Jump to content


  • Content Count

  • Joined

  • Last visited

  • Days Won


Everything posted by Trunkers

  1. I dunno, I am kind of coming around to Cratos. Competitive communities usually benefit from that type of personality. What I'm trying to say is, I really miss Zyos.
  2. The xbox.com website won't let me purchase the team skin packs because my region doesn't match my current location, nor will it allow me to change my region.
  3. Alright, I found a way to make the plots look *nearly on good on imgur as they do locally on my machine. (click on the image to enlarge) This is RNG. They started of on top of the world, with a 4-1 record. Since then, they've dropped three on the trot. Although it's tempting to point to individuals, the RNG story is a complete failure of the collective. Commonly, Victory X and Ninja have each posted their three season lowest k/d's in the last three games. Let's let that sink in. Individually, three of the team members have been at their worst. The story doesn't get much better for Penguin. His last 3 games represent three of his four lowest k/d games in the league thus far. In terms of slaying power, Commonly does appear to lag behind his teammates somewhat. He's posted the team-low k/d in 4 of their 8 matches. Ninja, on the other side, has posted the team-high k/d in 4 of 8. Regardless, with the transfer window closed, it's up to Renegades to find some way to right the ship with the crew they've got. (all data collected from halo5arena.com)
  4. (click plot to enlarge) Using white here really made the plot pop, but the downside is that it messes up the legend. Whatever, you guys can figure out which player has "gone missing", I'm keeping it this way.
  5. Just a preview, I am making these for every team. (click on the image to enlarge) I wrote a function (or rather, set of functions), to be able to visualize a team's data by matchday for a given statistic (in this case, 'kd'). Then, I plot the average kd for each player as a vertical line, and their k/d for a given series as a dot on that line. Everything is color coded, so you get a players' color at the bottom, and you can see the team colors on the left hand side (these colors are consistent with the other plots I have made). I'm making a custom color palette for each team, for visual distinction. Allegiance gets an orange-based palette. Main messages here: Rammyy's two sub appearances for Allegiance were far, far better than any individual performance other Allegiance players have put up. Notice that his sub appearances account for 2 of their 3 league wins. The other league win? Ryanoob's single match with a positive k/d, back on Matchday 3. This plot is still a work in progress, so any feedback is appreciated. Thanks as usual to halo5arena.com for giving me the data.
  6. Yeah, now that you mention it, Commonly does sound correct.
  7. Yeah, it's something I am trying to figure out the ideal way to do at the moment. The problem with just doing the change from average is that the percentage change will decrease over time. Versus previous week (as you mention) is an option, but so much of it seems to be opponent-based, so I want to be able to highlight things in a meaningful way. It may be worth starting to make time series plots to show variation in performance over time.
  8. Going to drop a couple of quick graphs, courtesy of the data @@charas just sent me. (halo5arena.com, muchas gracias) Look, I know it's a small sample size, and that ALG played against basement boys Liquid, but hot damn Rammyy. Those are the numbers you put up as a sub? Also, since I now have the raw data (read: fully correct data), I recreated this plot using the data from weeks 1-3. In the future I may do cool things like show the before and after on one plot, but for now you'll have to see it side-by-side. (click the image to enlarge) A five-game slugfest between CLG and nV brought both team's numbers down a notch. Huke maintains his killing rate, while reducing deaths per minute. All of EG's numbers improve, particularly Snipedown's. Ninja manages to increase his deaths per minute. That's all I have time for at the moment, feel free to add your own observations!
  9. I'll be buying a team skin, if nothing else than to send the message "thanks for promoting / reviving competitive Halo". All the DLC has been free so far, and regardless of what everyone says on Reddit about the volume of content in the game, I'm more than happy to throw some cash towards the teams/developers.
  10. Man, CLG has such composure. I remember a month back or so, CLG were trailing hard in another Eden TS game. Like 10-15 kills hard. Some other pro (can't remember who) was in the chat and said "CLG will come back, I guarantee it", and sure enough they did. What a series. Thanks Envyus for stepping up your game. At the start of the season I wasn't even sure you guys were Top 4 material, but you've really shown that you are a top tier squad.
  11. Thanks for the kind words. I emailed a couple places and never got a response. Brief eSports resume if anyone is interested: I worked MLG's Senior Staff Writer from 2006-2007, and later as a freelance writer from 2007-2012. During that time I primarily covered the Halo and Starcraft scenes. I've also written for ChannelFireball (top Magic: the Gathering site) and Melee It On Me. Please see below for links to previous eSports articles I've written: http://www.majorleaguegaming.com/news/the-beautiful-videogame http://www.majorleaguegaming.com/news/3-keys-to-a-great-competitive-game http://www.meleeitonme.com/recapmlg2006/
  12. To be honest, part of my hope when making these is that it makes it easier for the casters to make informed commentary about the different teams and players. Back in the day I would sometimes slip notes to Puckett in the commentary booth, to highlight interesting facts or trends. So much of selling sports (e or otherwise) is building up reputations for your players, so that you can craft story lines around that. Numerical analyses are one such angle.
  13. (special thanks to halo5arena.com for providing the data for this post) Spend a week in the Halo community and you'll learn that stats are controversial. Perhaps you've heard a forum poster, or a match caster say something like “stats don't matter”. But if you've played Halo, you've probably also had a thought along the lines of “man, we would have won if my teammate didn't drop a neg bomb”. Usually these arguments tend to hit the extremes. Stats matter, but they aren't the only thing that matters. You can even win games while being outslayed! Today I'd like to take a first step in exploring the concept of objective efficiency. I have so many questions! At the pro level, how often does the team with more kills win an objective game? Does it vary by team? Which teams have performed the worst while outslaying their opponents? Let's make a graph. Each point represents a single game for a team. Slayer games are not plotted. On the x-axis we have the team's overall kd. If your team had 50 kills and 40 deaths, you have a team kd of 1.25. On the y-axis, we have the percentage of objective points your team earned. Win a flag game 3-1, and you have 75% of the objective points (3 out of 4 captures). In total, there have been 53 objective games where one team out-slayed another team, and of those, the team with more kills won 47 times. 89%. Round it to 90, whatever. The other 6 games? Those are in the lower-right corner of the graph. The most severe instance of a team getting beaten despite outslaying their opponents was in last week's Truth CTF game between Evil Geniuses and Envyus. EG had a whopping 25% edge in kills (60 to 48), but lost the game 3 – 1. Snipedown went 23 – 11! But if you watch the replay, it's a trainwreck. EG also lost two objective games when posting an even kd of 1.0. The biggest repeat offender is Team Liquid. They've lost 3 objective games despite slaying superiority, and another game to Optic where the kills were equal. On the other side of the coin (the top-left corner), Envyus have taken two games while being outslayed (against EG and Allegiance). They've also had the most games where they've taken 100% of the objective points. Envyus' objective win-rate is topped only by CLG. Thanks for reading!
  14. Ah, good point. Here's a link to the match: http://halo5arena.com/matchup/1541 looks like no, there's only a single game there. I'll see if we can track down the missing game.
  15. Exciting news in the world of Halo5 plots. @@charas reached out to me about producing plots for halo5arena.com, and the end goal is that I will make some nice graphs that can be hosted directly on the site. In the meantime, charas gave me a table with all the raw data I requested, which is really a treasure trove of information. Now our data is much more accurate, and I'm no longer computing deaths and kills per minute as derivations from damage-based statistics. Now I have raw kill counts, raw game time, and can compute these things directly and without error. As a comparison, we can compare our last graph (http://i.imgur.com/CRmdUwt.png) to the corrected version of the graph (http://i.imgur.com/te84gZA.png). I have some hypotheses for the discrepancies (it's not really relevant here), but the main point is, from here on forward, we have totally correct data. Today, I want to scrape the surface of what we can do with this new data. –- Imagine a world where teams only play CLG. Picture it. Who thrives in this world? Put some names in your mind. (We're going to make a plot using only the data from CLG's matches. Note, Envyus has not yet played CLG) Were you right? No, you weren't. It turns out that the individual players who did best against CLG were from Optic and Liquid. CLG played both of these teams in Week 1 of the Pro League, while the gametimes were still fresh to both teams, and the playing field arguably more leveled. Ace's 65 – 55 performance stands as the best statistical showing against CLG of any player thus far. To date, Optic Gaming is still the only team that has taken a game from CLG. As for the other end of the spectrum? One man struggled far more against CLG than anyone else, and his name is Cratos. Now I've watched enough of Lethul's stream to know there's a bit of a beef between Cratos and the reigning champs. Was it in effect here? Maybe. The point stands, Cratos went 18 – 42 in the match. And it wasn't just his kills that were off, he also posted the fewest assists in the series. Thanks as always for reading. If you have a question you'd like to have explored statistically, feel free to send me a message. (disclaimer: no promises that any question has a clean statistical solution)
  16. Somebody have a summary of all the changes? Is it time to update the OP, or are things not yet official?
  17. I had my bumper break and my charger port stop working. Luckily the store replaced my elite free of charge. Or maybe I just have meathands.
  18. I like Halo 5 way more than Overwatch, but what I find when playing Overwatch is that there are items that I actually want. Not much in Halo 5 really gives me that, and the monetization scheme doesn't offer me much reason to open my wallet. This will change when the team skins come out, but I think 343 can do better. Some concrete examples: 1) Give all customization to the pro player accounts. What better way to show off skins than during the Pro League broadcasts? Unfortunately, this goes for the LAN events too, where we always see players without their actual icons. 2) More opportunities to show off our characters. When a game ends, I don't even know who is who in the Spartan line-up. Put the player names on there! Highlight one player as the MVP or something. Steal the Play of the Game idea! 3) Make the REQ system more of a lottery and less of a grind. I am pretty convinced that not all items of the same class have equal rarity. There's a reason people get all their DMRs at the end of unlocking silvers, it seems to kind of withhold items. All-in-all, there's not enough "cool stuff" that I actually want in the packs, and even the things I want, I don't have the perception that I could get them without acquiring tons and tons of packs. All said, I do like the direction Halo is moving in, and love the current Pro League action. Halo as a franchise is still in damage control mode from consecutive flops, but Halo 5 represents real progress. It's a shame so much of the discourse around the game is so negative.
  19. Yeah, I can do that, it's all just script-based, the only manual part is tweaking labels. Ideally I would have a site set up where people could flip through different graphs. Other issue: currently I am making graphs from all the total data. Graphs for an individual week's worth of data are potentially interesting, even though the sample sizes are a bit dated. You could do a lot of cool things though: e..g. graphs showing how all players have done against CLG, graphs showing a player's k/d over time, etc.
  20. On the subject of doubles, I would love it if somebody made a 2v2 map with two strongholds, but in order to be scoring you have to control both of them. I think it would lead to some pretty interesting gameplay.
  21. New plot this week. Today we're going to look at a couple statistics that we've never really examined before: kills per minute and deaths per minute. (Fine print: Before we dig in, I'd like to provide a disclaimer that these are both derived statistics. I pulled the data from halo5arena, and we get deaths/min by dividing damage/minute by damage/death. Hooray unit conversions. Then kills per minute is k/d * deaths/min. Since these are derived from damage, and damage doesn't always seem to be 100% accurate in Halo 5, there's some margin for error here. Also, this week, I updated my labeling function to make it easier to label every single player without overlap.) Reading the plot: the diagonal line here is the line where kills = deaths. k/d = 1. Players above that line have a positive k/d. In the abstract, the best players are in the top-left: high kills and low deaths. Let's focus on the y-axis for now: kills per minute. Interestingly there appear to be five “tiers” of players when it comes to kills per minute. Top tier – Royal 2, Frosty, Huke, Snakebite, Snipedown, Ninja Once more, Huke and Snipedown are both going off, despite all the other Enigma6 and Evil Geniuses players going negative. Ninja is off in the Ninja corner. High tier – Pistola, Stellur, Mikwen, Penguin, Contra, Str8 Sick, Ace Mid tier – Lethul, Victory X, Lunchbox, Suspector, Commonly, Predevonator, Roy, Ryanoob, APG Lethul is an interesting case here. He actually has the 5th highest k/d in the league (Huke is the only non-CLG member who surpasses him), and the 2nd lowest deaths per minute. Lethul doesn't play at a blazing pace, and consequently his stats look much better “per death” than “per minute”. Also note that Lethul is the only player from this tier or lower that has gone positive (k/d wise) in any meaningful way. Bonus: Commonly and Suspector have virtually identical stats. Low tier – Spartan, el town, Rayne, Eco, Cratos, Maniag, Goofy, Shooter, bubu dubu Bottom tier – Danoxide, Assault We'll take Danoxide's stats with a grain of salt, as they're based off of just one week of matches. He's looked better than Assault, but worse than the rest of the pack. Thanks for reading. So far I've made a different type of graph each time, and I'm interested in hearing what you guys find is useful. My intuition is that we probably need a set of different graphs each week.
  • Create New...

Important Information

By using this site, you agree to our Terms of Use & Privacy Policy.