This would be easily abused. The moment a team of 4 decides a game isn't going their way, they have one person on that team quit and save the rest of the team from any rank loss. A party could go in with an alt or smurf account as the designated "quitter", so that their main account isn't negatively affected. Then they could just make a new account anytime their smurf gets banned, becoming an endless cycle of cheating the system anytime a game doesn't go their way.**So in the rare occurrence that a to3 actually beats a to4, the system recognizes that as the to3 performing much better than expected and deserving of a high CSR gain. If the to4 wins, that's normal and doesn't need to be rewarded any extra. A to4 beating a to3 isn't exactly impressive; it's expected.**
I am inclined to agree that a to4 over to3 victory is expected, mathematically speaking. However, I make my statement based on my over sixty days of Halo 5:Guardians arena experience coupled with opinions I've received from various experienced players. Not to mention a total of fifteen years of Halo multiplayer. More often than arithmetic would like to support, to3 over to4 victories do occur due to manipulation.
I'll provide two hypothetical scenarios for examples.
Player one of Team A achieves an in game score of 12K, 5A, and 9D midgame.
Player two of Team A achieves an in game score of 9K, 8A, and 9D midgame.
Player three of Team A achieves an in game score of 10K, 7A, and 8D midgame.
Player four of Team A achieves an in game score of 0K, 3A, and 11D Midgame.
Team A has a score of 31
Team B has a score of 37 and is expected to win according to mathematical prediction.
At this point, Players One, Two, and Three can easily manipulate the system by harrassing an obviiously frustrated Player Four into quitting in order to secure a potential victory (As the game is not going their way with player four contributing to Team B's lead significantly). In this Scenario, a designated "Quitter" fails to exist as a variable and the ranking system is just as easily manipulated by a yet to be to3.
Often times, Player four of Team A, will quit out of sheer humiliation. Now given the statistics relative to Player One, Two, and Three, it would appear there is an astronomically greater chance of a "comeback victory" from Team A when player Four is manipulated out of the equation.
On this occasion the ranked playlist in review is SWAT.
When a hypothetical quitter (A Player Four for example) removes themselves from the game,a massive imbalance occurs in the favor of the to3.
"Why?"might one ask?
Without shields Players are eliminated by a single headshot.
Considering all to3 players and to4 players have a similiar accuracy, you will be able to observe a simple pattern of addition occur when the point begin to add up.
Team A (to3) may only concede three points in a single rotation whereas Team B (to4) is able to concede four points. 3+3+3+3=12 compared to 4+4+4+4=16.
The reality here is that in SWAT, it is much easier for a to3 to even the odds after eliminating one player. In regular Slayer, shields make overtaking the opposing team near impossible. In shorr, in ranked Swat the system can simply defeat itself without even having to be manipulated .
If we're gonna consider there being a possibility of a 4stack having a designated "quitter" to manipulate the system being a reason to not balance the loss and gains of csr inn these situations, we must acknowledge the facts.
Teams can still easily manipulate the system in the exact manner in which you stated previously..
No matter what, there will be people who are going to manipulate the MMR,CSR ranking model.
I stand firm with many others that it does not seem nor feel fair to be punished for performing so well in a 4v4, that an opponent quits out of frustration, out of coercion, or out of simple will to manipulate the system to save the ranks of their teamates..
If Team A wins by a steak dinner measurement without any opponents quitting, they are rewarded in handsome csr gains.
If an opponent quits though, Team A is penalized with a gain of +1 CSR which may as well be nothing.
I believe a severe punishment ought to be implemented for quitters in the form of a 24HR or greater IP address ban depending on the number of times quit from a particular console in a given period of time. This may help reduce the rate of quitting and eliminate the potential for abuse by designated "quitter" alt or Smurf accounts as you said.