Honestly, I'm not a fan of how the rankings are implemented here on Smashboards. I don't like the "flat" scores for events. A local is ALWAYS worth 25 points; wether it has 8 people in it, or 32. Meanwhile a local with 33 people instantly doubles to 50. There are no calculations being done to handle the size of tournaments.
The system can be set up so that a tournament is worth 1 point per player in the tournament. So a tournament with 8 people is worth 8 points, a tournament with 32 is worth 32 points, etc... Now you may be thinking "doesn't this theoretically produce situations where a large local that randomly gets 80 people in it could be worth more than a major with only 60 people in it? Yes, it can... however, I've also put in safe guards against that.
Firstly, you can set it so that a major is worth 2 points per player instead... so it scales faster. Secondly, you can put in limits for each of the categories. So you can say a local can only be worth 32 points at MAXIMUM; no matter how many people enter. So if 80 people show up for a local, that local will max out at 32. This can actually produce more accurate rankings since if 80 people show up to a local, that doesn't necessarily mean it had stronger competition than a 32 person local; since the majority of those 80 people probably weren't that good in the first place. Meanwhile there are regions that only get 32 person locals, but they are full of killers.
Also, there are specific reasons why the ranking system doesn't use ELO. Most of these reasons, I've spoken about in the past.
The problem with ELO is that it has flaws in tournament design. Its really made for 1 on 1 matches. It is considered a "race" system, it can end up rewarding people simply for attending the most events... and it has a problem where it can severely punish people for simply having a bad day (something very common in fighting games). The Association of Tennis Professionals Entry Ranking system was designed to combat a lot of the inherit issues in ELO.
The tennis ATP system is probably the best "neutral" ranking system; since it bases itself solely on tournament results. It doesn't care who you played against, just how you performed. Player A can be the best player in the world, who only ever loses to one person, his training partner Player B... but Player B has never won a tournament... with ELO, there is a chance Player B could be ranked higher than Player A. Since ELO would consider B beating A, as a "bad beat", and include it in the rankings. Meanwhile, B gets a large windfall because he beat the top ranked player.
ELO also suffers from the issue where it actually promotes top players into NOT attending events. If you are high ranked on the ladder, you may not want to attend anymore events because you put your ranking at risk. This is where you have issues of when a new game comes out and some player abuses the meta; in time people will learn how to beat him, but by then he has earned himself a top ELO rating. Then he drops out from tournaments and simply plays online and trolls the forums. Unless you do a yearly ELO rating reset, he's going keep that rating forever... and a yearly reset would upset the people who still attend events, by throwing their hard-earned work in the trash... for a non-league tournament structure, this is not recommended. In addition, you may have issues where tournament players will selectively choose what events they go to, depending on who else is going to be at that event. Maybe they don't want to play against a specific player and ruin their ranking?
The ATP system is designed to "stale" older rankings. So it promotes people to continue attending events to meet the minimum event requirement. In addition, it automatically rolls retired players out of the rankings. As well, since it doesn't care who is actually in the event, people don't have to be selective about what tournaments they attend. Basically, the issue with ELO is all social problems; not mathematical problems.
The biggest problem with ELO however is the timing mechanisms:
ELO is also sensitive to the ORDER of matches... because each match is representative of itself. Its not just about who you beat, but also when you beat them. Yes, there are algorithms to take care of ELO ranking in single events, but not if events are submitted in the wrong order. For instance, I submitted an event that happened yesterday. The rankings would be handled based on that date and would preclude the adding of new events before that date; since ELO is based on order. In order to add new events previous to the date; you would have to rebuild the entire ranking system after every single tournament that is submitted. The ATP system allows a rolling ranking, without a concern for the order in which events are submitted into the system.
When you have hundreds of tournament organizers running events, you can't really expect them to all be responsible and submit their events in a timely manner. GarPR.com is now doing power rankings using ELO; but this works because its ONE GUY (or one group of people) submitting and managing the entire thing. He doesn't have to wait and hold on the irresponsibility of others.
