The Pochinki Post Issue 5


Let me start this off by stating that I am a person who deals in statistics for a living, so I am definitely pro-statistics, but even in the professional world, most people want to use stats in a way that they should never be used – without context.

So with that said, I want to talk about the use of stats in professional PUBG - but first a detour.

For a long time in Basketball, the statistics used to determine how well someone played were standard and, importantly, easy to track. Points, Rebounds and Assists. The NBA didn’t even start tracking Steals and Blocks until the mid 70’s, and turnovers in the late 70’s, which all seem like absolute no brainers now. Eventually the “Advanced Statistics” began to find their way into the sport – initially with just some simple formulas run over the existing stats to give more insight and eventually (when the technology was available) all the way down to player tracking for every player on every play.

The revolution in statistics for anything always comes when people stop looking to statistics to try and confirm what they already think, and move towards asking what can statistics highlight that we may not already see. When people start to ask what the statistics actually tell us. Does having a high number of assists mean much if you turn the ball over to the opposition just as much? Does scoring 40 points mean much if it takes 50 shots to get it?

The other way that statistics begin to be useful is when there is an understanding of what is expected from players in certain roles. Do you want your Point Guard or Power Forward getting 10 rebounds? Is it more important for your Shooting Guard or Centre to have a good 3 point shooting percentage? The answer to some of these questions depend on game plans and the way a team decides to play. Context is super important to ANY statistics.

So I turn now to the way stats are presented in PUBG.

First of all, the stats themselves are currently in the Pre-70s NBA world where there are seemingly only a few stats that are deemed important. Kills, Damage, Survived time, Knocks. It is difficult to even apply any low level measure of advanced statistics to just 4 simple data points. People involved with PUBG – be it players, coaches, casters or whoever need to start pushing to work out what PUBG’s other stats are so that we can have something to work with. What are the Blocks and Steals of PUBG that we currently don’t see? Can we have Kill Steals featured? Do revives matter? Can the number of times a player gets knocked tell us anything? If the answer is no to any of them, can that information work in concert with other data to tell us anything?

We also don’t have to wait for technology to catch up to get more detailed data. This is a video game. The ability to track literally everything exists and is captured in a database somewhere. Is there any value in looking at the “shooting percentage” for players in relation to the number of hits compared to bullets shot? Could this be split between shots less than 150m and over 150m similar to a 2 and 3 point shot? Could we get data on how individual teams perform after losing 1 player early? What about when they go down to 2? Does it matter which player(s) it is?

These type of things can be looked at very quickly, but what will take time to evolve and understand are the roles of players within a team and what systems teams use. There are no real “positions” on a PUBG Team. Roles are loosely defined as being a “Fragger” “IGL” or “Sniper” but there is plenty of crossover within that. Some teams have multiple IGL’s, many teams run with no real snipers and Sambty from Team Liquid is leading the kill charts in the PEL comfortably (see below) making him a “Fragger” but he is also a Co-IGL so how do we classify him? This is where the context matters.

To get any use out of the current or potential stats, you need to know how the teams play and how their game plans work. Over the course of a season, players making sacrifice plays to help their team get the win will level out, but for anyone, including (and especially) coaches, to be able to get anything out of stats, we have to know what a player’s role in the team is. If a team’s game plan is to have Player X be their lead when heading into fights, then you might expect his Damage numbers to be high due to opportunity, but if their numbers have them in 20th in the league, that might be lower than expected because of the opportunities. Could we find through their “Shooting Percentage” in close range that they are actually doing a great job of hitting their shots? Could we find that due to their position they get knocked far more than others which slows their ability to get the needed damage? There are hundreds of scenarios and game plans that data could help to analyse

There are no easy answers here, but the way to get there is not to be afraid to try. I would love to see Tournament Organisers looking more deeply at this without fear of the players and coaches complaining that the numbers are useless. At this embryonic stage of competitions, trial and error is not only necessary but essential.


Sambty is Nuts

Team Liquid had a fantastic weekend in the PEL to drive them to the top of the leaderboard after week 2, and a lot of that can be put down to the undeniable Day 2 performance by Sambty. In one day – 5 matches – Sambty had one of the best individual point scoring days of any player in the competition racking up 24 kills. Of course he can’t do this without his team (though he did get several of the kills as the final player on the team) but his 24 kill points on day 2 was more points than 8 whole teams were able to accumulate from both kills AND placement points on the day. That’s half the lobby. Liquid is a team full of killers, but those types of kill points are expected to come from the likes of Jeemzz and Ibiza. If Sambty can keep up these types of performances it will be hard for anyone to take over the top spot from Liquid.


Wildcard Don’t Like the Desert

It is only 1 weekend, but in a tournament that is only going for 4 weekends in total, there is very little room for error and it only takes problems on one map to cause some serious issues. This has been the case for the newly formed 4 from Wildcard Gaming in the NPL. After 12 games in the NPL so far, we have seen some great performances from players having their first shot at the NPL. While Wildcard have been in the NPL for a while, and 3 of their players have been playing since the start of the season, as a group of 4 they haven’t played together before and were only formed shortly prior to the beginning of this phase. After a slow start on day one, they had a great second day with a couple of second place finishes and a win in the last game of the day and with plenty of improvement still to be expected from them, the team have a real chance to make a run towards the top of the leaderboard. To do that however, they will need to work out what is happening with their Miramar matches. In 4 matches on the map they have only managed to accumulate a total of 5 points with two 14th places, a 16th place and an 8th to go with their 4 total kills. The talent is there, the potential to gel more is clear and the path to having an impact on the final standings is obvious. Time will tell if they can do it.