Boards›Gameplay

Why and how placements should be changed next season

G0rkster·9/30/2015, 3:15:59 PM·3 votes·948 views

Placements as currently implemented work well for people who are high elo and should not hang around low elo very long. But I argue that it is broken and negatively contributes to the low elo experience as currently implemented.

An example of a newly placed player in silver 5

If the elo system works everybody should be close to their true rating indicated by their win rate having balanced out near 50%. This is true only of the established players here. The newly placed player clearly does not succeed in Silver 5 and has lost 20% of his recent games and 25% overall. This player will stick in silver 5 until his MMR is bronze 5 then drop to where he belongs.

The problem here is multiple:

  1. It happens frequently, much more often than smurfs leveling.

  2. It adds relatively longstanding biases to low silver, in particular silver 5.

  3. Silver 5 is already the most populated elo with too much variety.

  4. Misplaced bronze in silver encourage dodging as the "better" playing strategy. Ideally playing should be the best strategy in as many situations as possible. If all players are near 50% win rate there is no incentive to dodge.

What needs to happen: Bronze level players (many new ranked players) should not place above bronze.

Possible fix: Every player starts at Bronze 5 after placements, but their MMR is determined by placements. So you do get new players that the system deems to be silver have silver MMR.

If over the next N games it turns out that indeed they can compete at that MMR they will climb. If it turns out that cannot their MMR will deteriorate. The promo and climb mechanism will create enough of a sample pool to do this. Skipping divisions will make high elo players climb very quickly and have minimal lasting impact at any one particular MMR.

This shifts randomness to Bronze 5 but makes Silver 5 more stable. I think it is correct to place the maximum amount of variability at the bottom tier.

P.S. Some numbers to show how strong the Silver 5 bulk is compared to other divisions. These numbers are from NA lolsummoners

Using this equation:

Effect degree = Numbers at tier 5/All players in division below (gives a percentage comparisons)

Bronze 5: NA

Silver 5: 0.396

Gold 5: 0.182

Plat 5: 0.153

Diamond 5: 0.175

In all divisions Tier 5 of the higher division is about 15-18% of the players of the division below. Except silver 5 where it's near 40%.

10 Comments

Ternt9/30/2015, 3:25:32 PM3 votes

I also believe that new players should never place higher than Bronze 1. I cannot count how many times I've had brand new 30's in my Silver 1~2 elo games. It's automatically having a disadvantage because they do not have the experience that the rest of us who have climbed from Bronze upwards this season. I know this from personal experience, because when I hit 30 after worlds last year, I got placed into Silver 2. Do I honestly think I was good enough to play in Silver 2 last season? No. Do I now, Yes. But it took me almost an entire year to gain the knowledge and skill to maintain Silver 1 or 2 status. New players should NEVER be placed into Silver, easy as that.

EDIT: This has kind of forced me into checking each person on my team before the game even starts. After a game where my 12 game Tristana died 13 times before 20 minutes, I've decided I'd rather dodge games than play them because of this.