This was the old scoring system for when I tried to run this during FL13: https://docs.google.com/spreadsheets/d/ ... sp=sharing
Parts of that scoring system are no longer relevant since we had more medals to work with (such as the passing/interception type medals) in Halo 4 between ricochet and TFC's stats site. I believe another variation I was planning for SML14 (might've been after that, actually, I don't recall) used a team defense system rather than a pure defender position, and that variation also normalized all tanking/defense stats per minute rather than per game, but since there isn't an easy way that I know of to get game time for stats in Halo 5, that might not be possible.
I will say that, tanking-wise, I like using stats like multikills as barometers rather than pure K/D. Successful double/triple kills and sprees put your team in advantageous situations that generally lead to pushes and goals. In the case of H5, you also have tracking for Goal Offenses and Defenses which can indicate an ability to finish off pushes at the opposing goal or land clutch kills on defense, respectively.
Here is a more modern system that I had originally created for use in a GPT fantasy league that never came to fruition (RIP), based on a patented (not actually patented) "player rating" formula I used for an old AGLA stats article series (which also might come back some day):
Runner: (3 * Goals) + (1.2 * BallKills) + (0.1 * KDA)
Tank: (0.4 * Multikills) + (0.5 * Sprees) + (1 * Exterminations) + (0.2 * Goal Offenses) + (0.2 * KDA)
Defender: (3 * Stops) + (0.5 * Goal Defenses) + (0.1 * KDA)
Hybrid: (1.2 * Goals) + (0.4 * BallKills) + (0.1 * Multikills) + (0.15 * KDA) + (0.2 * Sprees) + (1.2 * Stops) + (0.1 * Goal Offenses) + (0.1 * Goal Defenses)
The reason the numbers here might look odd decimal-wise is because I had specifically tweaked them so that the average points in each position league-wide was relatively even. I like the hybrid scoring system here better than in the older system posted above since it allows players who dominate in two areas of the game to not be overly penalized by the third (if you just take a straight average of the other three positions for hybrid scoring, you generally won't see very extreme values either way for hybrids, even if they had a monster or terrible game).
I used the stats from all the GPT Qualifiers at the time (probably somewhere in the range of 8 games/player on average) as a sort of control data to test the formulas, and these were the results:Runners:
Min avg/game: -0.6 points, Max avg/game: 13.5 points, Mean avg/game: 3.6 points, Median avg/game: 2.9 pointsTanks:
Min avg/game: -1.5 points, Max avg/game: 17.0 points, Mean avg/game: 3.6 points, Median avg/game: 2.8 pointsDefenders:
Min avg/game: -0.5 points, Max avg/game: 11.1 points, Mean avg/game: 3.7 points, Median avg/game: Median: 2.7 pointsHybrids
: Min avg/game: -0.7 points, Max avg/game: 11.8 points, Mean avg/game: 3.7 points, Median avg/game: 3.2 points
Generally with that system, tanks tended to be high-risk/high-reward type plays due to having the great impact from poor KDAs if a player has a bad game (6/32 players averaged <=0 points per game average, compared to an average of 1-3 such players in the other three positions, while 4/32 players averaged >=9 points per game, compared to an average of 1-2 such players in the other three positions). Defenders and Hybrids tended to be more consistent performers with a lower ceiling (only 1/32 players averaged >=9 points per game as a hybrid, but the median average of hybrids was considerably higher than the other three positions, while defenders had 3/32 players average the >=9 PPG mark but with a bit more variance). Runners tended to fall somewhere in the middle between tanks and defenders.
That system had been planned using seven-man rosters - one runner, two tanks, two defenders, and two hybrids. I think there's also a case that could be made for six-man rosters (1 runner, 1 defender, 2 tanks, 2 hybrids) since there aren't many purely-defensive players nowadays. Keep in mind, however, that this was for a 8-team, 4-player roster league, so the player pool was considerably smaller than the AGLA's would be, and you could certainly get away with larger rosters (or even a draft/auction bidding type thing, as has been done in past fantasy setups) in an AGLA variant, in my opinion.