|
|
| Author |
Thread Statistics | Show CCP posts - 0 post(s) |
Beren Hurin
Onslaught Inc
RISE of LEGION
1973
   |
Posted - 2014.01.15 19:19:00 -
[1] - Quote
I was wanting to try to figure out about how many people I might be queueing with at any given time so I started making some assumptions and throwing some figures around, so if you don't mind, follow me down the math path... Some data is debatable, but hopefully not too far off.
Average Player population: 3,000
Game modes: 5 (Ambush, domination, skirmish, FW MIN/GAL, FW CAL/AMARR) For simplicity of distribution I assumed that the same amount of people would be grouped between the allied factions as were also queued in the other game modes. Could be a high estimate...also I'm ignoring PC
Players per game mode: 600 or 20%
Now not everyone is in every game mode queueing at the same time. Some are in battle, some are idle, some are queuing. I estimated that on average, 85% of time is in game or match, and 15% is queing or idle.
So now what we need to know is how many people could be queing at once. Let's say that half of those players who aren't in game have also just selected to find a battle. So that's 15% of 20% or 3% of all players (45 players).
Also, let's assume that there are a chunk of players who have just left a match and are getting ready to re-queue. 85% of these players will immediately re-queue rather than idle around.
...And we will also assume that the size of this group is (game mode population)/(avg. match length)/(avg match queue time)...
Since there are 510 players playing at this time, and we'll estimate the average match length is about 10 mins for all modes, and the average wait time is one minute, that means that (510)/(10)/(1) or about 51 players are getting out of battle and immediately queuing. This would be around 10% of the players in a game mode at any time.
So with the 51 re-queuers and the 45 players leaving their idle state, that is 96 players or just enough mercs for 3 possible matches in a given mode. That doesn't leave much room for matching squads against each other.
In summary:
When you select one game mode to play you will queue with:
Total Population: 3000
Game mode: 20% of all people in the game (600)
Activity rate: 85% of all people logged in the mode (510)
Match turnover: 10% of all active people in a specific game mode at a time. (51)
Returnees: half of the 15% of people that were idle (45)
(96 total players queued)/(32 players per match)= 3 matches
Depending on your game mode (like ambush) you might have a few more people to match against, or a few less (like FW) but until we get 5000 to 6000 players, the matchmaker gets extremely limited in the matches that it can make.
Bottom line: The only possible better results to matchmaking are longer queues or more players. It cannot possibly do much better with what it's got to chose from right now. |
Beren Hurin
Onslaught Inc
RISE of LEGION
1973
|
Posted - 2014.01.15 19:39:00 -
[2] - Quote
Ghosts Chance wrote:except the main fails of matchmaking are
putting multiple squads on the same team instead of opposing sides
starting games before teams fill wich causes:
one side is full and the other is half empty
one side having a massive head start capping most if not all the bases before the opposing team can even spawn in.
your analysis fails to take this into consideration.
I don't really think it fails that much. You are in a tryhard squad. The matchmaker will place you in the best match. If your squad just stomped a match and another squad was with you and it got 25% of the points of your squad it would LOOK like it's underperforming against maybe 10 players on the other side, depending on how bad the stomp was. The matchmaker could look at those 10 players and rank them higher than your 2nd squad.
Thus dooming the loosing solo players further...
When you have any 6 players getting 50% of all WPs awarded to both teams in a game, it then becomes difficult to have some way to determine the next best 6 players' relative value. |
Beren Hurin
Onslaught Inc
RISE of LEGION
1973
|
Posted - 2014.01.15 20:06:00 -
[3] - Quote
The Attorney General wrote:SteelDark Knight wrote:The did not see it mentioned so thought I would bring it up as it probably effects the numbers. What about the different server regions and how that impacts Ambush, Skirmish, Domination, matchmaking as it further splits up the player base in which to pull matches people from? Good point. Depending on the time of day, there might only be one battle available in your region.
Yeah I didn't even think about server regions. Good thought. It's likely though that if you are in a sparse region, queue times will be a bit longer to even things out.
The thing we don't know is if queue times re-tune themselves based on how lopsided a match is. How would you feel if the reward for a pub-stomp was a +5 minute longer timer to your next queue? |
Beren Hurin
Onslaught Inc
RISE of LEGION
1973
|
Posted - 2014.01.15 21:31:00 -
[4] - Quote
I was just brainstorming a kind of system where, if your squad starts to perform pretty highly stomping a few times, it starts to increase it's "Morale Rating".
Each individual's morale is tracked. The higher your morale, the more you contribute to a longer queue because the matchmaker is trying to find a harder opponent. But additionally there is a reward of a possible kind of SP boost.
I don't know how it would be best to enact, but morale would then have a kind of half life. If you stomp a bunch of matches and get 100% morale, then one of these could be a possibility to implement. Morale would have a half life of 48 hours and would increase as a proportion of the team WP you get in a match.
The higher your morale, the longer you contribute to match length, but the greater your SP rewards.
You could alternatively make squad morale be the factor that increases SP gain. So the higher the total squad morale is the longer you wait for battle, but the better your SP gains.
Ultimately, it would have to be tuned so that the longer wait and the better competition is still worth the increased SP. |
|
|
| |
|