adding season tabs so you can now and in the future track your progress. Resets like this make it possible and since everyone is being scored by the same system it doesn’t take anything away as long as you’re willing to play a few games.
For any site users if you don’t understand something please let me know. I am very reachable on Twitter and answer most support emails personally. It is much better than speculating on something that might be incorrect, a bug or something you don’t fully understand.
Thanks for reading. Here’s a SmiteGuru icon if you were fast enough: http://eb.godankey.com/?id=VUQ0T2JBVXdBbU09