Edit: Post #8 contains draft updates, and the region breakdown can be found here It's officially time for the World Cup cycle to get underway, and first that means announcing the international team draft order! After many long hours among TFS management that surely was similar to the March Madness selection committee, we've finally got the system we want that takes into account the factors we considered most important. I'll explain those factors shortly, but first since you are all anxious, we'll start out with the draft order: So how did we get to this order? Bribes of course!!! But more seriously, the google doc below lays out everything that went into the determining the draft order: (If spreadsheet doesn't appear, try going here: https://docs.google.com/spreadsheet/ccc?key=0AlejV6SLlSf0dDJpNnBqS1UtTnRsSEZnQ0NGeVNVUkE&usp=sharing) Woah, what the hell is all this mumbo-jumbo? Well grab your favorite drink, as your head's about to hurt... OVERALL/GENERAL STUFF - In general, what we have here is a weighted system of (1) results from S14 and S15, (2) total TFS seasons played, and (3) match completion. So in short, the better results you had over that period, the worse your draft position would be. But if you had long service time and a high completion rate, that would help your position. - We assigned weights of 65% to results, 10% to seasons played, and 25% to match completion (the numbers highlighted in yellow at the bottom of doc). - In order to mesh all 3 components together, we have to somehow have all 3 with a common unit set. So to do that, there's rankings (1-32, everyone participating in the draft) for each of the 3 categories, and then the 65-10-25 weights are applied to those rankings. In other words, you can't just add the raw values of points per game to number of seasons completed. So this allows us to create one final number, the WEIGHTED RANKS column. - For the 3 RANK columns, you can think of it as being 1=helps draft position and 32=hurts draft position. That's why shiftee is ranked #31 for PPG, but #3 for SERVICE TIME. - For any ties in the RANK columns, the RANK was adjusted to the average of the ranks. For example, 5 people have 15 completed seasons, but rather than all 5 ranked #1, or randomly choosing to rank them #1 through #5, we took the average of those ranks, thus giving them all a rank of #3 (this also explains why some ranks end in .5). - Looking at the DRAFT POSITION compared to the TOTAL wPPG RANK (more on this calculation later), you can see how service time/completion affects the overall positioning a bit. TOTAL SEASONS PLAYED - Titled SERVICE TIME in the doc, and taken from the TFS archives doc. Just shows how many seasons have been completed. - People who have at some point dropped out mid-season are highlighted in gray and have had their numbers rounded up to show the total number of seasons they participated in. The reason for doing this as opposed to noting a half-season played (such as 14.5) is due to the match completion criteria (explained next). Had we kept service time at a half season, the half-season'er would have incorrectly been double-penalized from the two categories. Another way of looking at this is that match completion in the next section represents the person's completion rate over the number of full seasons they have participated in. MATCH COMPLETION - Last time we tried the World Cup, we used completion percentage. However, in using percentage, that gave a huge benefit to new people having completed only 1 season at 100%, while making it tough for others to rise up the rank. What we really wanted was to reward people who are dependable and always get their matches played, and it wasn't too fair to previously rank someone with 100% over 1 season higher than someone who missed 1 game over 15 seasons. So screw percentages, in with something new!... - The match completion "points" system can be best explained as such: each season, each person receives 1 "point" for completing all of their matches. Miss 1 game, and you get docked a point, thus finishing with 0 points for that season. Miss 2 games, and you get docked 2 points, so you actually finish with -1 point for the season (the full per-season table breakdown is in the bottom-right of the google doc). - Seasons played and games missed were both taken from the TFS archives. S14-S15 RESULTS - The most complex part of the whole thing is quantifying the results of each user from FIFA 15. So I'll try to go somewhat step-by-step in how these were made... - The results component is exactly the same as the one we originally used when we tried the World Cup on FIFA 14. Back then, we evaluated people from S7-S10. But with many new users and the transition to PS4, we kept it to just S14 and S15 results this time. - For both seasons, the doc shows each user's Serie level, games played, and total points. We then made a PPG column so you can easily see those totals too. - After entering in these totals, we made a wPTS column (weighted points). As you can see by looking at the Serie Level weights at the bottom, all Serie A points were weighted by 3, all B pts by 2, and all C pts by 1. (Also, anything in the doc with the "w"-something stands for weighted and that just means adjusted for Serie level weights) - We have to still make these totals on a per game basis due to incomplete games, people who dropped out early, and differing number of teams per league, so the wPPG column is wPTS/game. So this would be the weighted (by serie level) points per game. - To get a total number for S14-S15 totals with most recent results taking slightly more precedence, we kept the same season-by-season weights used from the 4-season format, which at the time was S10 = 40%, S9 = 30%, S8 = 20%, and S7 = 10% (If you look down at the weights box at the bottom, you'll see these weights). But, we don't have 4 seasons of data, so... - For people with 2 seasons played (from S14-S15), we still need to weigh everything they've played out of 100%. Therefore, we took the same percentage of distribution used season-to-season above (40-30-20-10) to create the "Distribution" line in the weights table (100%-75%-50%-25%). That means looking at the most recent season's (S15) weight as 100%, the previous season before that (S14) would be 75% of that weight total, then the season before that (which would've been S13 had we chosen to use a 4-season system) would've been 50% of the most recent season's weight total, and so on. With this distribution in mind, you can see in the weights box how each season would be weighted if a user played less than 4 seasons (so, all of us in this case)... So for people with 2 seasons played, it comes out to roughly a 60-40 weight distribution, but just adding math to create numbers more in line with the 40-30-20-10 original distribution and not just arbitrary numbers. And for those with just 1 season played, obviously their one season is 100% of their results total (so they aren't penalized for having played less seasons). - Alright, so after weighing the FIFA 15 seasons with more recent results carrying more weight, we get the TOTAL wPPG number. So for example, for someone like myself with 2 FIFA 15 seasons played, my TOTAL wPPG would be ~60% of my Season 15's wPPG and ~40% of my Season 14's wPPG. So in the end, this step creates an overall weighted average of my wPPG each season. - And with that all completed, I can create a ranking like the SERVICE TIME and COMPLETION PTS columns that go into making the overall "magic" number (the WEIGHTED RANKS number) And in the end, we can put in order from lowest to highest the WEIGHTED RANKS column to get our overall draft position. FAQ Why are some people in highlighted in red? These people already have a team due to placing a Heritage claim, and thus will not be participating in the draft. So looking at the top 3, PhiGuy179 has pick #1, Lions67 at pick #2 has already claimed Scotland through the Heritage claim process, so then Andanas at pick #3 will essentially pick 2nd. There are three ties in the draft order, what are we going to do there? Through all the tinkering of weights and components, it was pretty rare to see ties, so that's sure an oddity. Somehow, abolishthefed and Lou both have played 2 seasons, both have completed all of their games, and both have totaled the exact same amount of points each season. But luckily, we get to avoid the Don Garber-approved blind draw since Lou already claimed Italy as his Heritage claim. As for the other two ties, SERVICE TIME is the tiebreaker, which means Jugster will actually be pick #22, and wSeals9 will be pick #25. Will this whole method be used again for the next World Cup we hold on FIFA 16? While we could make changes here and there (and plus it is kinda early) we will still likely find a way to incorporate past results, service time, and match completion at the very least. And this process did end up being very similar to the one we set up when we tried the World Cup on FIFA 14. So this may be something to keep in mind as each future season goes by. Hmm, so I should really work hard to make sure I complete all my matches from now on? Yes Next time can I just send in my bribe money so that I can avoid understanding all of this and be assured of the #1 overall pick? Sorry. Unfortunately, Sepp Blatter does not have a say in TFS proceedings. If you have any other questions, post here and I'll be able to explain it further. The draft is set to officially open this weekend, so get practicing with those int'l teams!