I have a suggestion to address the issue of new players being placed at a much higher Elo than their actual skill level.
All new players would start at 500 Elo so they are matched against players closer to their skill level and do not have to lose 10 to 15 games just to reach their appropriate elo. In addition, new players would receive 500 extra hidden Elo that is only used when they lose games and doesn’t affect their placement on the ladder. Losses would first deplete this hidden Elo before affecting their normal Elo.
This hidden Elo does not have to be literally hidden and could be shown as something like shield Elo or provisional Elo. Once it is fully depleted, losses would start being subtracted normally from their Elo.
The goal is to let new players face similarly skilled opponents from the start instead of losing repeatedly and quitting, while also preventing Elo deflation on the ladder caused by having new players start at only 500 Elo.
so basically the extra 500 shield elo makes sure the elo average stays the same and not drop.
as matter of fact in the current system the elo average of active players keeps rising (Elo inflation) every time a new player loses couple of games and quits the game forever and this new system would actually help with that elo inflation as well.
Let me know what you think or if you see any problems I might be overlooking.
if a new player with the skill of a 200 elo player joins the ladder, they can’t fall down to 200, as the shield protects them. this means they will still need to lose a lot of games (to deplete the shield elo) and then some more to fall down to 200.
if a new player with the skill of a 500 elo player (or higher) joins the ladder, they will gain elo for every win, but won’t lose any for their losses, so they will climb the ladder. Eventually they will reach a point where they can’t win anymore, so they keep losing, but they aren’t falling yet because the shield protects them. Only once the shield is depleted will they start dropping back down to their correct elo.
I think overall this will mostly just delay the loss streak. Admittedly this might prevent some people from leaving out of frustration in their first few games.
I think a better system would be to simply start new players at 800 or even 500 elo. The effect on average elo would be very small and take a long time to become noticable. Yes, eventually the starting elo would have to be re-adjusted. I think this could be automated by having a few AIs play on the ladder to see where each AI fits in.
Thanks for the reply! you made some good points and here are my answers:
• the 200 elo player would need to lose the same amount of games as in the current system the 500 shield won’t make him have to lose more games. however most players drop to about 500 before they start climbing back again so 500 is probably where the average new player’s skill is at. However the starting elo doesn’t really have to be 500 it could be lower and the shield could be increased. maybe the game could even let the player choose where he should be placed by asking for his skill level. for example an advanced player could be placed directly at 1000 new to pvp but knows how to play the game from campign at 500 with 500 shield and completely new to the game at 100 with 900 shield.
• I see what you mean but a new player would never be abled to pull a long winning streak and is probably going to have a 50% win rate and the 500 shield would probably get depleted before he reaches 800 elo. also losing against players with similar skill is much more forgiving than being stomped by players with higher skill because you’re at least you would be having fun.
Losing alot at the begining is unavoidable but the idea is to not let players experience a long crushing losing streak and not have fun during it and instead have some wins between the losses and even enjoy losing and strive to get better.
also one thing i would like to add is the shield is only useable when you are under 1k or maybe 1.2k elo just so no smurfs start with 500 elo and climb up and use their shield in higher elo.
starting at 500 elo without this system wouldn’t work because eventually the new players will climb to 1k elo and where are they going to get these points from? from players with higher elo which would cause a deflation and a drop in everyone’s rank and people would get mad.
however with your proposed system he doesn’t need a win streak. at a 50:50 win-rate he will gain 500 elo while losing 500 elo (but those 500 elo are absorbed by the shield), so he gets shot up in the rankings (ofc the higher he gets the faster his shield depletes). So even if he manages only a 2:5 winrate in his first games until the shield is depleted, which will still be quite frustrating for most people, his elo will end up being inflated by 200 points.
this is true in theory, but not really relevant in practice. there have been several posts about this on this forum, with the current number of players and the current influx of new players it would take years before it becomes noticeable (iirc, the estimate was that with a starting elo of 800, it would take over 10 years for the average to go down to 900. a starting elo of 500 would approximately half this time, but that’s still a long time)
If you reduce starting ELO the average ELO will go down over time so to keep the same effect you have to continuously reduce the starting ELO.
The other big issue with reducing the starting ELO, that I haven’t seen anyone mention, is a psychological one. If the average ELO is slowly going down over the curse of years then most players will continuously lose ELO over time unless they are getting a lot better very quickly. This constant slow ELO drain will definitely not be motivational for most players. Many players might quit in frustration because they only slowly see their ELO go down or stagnate despite them actually getting better at the game.
If the ELO was hidden then this would be less of a problem. So maybe hidden ELO would be the best solution. Instead of showing ELO players would get seasonal named ranks like “Gold III”.
I think you missed the point of the whole post. reducing the start elo to 500 has been suggested by many people and pro players but then there was the argument of the average elo dropping.
In this post i suggested a way to reduce the elo to 500 without elo deflation dropping the average elo. so yeah the average elo on the ladder would stay the exact same it is right now.
The shield ELO doesn’t work either because of things mentioned by TheTowerDefender.
That’s why I proposed fully hiding the ELO and then giving players named ranks. But that solution needs be be careful adjusted. You don’t want the visuals ranks and hidden ELO to be too different because then people will match against others that have very different visual ranks and they think they are matching against the completely wrong ELO even if they aren’t.
I get what you mean, most players will get an inflated elo by the time the shield depletes. however to get to this point where they depleted all of their shield and got a bit inflated they must have played at least 20 games which means they are already liking the game and are hooked and already got a bit of experience in the game which would make losing this inflated elo not that bad and it is certainly better than starting at 1000 elo and having to lose 500 points through a crushing losing streak.
also what i have here is just a starting idea that could be improved for example losses could take 50% of points from the shield and the other 50% from their normal elo.
Or whenever winning they could get only 20 points from the opponent and the shield doubles it and gives 20 points as well and losses take points directly from normal elo.
there are so many options to improve this system and it is just a starting idea. it doesn’t matter how it is done but the goal is to put new players in low elo but still make them contribute 1000 points to the total “economy” of the ranking ladder.
i read the post in the link and it shows that most players end up quitting after experiencing a crushing losing streak or even just one major crushing loss. if you drop the elo to 500 players might just continue playing and not leave since they would be already in their apropriate elo so numbers of new players on the ladder would increase and it wouldn’t really take that long for the effects to show in the lower ranks. it would take years for the effects to show in the higher ranks though. it doesnt take long for a 500 elo to reach the skill level of a 1000 elo and then points would have to be shared and everyone in the 900-1100 range would start dropping.
It is true the total average would slowly drop over the years but the average of the lower ranks would definitely drop much faster probably to 800-900 average instead of 1000 within a year or less.
yeah it would probably take a longer time for it to drop all the way to 500 average but even a 100 elo drop would be horrible imo.
Ok I have an idea for an advanced shield ELO concept:
Every player does 4 placement matches VS. AI.
Every time they win they get 200 ELO and have to face a stronger AI.
Every time the lose the gain nothing and the AI difficulty stays the same.
So they can get between 0 and 1000 ELO.
The ELO is “filled” up to 1000 with Shield ELO
Players who win all placement matches start normally at 1000 ELO. The others get some Shield ELO.
You lose some real ELO when you lose a match.
If you lose a match and you have Shield ELO you lose a percentage of Real ELO.
The percentage is equal to ratio between shield and real ELO.
At 100% Shield ELO you only lose Shield ELO.
At 50% Shield ELO you lose the same amount of both.
At 0% Shield ELO you just lose normal ELO as expected.
You always lose at least 1 Shield ELO if you lose.
That means the higher you are on the leader the less “powerful” your shield is preventing you from ending up with a too high ELO.
Inactive players get Shield ELO
Why limit this system to new players? Players get some of their ELO converted to Shield ELO for every month they are inactive. That should help them get back to the game more easily. The number could be relatively low like 10 ELO per month of inactivity. This timer could also be limited to only start after a certain amount of inactive months, maybe like a year. So after 2 years of inactivity you would only have 120 Shield ELO.
I don’t think it’s a good idea to force players to face AI before playing against real players i know alot of players would hate it. also losing elo to AI defeats the purpose of the shield elo. the purpose of the “shield” or “hidden” elo is to replace the missing elo when players start at 500 or lower just so the current ladder elo economy doesn’t deflate resulting in the drop of the average elo. losing that elo to AI would mean this elo is gone forever and not passed to other players.
It’s like if 1000 points is the entry fee that every player must pay to play in the ranked ladder but if you give the new players only 500 points then it wouldn’t work so you need to find a way to give them 1000 points and still place them at 500 elo and that is the job of this hidden elo.
the hidden elo isn’t really here to offer new players some type of protection from losing.
yes, but as I wrote, with the current rate of new players vs existing players, a change of starting elo from 1000 to 800, would take between 10 and 20 years to shift the average by 100 points.
So this “continuously reduce starting ELO” is theoretically true, but in practice this solution would fix the system for a decade without any intervention.
this is also only relevant if the average ELO changes are noticable. 100 points of change over multiple years are not noticeable.
this part I definitely agree on. I think your proposed system is still better than the current system. I just think that there are still issues with it, and better systems exist.
Ideally we would have a way to place people and their “true” elo from the start, an approximation of this would be placement games against AI.
But I think even a simple lowered starting elo would still be superior.
Both of these system have the downside of elo deflation in the long run, but we are talking about literal years or even decades.
I think the general idea to place new players lower is great, but the system to combat elo deflation seems over-engineered with its own downsides. A simpler solution would be:
add +1 elo to a bunch of random players ever so often (hide it by making one player lose 16 and the other win 17 points). Iirc the current average elo is slightly above 1000, as more people below 1000 have dropped the game than players above 1000. shifts in average elo can have multiple causes.
I also think that lowering the starting elo would keep more people in the game, but I don’t think it would be an enormous change. even now 25% of the people who win their first game don’t play a second (compared to 40% of those who lose it)
The analysis suggests a starting elo of 600-700. But even with twice as many people staying on, and the starting elo being 600 instead of my suggested 800, a noticeable shift in average elo (eg 100 points) would take several years.
I don’t think elo inflation/deflation should be a major concern for designing a rating system when it happens this slowly. The devs also ignored this issue when they added elo decay for inactive players, and points leaving the system when players quit.
a 100 elo drop over 5 years would not be noticeable. changes in how much you play and how well you understand the game (eg new civ additions, map pool changes, meta changes) will have a much bigger effect on your elo. + most people’s elo varies by more than 100.
(btw, the closer you get to the new starting elo the slower it would converge)
A whole different point but wouldn’t it be possible to just add some more ELO to players randomly to keep the average at 1000? Like 1 Point to to a random selection of players once a year would probably be enough and practically not noticeable.
Starting at lower ELO sounds probably good. I didn’t know the game throws you into the deep end until I tried to play some Ranked long ago and got throttled because I didn’t know everyone plays the same ”early rush"‘ way, likely taught to then by their favorite eSports player/caster
As a “vs AI” player, it was culture shock and a very poor experience. I never played ranked again. Everyone seemed to study and know the latest meta and early rush strategies, and were hyper concerned with very early econ disruption, which I find very annoying. It’s not my style of aoe gaming. But I also don’t want treaty.
I’d probably do better now because the AI was trained for a while to do early rushes in every match and I have more strategies now… and I’ve watched a few eSports tournament casts. But you’re right, it felt like a high barrier to entry and I haven’t played Ranked for years now as a result of a few matches that left a bad taste in my mouth. It felt like everyone had mad skills in Ranked and loved playing Call of Duty style. I dropped matches fairly early and looked forward to just going back to my “vs. AI” safe place.
I know very little about ELO, and didn’t know there are much lower scores out there than what the game starts you at and dumps you into. It could educate new players much better on that front, imo. I still don’t really get it. All I know is Hera has a near perfect score at times, the Wololo Tournament core players are close behind, and people like me are somewhere well below that but no idea where.
I’ve long wanted and asked for an offline ELO scoring system for “vs. AI” skirmish players. If nothing else, it might help us understand the system better before we jump into the deep end of online Ranked. My main rationale, though, is that it would be fun to have “vs. AI” games actually mean something to the player, and a reward for playing them over a length of time. Right now, “vs. AI” matches are meaningless blips in time