Greedy Goblin

Wednesday, August 2, 2017

My PUBG stats and the top monopoly of the "dedicated"

I took notes of the last 112 games I had in PlayerUnknown's Battleground. This was a smart choice, as the leaderboards were reseted. Out of these games, 10 crashed, it definitely felt more. Here is my progress during the 102 non-crashed games:

As you can see, I'm improving fast. The first half of the games had 29 median, the second half had 21. I definitely see a below 10 median a reachable short term goal. However it raises a question: we saw that my winrate is top 4%, while it should be around 25%. After all if my average (rather median) performance is around the 25th position, than 24 players are ahead of me and 75 are behind, so I should be at 25%, that's no-brainer.

To understand the difference, let's consider a simplified example: a match has 2 players only so you can win or lose. There are 10 players. Adam always wins against all the other players, Betty has 75% winrate against the rest, who have 50% winrate against each other. You'd expect the toplist with Adam at 100% winrate, Betty at 60% winrate and everyone else at 37.5% winrate. Now imagine that Adam and Betty are always online and play 100 games with each other. The others are rarely online, so when they play, they always play with either Adam or Betty. They play 10 games each (5 against Adam). This would mean that Adam still has 100% winrate, the casuals have 12.5% winrate and Betty has 21.4%!

We see the same thing in PUBG: there are awful lot of casual players with only a few games per week who have horrible winrate while a few "dedicated" players monopolize the top positions. While I'm currently around the 25th percentile in an average game, that translates to top 4% among the players, just like the abysmal winrate of Betty translates into top 20%. Please note that it's not a bug or corruption, it's the reality, just like if you go to any World Class tennis game, you see the same few players winning, or if you jump a titan around in EVE and take note of local, you find that most players in EVE are in Pandemic Legion.

This is in line with what the third party tracking site found, look at the large number of baddies at the low end. I'm sure that if everyone were playing the same amount of games, the distribution would be bell curve:

What I want to say is that you can only track your performance via objective statistical trackers and and their lack is a serious mistake on the part of the devs, because the personal experience of the average players is much worse than the reality is. Betty can honestly think that she sucks due to her low winrate, despite being one of the best players in the game. Due to the dedicated players monopolizing the top positions, the improvement from the bottom to the top 10% is nearly invisible.


PS: my statistics since the reset is looking really good:
I played 10 games, got 2 crashes, one unlucky landing (#88), one near-top 10 (#12), everything else is top 10. Getting to the toplist is definitely a possibility, top 0.1% is more or less done deal. I love how this "dumb shooter" is actually a strategy game and someone with a brain can easily outplay all the "l33t pwnzorz".

13 comments:

Steve said...

You'll probably dismiss this but I think the problem is you are selectively focusing on one stat to the exclusion of other stats...the meta of the current game is such that you can fairly easily get at least top 20 if you play conservatively, avoid fights, hide, etc. The problem is you will pretty quickly plateau w/out learning how to truly succeed and win games.

My guess is your games generally go something like this: land at a less populated area, loot as you move slowly/safely towards the circle, avoid contact if possible (run away from the sound of bullets, etc), and stay w/ the circle, hiding and keeping yourself alive as long as possible. Then you ultimately die when the circle gets small enough that you no longer can avoid combat.

This is a perfectly valid way to play, and will definitely result in a high final rank, but my guess is you will very very rarely be competitive for a #1 place...probably only in times where the placement of the safe circle allows you to bunker in for the entirety of the late game.

The problem is this strategy is giving you a false sense of success, where you think you are improving towards ultimately regularly being competitive for a victory...in reality the person who is spending his games dropping in to higher contact areas (school, military base, etc) is developing the skills that will make them competive in the games they do reach top 10 in. They are also developing much faster than you, because those games are shorter, they aren't spending 15 mins moving across the map and looting but then dieing as soon as they run into another player, so they are getting experience at a much faster rate than you are.

You should consider (if your connection will allow it), playing on a different server where you can ignore your rank statistic and rather just drop in to the dense areas right at the start...you will die quickly alot but you will always be engaging other players right from the start, building experience that will eventually help you as you get to those top 10 endgame phases.

maxim said...

@Steve
It is a rather common pattern with Gevlon. He isn't really interested in #1, but rather sees "top X%" as a sign of excellence, and he is more interested in discovering patterns than honing twitch skills.
This gives him a lot of interesting stuff to say at the start of the game, but continually leaves him hanging and unable to really advance as he plateaus in a territory, where 1000 words mean less than a few more APM. He is mostly satisfied with the result, though, because most normal players don't even reach this plateau and makes it his mission to write stuff to help people reach the plateau.

The only time this tactic really backfired so far was in LoL, where the plateau is also a trap (and he ended up inventing explanations like "elo hell" to shift responsibility for the poor decision elsewhere). Let's see if PUBG is the same.

Gevlon said...

@maxim: I'm already at top 2% in PUBG which would be lower platinum in LoL.
The reason why I'm interested in patterns is that honing twitch skills provides exactly zero thing to say besides "hone twitch skills". Someone with twitch skills has nothing to offer to the community as they already obviously known that twitch skills help. By figuring out a way to the top X% that is available to anyone without twitch skills, I can help these readers to achieve something they couldn't.

Anonymous said...

1) Well, being in top % does not mean you are good just better; so as the game grows, even if your skill does not change in the slightest, you should expect your rank should improve as the more popular it is, the lower the average skill. I.e., it seems to me that instead of playing, your time would be better spent on a Wintergrasp strategy. I.e., encouraging more bad players to join. Dubious ethics promoting it since this is a company that is already selling lootboxes in Early Access after saying they wouldn't. I wonder if it is coincidence that it is in your rating's best interest to encourage more people to buy.
2) There is a further complication, since you have to reach it prior to the next reset - e.g.now 29 days from the next reset. The optimal strategy may involve finding when during the week the worst players are on, get in a couple of great finishes and then coasting for the rest of the month.
3) How many readers do you think there will be who care about PUBG, haven't already settled into how they like to play it, and have the same victory conditions as you (I assume someone who cares most about k/d or k will find your advice less helpful) Can that be more than a few dozen readers?

Unknown said...

He might also be able to learn the map, spawns and best trap locations really well and use those to progress.
There might be tricks to use even in the confined space of top10 to allow others to kill themselves.

It depends on the weight of top positions in the final ranking, too. Is top1 much better than top2?

Gevlon said...

@Anon:
1) there is no "good", just "better than the next guy" where the task is not yes or no.
2) I don't care about gaming the toplist, because sooner or later they fix it. I want a strategy that can't be fixed server side because it's against the morons and slackers.
3) I don't know how many CURRENT readers care about PUBG. PUBG players are all perspective readers of a PUBG guide that is written by a top ranked player. Most of my page hits aren't from "faithful readers", but from random guys reading one of my permanent pages that google and links gave him. My top hitters are the BDO wealth guide and the EVE anti-ganking guide.

Anonymous said...

First person view only mode is coming in the next patch. Do you think you'll play in that mode or stick with the third person mode?

Gevlon said...

@Anon: it depends on how it integrates to the toplist. I mean will it be a niche with a few thousand players and its own toplist, or can you get to the same solo toplist as 3rd view players by playing FPS?

Steve said...

Imagine you had a blog about automobile racing and you put up a post "How to become one of the fastest drivers in the World!;" aspiring racers see this and click thinking you are going to give them some info to help them possibly become a race car driver. In the post you instruct people to go to Germany, rent a fast car, find a straight stretch of the Autobahn, and drive 150 mph for a minute. Congrats, the person who follows your advice is now in the top .1% of drivers by lifetime highest speed achieved.

But they are no closer to being a race car driver than they were when they started, and if they tell their friends how they are basically a race car driver at this point (after all both they and Formula 1 driver Lewis Hamilton are in the top .1% of fastest drivers in the world) they will be laughed at.

You've set up a goal/defined an achievement that is technically accurate but is not in reality anything meaningful or profound.

Gevlon said...

@Steve: if you don't find that advice valuable, then you are probably already a car racer, already drove at 150mph and interested in how to be even better. You might find this advice useless. You are not the 99% of the people whose "driving experience" is going from home to work at 60 mph and could have the time of their driving life on that German Autobahn. As a content creator, you must aim at these people.

Cathfaern said...

@Anon:
Not really. Gevlon is more like that you have 30 race car in a race. The best is the pole position of course but that needs high twitch skills. Gevlon says that yes you need twitch skill to be the number one but you can be number 2-5 (or even 1 with luck) with good pit stop and overtaking tactics and average twitch skills. Yes, if someone with high twitch skills implement these tactics he will be better than Gevlon but most of the time the others will just one to push the gas / break pedals and twist the wheel and don't care about tactics.
There is no separate win condition just the aim is not the first place.

Zyrus said...

You're all missing the point, what Gevlon wants to do is write about something different than the countless "GET GOOD AND CARRY" guides, because while "GET GOOD AND CARRY" is factually correct advice, it isn't simple, nor particularly viable to everyone.

Mind you, that's from a LoL perspective, doubt you can "carry" in PUBG, but it applies to all games in one way or another.

Unknown said...

You will get better at FPS mechanics the more you play, as will everyone else. The question is whether you sacrifice winrate to get more intensive practice - dropping in hot zones like school/crater/base and taking any fights you can will mean you spend most of your time either fighting or waiting in the lobby; or you use gevlon's approach and try to learn the strategy/tactics part first. I prefer the strategic approach, but since I am not playing for a goal and hence do not need to keep my stats pristine I am free to sacrifice some games practicing shooting in dangerous places.