Greedy Goblin

Wednesday, January 18, 2017

The dodge prison did not skew the League data

I was worried that being placed to the "prison" causes more than wasted time in League of Legends. After all, one can claim that there is no prison at all, just random events created my previous results and now the same random events took them away. Luckily it's not the case, probably because the same algorithm is used to rig for paying players that is used to rig against "offenders":
Actually my prison weekend data increases the result that the biggest predictor of win is having "type D" players in your group. Every single group has the same representation disparity in both dataset. Please note that the scale is % of teammates, so on both graphs the sum of the green bars is 100 and the sum of the red bars is also 100, so the difference in winrate is not visible. Which is also not important. You can claim that on the weekend I was unlucky or I played badly for some reason but it doesn't matter: the games were still stacked in one way.

I'm now working on the final iteration of the exploit mode and will publish the results, regardless of my position on the ladder. Sure, I prefer to climb higher than ever to prove how well it works. From that perspective, the prison island was a setback. On the other hand, it's an extra proof that Riot manipulates your winrate whenever they please.

Also, I'm clearly out of the prison now:


maxim said...

This is getting really boring.
The payoff better be worth it.

Honestly, you would do better just writing battle reports of some actual games you did (certainly, there were at least a few interesting ones?) than spending all of this time on some meta promises bs. Just publish the results when you get them.

Gevlon said...

I just did as you suggested: I published that I've just found that
- by stopping dodging, I'm getting my winrate back
- when I lost winrate, I only lost winrate and not the structure of the games, so most likely the system is the same, just dialed higher

Gesh said...

It seems that there is no prison, but you were on a losing streak.

Gevlon said...

@Gesh: the chance of that happening randomly is less than 1%

Anonymous said...

At times, I think this blog is more like a social experiment than real talk. Spoonfeeding us speculations as facts, then analyzing the results. This prison island is just one of many and not the best of yours. 40 games with 35% WR ? Never happened to anyone before.

Rigging is still yet to be seen but by judging all the articles so far, you have not achieved anything. Your best bet is to prove, that matchmaker is not perfect, but that is pretty obvious.

I wish you transitioned to something of your domain (MMO). Why haven't you ever bothered with well-known, established titles such as GW, Lineage, Runescape ? At least, your readers would get to know these games from your perspective.

Gevlon said...

@Anon: of course it happened with others before, but it shouldn't. Actually my earlier claim that the chance of it is 1% is wrong. It would be 1% if the game results would be uncorrelated which they shouldn't be: with every defeat your rating decreases, so you get easier enemies, increasing your win chance. I lost two leagues and still had the same abysmal winrate. Then suddenly this went away.

Of course I haven't proven the rigging yet, and it's on purpose. Right now I'm just another "whiner kid blaming Riot" and they have no reason to care. If I came out with the "big find", they'd immediately shot me down. So I must not publish until I'm done (read: decided that I won't play LoL again).

Anonymous said...

There is certainly something strange going on in LoL matchmaking, see the placement of rank newbies in Silver. Or the fact that you can trade roles with other members of your team when being auto filled (if you can fix auto fill issues in two seconds in chat, then the matchmaker is clearly not working). All of this evidence doesn't necessarily point to rigging, but it does wildly gesticulate in its general direction. I think Gevlon's upcoming post explaining A, B, C and D players and what "exploit mode" is will be the nail in the coffin.