Well, I've seen lots and lots of speculation and suggestions for the Halo 4 Ranking System, but in my view, they all seem to fall into the same traps as all of the previous ranking systems did. I thought I'd give my two shits about it too. Here we go, Problem 1: "I never get paired up with a good match! I either have an easy time or get totally owned by the other team!" simple fix: have a skill ranking system by which everyone gets paired with someone within their skill range. BPR should do. Set it so that you never get paired with anyone outside of 5 ranking (units?) from you. Problem 2: "Ranking systems like these as well as all statistic averages are too fixed. Now I'll never get a good ranking I can boast about! Oh well, I guess I'll have to buy an account with a high BPR." Do not worry, my imaginary whining example of the halo community, if all statistics were drawn from the past two months, then your BPR, as well as your K/D ratio and all of your other statistics will be a more recent example of your skill than in past halo games (Reach was my first shooter. Look me up. I'm a 26). As well as the fact that it would discourage selling accounts because all of the statistics would only last for two months. Problem 3: "I like the Halo reach ranking system and how it gives me a visual initiative to continue playing (hmm... that doesn't sound whiney enough...) WAAAHHHH" By all means, have a linear ranking system as long as Halo 4 uses the BPR system I described above in the background in order to find good (and fair) matches (as well as bragging rights) then there should definitely be a form of linear superiority that dictates whether or no you can call another player, "less experienced", or, as we normally call them, N()()Bs. Recap: A Halo Reach-Style linear ranking system backed by a BPR matching system (that only uses information in the past two months). How does that sound? Please let me know of any flaws you see (or foresee). I would very much love to have a final, perfect ranking system written down somewhere.
But after every BPR reset, you could have Pros fighting N00Bs. Of course that is if they pre-order it.
by two months, I mean the past 60 days. It wouldn't average in days from before that. it wouldn't reset. It would just replace the statistics from 61 days ago with today's statistics.
I'm not sure BPR would work too well. My BPR is 81 and my friends is 71, yet he is a better player than I am, I can maybe score 5 kills on him in a game to 15. I guess what i'm trying to say is it would have to be a little more complex than just using BPR for a matching system. Maybe this is just me, but I much prefer the Halo 3 ranking/matching system over Reach's. Unless it kept your last BPR for the first few games after the reset.
There is no reset. And your BPR inaccuracy is most likely due to the fact that you friend must have been much worse when he started playing, and the problem is that it all gets averaged in all the same. What is this reset you're talking about. Instead of BPR calculating all games you have ever played, it would only use the ones you have played in the past 60 days, thus, giving a more recent representation of your skill as well as giving players an initiative to continue playing.
I'm not sure, my friend is an inheritor now and had plenty of time to get his BPR up. Besides, lot's of good players have bad BPR. EliteWarrior5 was talking about BPR resetting.
but of course, it's always better to use the most current statistics without forcing players to play extremely often or their rating would go away.
Maybe it would be better if they used something more along the lines of the arena ratings, where you play a few games and get rated for each game than averages them out. Though that is similar to the BPR now that I think about it. Still, they could do an average of how many games you played a month, than you start the next month off of that rating.
well then that's a less function version of my idea, because you would have a hard time changing your rating at the end of the month, and your rating would change instantly at the beginning of the moth. With my system, it changes the same amount for every game you play, and if the first game of the month you played was really awful, then the second game would pair you instantly with awful players because it wouldn't have anything to average it out with.
The BPR scale is from 1 to 100 points, and it is not meant to be even close to the Trueskill system that is used in Reach and Halo 3 matchmaking. It just depends on your overall k/d and other small things (the other stuff is shown next to the question mark on a 343 service record page). It does not count who you got these kills or deaths against, however; if within two months you happened to be matched up against more bad players than good players, your BPR would be unfairly inflated. By the way, a certain number of games makes more sense than 2 months. My BPR is 100, so obviously the cap is way too low by any standards. Believe it or not, a more sophisticated system (the Trueskill system I mentioned) accounts for wins and losses against opponents whose own trueskill ratings are factored in to determine how many rating levels (shown as 1-50 in Halo 3) you should rise or fall. It also determines the "certainty" of your rating based on the number of games played and the consistency of results. More "certain" ratings change at a slower rate. This allows new accounts to be properly rated quickly, which is good for anyone who would be matched up with a new account whose trueskill otherwise wouldn't be accurate, but it seemed to incentivize getting a new account in H3, when some players were half-obsessed with getting a 50 because trueskill was displayed. It was also disconcerting for anyone to not know how many higher-ranking players would have to be beaten before his/her skill would increase. The reason why the ranking system in Reach can't be precise is because the populations are too low. In H3 populations were higher and MM was more precise before Reach came out, at which point much of its population left, and I found that there weren't many good players to be matched up with frequently. I think Reach's system might let less time go by before less precise matches are chosen, but your hard limit wouldn't let players be matched for long periods of time. Personally, I would like to see an variant of the H3 system, with experience for games that the player won, and a limitless number rank with increments that were finer than the 1-50 increments (for the ranked playlist). I would prefer that the playlists match using trueskill, but that the number rank displayed does not change at a slower and slower rate. I don't know how many games in the past are averaged, but I would prefer that wouldn't be more than 20.
I just want a system that shows off skill more rather than experience/time spent playing... kinda like Halo 3 but better. I always enjoyed the challenge of trying to get just 1 more point, just 1 rank higher so I could to the next series of ranks. Good ol times.
I had advocated a Starcraft Divisional/1-50 hybrid system on Waypoint before. Ranked ladders would have certain traits: -Separate Ladders for organized team and random team stats. For example, you and 3 friends can play on the ladder with a skill combined, or you could go in alone/with one other person and have a separate skill that takes into account the fact that you search alone. -5 or so Divisions, lets say Iron, Bronze, Silver, Gold and Onyx, like Reach. Each Division gets divided into groups of 200 or so. These smaller groups give players a goal, to reach the top of their group. Armor/Commendation awards could be given to players who finish within the top 16 or so of their group at the end of each season. -Groups divide players further based on a 1-50 system within that division. For example, a 45 Bronze wouldn't be as good as say...a 22 Silver. This would make the 1-50 more precise by essentially dividing it into 250. Numbers would change more frequently helping to show progress better. -Groups would be reset at the end of each Season, which would be 3 months long, like Reach. Trueskill could also be loosened up somehow at the end of it. It would be more lenient at first with who you get matched up with to give room to move up and down more easily at the start of each season. -Group rankings would be decided by points. Points would be gained for each victory and lost for each defeat. More points would be earned from winning than lost from losing. The amount of points you win or lose each match depends on the average trueskill of each team. If you search solo, players who leave the game within the first 2 minutes significantly decrease the amount of points won or lost. This is to prevent quitters from ruining rankings of others. Past 2 minutes, players could exploit the system by quitting if they are losing, so the amount of points stays constant afterwards. -A group consisting of the top 200 players/teams in each Division would be created, taking the top players from the last season and giving them an invitation into this special group. Inactivity or opting out of the ladder would trigger an invite for the next most qualified candidate. The top group ratings, or just the top Onyx group ratings could be displayed on Waypoint to encourage players to try and reach the top spots. -Other ideas include tournaments with different division requirements from time to time and a more sophisticated/built in team system to help teams organize and play each other, kinda like an improved Halo 2 clan system. A 'Bonus Pool' system like Starcraft could be added to allow players who stay inactive for periods of time to come back and have a chance to catch up on their ladder. The goal of this system being a way to encourage constant competition and reward, by making moving up and down a lot more fluid than 1-50 by offering a group of similar skilled players that you compete indirectly against throughout the whole season on top of direct competition in each game. More consistent and visible rewards for victory (moving up on the ladder, climbing the 1-50 ranks, reaching a new division and competing for the top spots in each division) encourage players to stay on the ladder and fight to keep their place in their division and to try to climb higher. It would be complicated, but in my experience I've had more of an incentive to play on the ladder when I can immediately see the fruits of consistent victory and reason to defend my ranking.
What if someones BPR is 100 and he/she's and inheritor/reclaimer? Then who will they be matched with? [br][/br]Edited by merge: I'm only 25 BPR.
I just don't think that this has much more to offer than my system, and it's a hella of a lot more complicated. other players between 95-100
Eh, I haven't been too much of a fan of BPR. You can K/D ***** your way to victory on BPR. The number doesn't mean much. I also don't think the BPR system is really all that rewarding. It's relatively easy to ***** out a 100 if you really wanted to/had a team from what I've seen, and two months of recorded stats isn't a lot, really. Especially if its split between different playlists. I'd recommend more. Also, would there be distinction between ranked/social? If not, how do guests factor in, since their skill can't be tracked? Would there be improvements to the BPR system, and if so, in which areas? Would it be visible? I don't mean to sound rude/butthurt, I'm legitimately curious. Sorry if that comes off this way; it's hard to show mood accurately over the internet. A lot of people play ranked to see visible improvement and a way to distinguish themselves from the rest. BPR isn't done on a percentile-based system, meaning there is no set percentage at 100; It's just a compilation of stats that determine the number. You could have half of the population at 100 eventually. For that reason, even matches can become hard to find with BPR, being that any number of people could be at any rank. 1-50 graphs in Halo 3 showed that the system had some connection to a percentile, being that ranks 45-50 had significantly less people than 1-5. BPR doesn't have that curve. You aren't ranked against other players; you're ranked against a calculator that gives you a number. That's why I'm in favor of a more complicated system that puts a higher emphasis on determining skill by comparison to the rest of the community. Wins and losses against other players determine where you stand.
thanks, this is helpful. I'll put some additions in to make it better. 1. the amount that any given kill influences the BPR changes in accordance to the rank of the person you killed. (However, just as a side note,this isn't completely necessary because of the skill of the player you face regularly in your matches. - since the skill of your opponents is more or less the same, killing awful players won't dramatically inflate your BPR. and even if it somehow did, it would vanish after two months). But you are right. It really helps it from becoming cheatable. 2. What you said about the curves- first of all, if we had them, then the lower 20 percentile would be filled with players who haven't played in the past two months. If the computer calculates that you are a 100, even after factoring in the skill, then you truly deserve to at least have a chance at competing with the big guys. If you do poorly, your BPR would reflect that accordingly and you would move down. It's a self balancing curve without an overly-padded middle. 3. Only competitive playlists affect your BPR. Things such as team slayer, swat, rumble pit, team snipers, and maybe BTB. Not things like multi team, grifball, and race (fingers crossed), that people don't take very seriously and just want to have fun in. Also, not in objective gametypes so that we don't end up with whores who don't PTFO. I hope this has been thorough, and I thank you for your criticism. Feel free to point out any more problems if you spot them. I'm going for simplicity and efficiency, just so you know.
That cleared it up, I can understand. As you may tell, I'm a fan of the more complex systems with a whole lot of depth, but that isn't to say I think this is a bad system. I think it would work well. I think that a cumulative BPR should be considered when a player is completely unranked, having no games played in the last 2 months. That way, they will get some-what accurate matches when starting back up again. I still have a problem with BPR though. If it works like it does in Reach, it's far easier to move up than down. Over time, everyone will migrate upward, even if they don't really deserve it simply because they would have to really consistently do poorly against people deemed their skill to move down. Otherwise, it's all moving up. Unless the system worked more like 1-50 than BPR, which would probably be more accurate. Judging their last 2 months against peer performance, under the same set of rules you outlined, would keep the curve. I'd be happy with that. Other than that, I still have to ask if you would take parties into consideration within the BPR system. Parties and communication gives those players a series advantage; decent players working together will beat great players working alone nine times out of ten. Would party or random searches be considered? I'm glad to see you didn't misinterpret my criticism. I think you have a good idea here (although I still have a fondness for the divisions), and I'm genuinely interested.
Alright: more edits 1. scratch the two months. It's not helpful and it's inconsistent. How about it uses the past 50 matches? that way people who haven't played in a while wouldn't have to wait two months and even as halo 5 comes along, the rank will still be preserved. 2. in regard to parties, the only big advantage I see is the higher frequency of headsets. at the higher level of competitiveness that this ranking system would provide, I think there would be more headsets anyway because people would always be looking for that extra advantage. besides, it's still a huge improvement and I don't really think it has much to do with the ranking system itself. The only perfect solution would be to get rid of parties altogether... but then people would get really pissed off
I think 50 matches isn't enough. I don't think a system gets a fully accurate read on players until...100-150. You need to be somewhat sure when you give people a rank. Maybe 5 to get in, and the most recent 50 have a higher weighting, but I don't think it's enough to define a player. Otherwise, I think it's good. Parties are a bigger advantage the higher you go. Players are much better at communicating with each other, which is very important in controlling the map and finding and killing enemy players, as well as helping each other out of sticky situations. I know in Starcraft, organized teams are usually give weighting of an entire league higher (20% better in the general population) than similarly ranked solo players matched on to a team, and the parties still go 50-50, even out-skilled individually by that much. Maybe match organized teams against players that are technically ranked higher? Like, 70 avg team plays 80s and low 90s? Higher levels kinda get screwed over there, but then the only other solution would be separate ranks for teams/solo. And yeah, no parties is out of the question!
honestly, im not that good at halo, but better than the average player, and i preferred the halo 3 system better, but i love reach's progression system, so a combination of the 2 would be great, and also i managed to get to high 40's in most of the ranked playlists on h3, however by bpr is 45, so i personally think its not accurate, but that might be because all i play is grifball