Hi. You might know me as that one scrub who writes fantasy guides on Reddit sometimes. In the comments, I got a lot of requests to do guides on Compendium Predictions, and I always said no because it wasn't my strong suit. But I thought about it some more, and realized, most of what I've done for Fantasy Dota has been gathering and reporting data, why couldn't I just do that for something else? So here's my attempt at it. This is PART TWO, where I cover the Team section and the Players section (Part One, which covered Heroes and Tournament picks, is over here). Also, the usual disclaimer applies. The thoughts and conclusions in this piece are my own. I encourage you to read it and take mental notes, but I also encourage you to listen to what others say too. I promise your Merlinis and Purges of the world are going to approach this stuff a lot differently than I will. Gather as much information as you possibly can, and then make your own decisions. And for those interested, I used Liquipedia, DotaBuff and Open Dota as my primary sources of information to pull from when doing each of these. I had no preconceived notions when I went in, I just filled out my predictions as I filled this out. Anyway, let's get started.  

TEAM PICKS

Which team will win? What a rotten question to start this statistical analysis with. This is the question that most depends on your gut, and that's not really what this guide is supposed to be about. Regardless, I think if you want to make a conventional pick, you only really have three choices: Liquid, VP, and LGD. VP being the team that won the regular season speaks volumes, and if you want to pick them based on that, go for it. Liquid are the defending champs, and not only that, they were the team that won the most recent major tournament. So they're a fine pick too. And of course, if you believe in the whole Even-Numbered TI Asian Team thing, then you go with the Eastern team with the highest ranking to win it, and that's LGD. I'm personally going to pick them because I found a tab on Open Dota that ranks teams based on ELO, and that has LGD as the highest rated team (Liquid is two, VP is three, VGJ.Storm (!?) is four). So that's good enough for me. THE PICK: LGD Team with the most Kills in a game Kind of a pain in the butt of a category for me to track because I had to browse each game one-by-one, like so. But I went and combed for any team who met 45 kills in a game (from the Supermajor and Qualifiers) and tallied them. According to my tally, six teams pulled this off multiple times. The first five I'll list did it twice: OG (51, 47), paiN (55, 46), VP (50, 48), Newbee (48, 45) and TNC (57, 45). But the sixth team, Vici, did it a whopping four times (56, 51, 49, 48), and that's who I'm going with. To make it all the better, Vici averages the highest match duration of any team at TI (spoiler alert for later, sorry). THE PICK: Vici Gaming Team with the highest Kill avg Two teams averaged 32 kills per game or more at the Supermajor (LGD, VP), and none quite got there in the Qualifiers. Since I personally picked LGD to win the whole thing, I'm going to hedge my bets a little and pick VP for this category. If you're seriously itching for a dark horse, the teams that cracked 31 at the Supermajor (Mineski, Vici) and the Qualifiers (Winstrike, paiN) are all viable too I suppose. THE PICK: Virtus Pro Team with the fewest Deaths in a game The first thing I did was check deaths/game averages and got Fnatic as the low water mark at the Qualifiers (17.73), so that's an excellent first impression (VGJ.Thunder was the lowest at the Supermajor). But we care about brutal stomps for this category, so I went and treated this like I did two questions ago, and tallied every game that had a team die 5 times or less. Only three teams managed it twice: EG (2, 4) did it against a trash team and complexity (not bad); IG (2, 3) did it against IG. Vitality and Keen (not bad); Secret (2, 5) did it against Vici (WHOA!) and Final Tribe. I think Secret is the best team I just named and thus most capable of delivering a vicious thrashing, so I'm rolling with them. THE PICK: Secret Team with most Assists in a game I'm not exactly sure how to find team-assist totals broken down per game, so I did the next best thing and found assist averages. Between the Supermajor and Qualifiers, three teams averaged 70+ assists per game: Vici (75.5), Mineski (73.1) and Winstrike (73.3). Remember, we're not picking for average, we're picking for who we think the very highest instance will be. Winstike is an interesting case because I remember when I was calculating game-time averages, Winstrike was by far the most volatile sample I dealt with (lots of quick 20 minute games, lots of long marathons, everything in between, they were all over the map). Yeah, Vici and Mineski both project as "better" teams...but this is a rare opportunity where I get to appreciate volatility. Consider this...Mineski and Vici both average longer matches than Winstrike. Winstrike had lots of brief matches as previously stated, and STILL logged a super high assist total. That either means they get a ton of assists in their quick games (whatever), or it means their long games have a TON of assists (intriguing!). Or a combination of both. Look, if you want to pick VG or Mineski here, go for it, those are great picks. But I had to make ONE wild card pick somewhere in this guide, and I think this is my moment. I've talked myself into it. THE PICK: Winstrike Team that wins the Longest game I crunched match duration numbers for the teams participating in the Supermajor and/or Qualifiers, and the four longest averages belong to VG (44.9), Newbee (44.0), IG (43.4) and Secret (42.4). You can certainly imagine all four of those teams in long marathon games. You can certainly pretty easily envision at least three of them winning a long marathon game at TI8. All I'll say is the longest game from the Supermajor belonged to Vici and Mineski (88 Minutes), and Vici ended up winning that one. It would be weird to me if they got the longest game two tournaments in a row, so I'm going with Newbee just because they're somebody else. THE PICK: Newbee Team that wins the Shortest game Conversely, the two teams who BY FAR average the fewest amount of minutes per game are OG (32.0) and Liquid (31.8). Both are fine choices, you can picture both rolling through a weaker opponent like crap going through a goose. I'm personally picking OG. THE PICK: OG Team with highest Game Length avg Go back and read my reasoning from two questions ago for the longest single game. While my gut tells me they won't necessarily be in the longest single game, I think Vici will still pull off the longest average based on past performance. However, Newbee, IG, and Secret are all very fine picks as well just like before. THE PICK: Vici Gaming Team that picks the Most different heroes Believe it or not, the team that had the most diverse hero pool from the Supermajor was Optic, who were credited with 72 different heroes in 26 matches. However, LGD and Liquid played just 16 matches a piece and tallied 56 and 55 respectively, which really isn't bad. I think all are fine picks, but since I haven't had the pleasure of choosing the defending champs on any of these picks yet, I'm picking the horse. It just feels like the right thing to do. THE PICK: Liquid Team that picks the Fewest different heroes This question has always represented something a little different than what it's asking at face value to me. For me, the best way to limit a team's overall hero selection would be to limit their total number of matches played. In TI8, that means picking the team with the lowest chance of getting out of group stage (or, if they DO get through, they get bounced in a BO1 immediately thereafter). According to Open Dota's ELO rankings, Serenity has BY FAR the lowest of any TI team. So I already lean heavily toward this team, but it gets even more compelling. Not only does ELO hate them, they also had the shallowest hero pool of any team that qualified for TI, playing a paltry 31 heroes across 12 matches (next lowest was VGS.Storm's 36 in 9, followed by OG's 41 in 16). I think you have the potential to be a Cinderella story, Serenity. It's nothing personal and I wish you well. I'm just playing the odds here. THE PICK: Team Serenity  

PLAYER PICKS

Player with highest Kill avg At the China Supermajor, two players stood above the rest in this department: Somnus ' M (12.0) and Moonn (11.57), and I will also mention that Armel simply because he ranked fifth in that tournament (10.23) and led all players in Main Qualifiers at 10.89. I would also add that Moonn's numbers are a hair inflated because he had a 30-kill, marathon Storm Spirit game, which led all players. Somnus ' M strikes me as the BETTER option (his average was higher), and he's on the supposedly BETTER team. So I'm going with him (it matches my pick of LGD winning it all anyway, which is nice and convenient). THE PICK: Somnus ' M, aka The Artist Formerly Known as Maybe Player with most Kills in a game Let's look at who's demonstrated that they score oodles of kills by tallying everybody between the Super Major and the Main Qualifiers whose put up 25 or more kills in a game. We get Moonn (30), Armel (28), CCnC (27) and Paparazi (26) from one, and nobody at all did it in the qualifiers (among players who qualified for TI, only ana and w33 even managed 20). So I think your choice lies in there somewhere. CCnC doesn't appeal to me because he topped out at 16 kills in qualifier season. Armel doesn't appeal to me because his own teammate (Raven) registered multiple games over 20 kills in his own right, and I'm worried about that thievery limiting him. I'm going with Moonn, in part because he registered the highest sample available, but also because he hit 22 on a different occasion, so I appreciate that he's demonstrated at least a small amount of elite-level consistency. THE PICK: Moonn Player with lowest Death avg I know I just mentioned him in the previous section, but to me, the clear favorite here is Raven. The man had the lowest death average (3.08) at the China Supermajor and turned that around by finishing second in all of the open qualifiers (2.33). That's supreme consistency. The one obvious alternative is Resolut1on (5th-lowest at the Supermajor, followed by THE lowest in qualifiers), but I put more stock into how players did against the higher level of opponents at the Supermajor. Other honorable mentions: Freeze, Sylar, RAMZES666, and Abed. THE PICK: Raven Player with highest Assist avg The three players I'll mention here are LaNm, Universe, and Tims. The case for Universe is that he led everybody in qualifiers with an incredible 19.5 assists per game (beware, that only accounts for 6 of his games, the rest are missing from DotaBuff). The case for LaNm is, while his number was less impressive, he also led all players in assists per game at the Supermajor with 18.13. The case for Tims is he's a consistency king, finishing third in assists (17.23) at the Supermajor and second in assists (18.11) in qualifiers. Since I just picked Raven, I'm not super pumped about the prospect of putting a bunch of eggs in my TNC basket, and between Universe and LaNm...man, that's a tossup to me. I'm personally going with LaNm for the reason that VG averages the longest matches, as previously mentioned. THE PICK: LaNm Player with most Assists in a game There were five instances of players hitting 35+ assists at the Supermajor: YapzOr (45), eLeVen (39) and LaNm (38) did it in the same match, and then Tims (37) and Kuku (36) each did it in the same match. In Qualifiers, it never happened (at least not among players who qualified). YapzOr also had a 29-assist game, while kuku had two of them. So I lean toward those two, and I'm just going to go with the guy on the squad that I think will be making a deeper run in the tournament (deeper run in TI8 = more games played = more opportunities to register a whole boatload of assists in one of them) THE PICK: YapzOr Player with highest Last Hit avg This one's a little strange. Of all players who went on to make TI8 at the Supermajor, Resolut1on led everybody in average last hits at 413.31/game. However, when qualifiers rolled around, he dropped to 10th overall at 327.33. That's a huge dip, and not only that, people's numbers are supposed to get BETTER against inferior opponents, right?. Maybe the competition going down made him feel comfortable not farming as hard? I dunno. So...I like his ceiling the best. But if you're looking for a more "safe" option, consider Sccc (381.21), rtz (359.63), Paparazi (353.13) or RAMZES666 (352.59). I'm personally going Sccc because he had the highest average among "safe" options, Newbee plays very long games on average...and he's also very handsome, which must count for something. THE PICK: Sccc Player with most Last Hits in a game Set the threshold at 775 last hits, and between the Supermajor and Qualifiers, it happened four times: Ace (830), Moonn (820), SumaiL (790) and Abed (785). I have no strong feelings about anybody on that list, but I'd recommend picking one of them. Beyond that, all four players are on teams that average over 40 minutes per game. So...personally, I'm going to pick Abed because I haven't had a Fnatic pick yet. THE PICK: Abed Player with most GPM in a game So...in the China Supermajor, there was one and only one game when somebody cracked 900 GPM, and it was Paparazi who did it. In Qualifiers, the second-highest GPM among players who wound up qualifying for TI was a paltry 800. First place? 1061!!! How did that happen? It was SumaiL...and it happened because he picked Alchemist. Alch is such a game-changer for this that the meaning gets morphed and bent and you end up asking something else entirely. You're only answering this straight up if you think Alch doesn't get picked all tournament. Otherwise, the REAL question you're asking is "who is most likely to pick an Alch?" SumaiL obviously did, but I'm not sure he does it again now that EG tipped their hand that it's in their repertoire (throw him in the list of suspects anyway though). Other than that, you have to get super obscure to find other examples. MidOne played him at the Thailand Minor (they lost). Ceb played him offlane against Newbee at EPICENTER (he won, but I doubt he took greed). Uhhhhhhhh. I'm not going back any further. Whatever. I'm going to go out on a limb and say no Alch's will be picked, and I'm just going to play this face up and choose Paparazi again. And hey, if HE somehow picks Alch, all the better. THE PICK: Paparazi Player with highest GPM avg Reso led this category at the Supermajor, and while he only finished second in Qualifiers behind paiN's hfn, I'm sold on that level of consistency. THE PICK: Resolut1on Player that plays the Most different heroes This isn't very readily available data, so instead of mashing my head against the wall trying to count this all out, I took a shortcut. As previously discussed, in the Supermajor, the team with the most diverse hero pool was OpTic, who, per DotaBuff, had a pool of 72 over the course of 26 matches. Thus, I studied their last 30 matches to find which of their players had the deepest individual pools. 33 had the deepest pool (16 heroes in 30 matches) on the deepest team, so I went with him. In case you want another choice, CCnC finished just behind him with 15. THE PICK: 33 And that's all she wrote for Compendium Predictions. Again, here's Part One with the Heroes and Tournament picks in case you missed it. Thanks for reading. PHOTO CREDITS: Kuroky and Team Liquid at TI7, courtesy of Valve; YapzOr at ESL One Katowice 2018, courtesy of Turtle Entertainment