Who wants to make calulating skill easier?

Now, I know what everyone is thinking, “Why, just use a kill/death ratio” because, kdr’s represent who that player has been since they started playing, take me for example, Halo 4 was my first halo, so I had no past experience, which is why I have a 0.84 K/D when most of the time I have a positive score in games, all the kill’s I got when I started still add up. So I was thinking: Why not make some way/measurement/calculation to figure out current skill, like maybe the measurement only uses your last 10 or so games. But I wanted some way to actually measure the players capabilities. not a clue how, but that’s where you guys come in. I could really use help, We could talk back in forth I waypoint, anyone want to help? Even if you don’t want to continuously help, even contributing just one idea would be appreciated!
thanks

So something like the A,B,C system? :smiley:

No really I don’t know what is considered an “F”/not good and something that is something that is great/“A”. Yeah 343 should put an explanation box with a K/D range.

> Now, I know what everyone is thinking, “Why, just use a kill/death ratio” because, kdr’s represent who that player has been since they started playing, take me for example, Halo 4 was my first halo, so I had no past experience, which is why <mark>I have a 0.84 K/D when most of the time I have a positive score in games</mark>, all the kill’s I got when I started still add up. So I was thinking: Why not make some way/measurement/calculation to figure out current skill, like maybe the measurement only uses your last 10 or so games. But I wanted some way to actually measure the players capabilities. not a clue how, but that’s where you guys come in. I could really use help, We could talk back in forth I waypoint, anyone want to help? Even if you don’t want to continuously help, even contributing just one idea would be appreciated!
> thanks

That doesn’t add up.

> > Now, I know what everyone is thinking, “Why, just use a kill/death ratio” because, kdr’s represent who that player has been since they started playing, take me for example, Halo 4 was my first halo, so I had no past experience, which is why <mark>I have a 0.84 K/D when most of the time I have a positive score in games</mark>, all the kill’s I got when I started still add up. So I was thinking: Why not make some way/measurement/calculation to figure out current skill, like maybe the measurement only uses your last 10 or so games. But I wanted some way to actually measure the players capabilities. not a clue how, but that’s where you guys come in. I could really use help, We could talk back in forth I waypoint, anyone want to help? Even if you don’t want to continuously help, even contributing just one idea would be appreciated!
> > thanks
>
> That doesn’t add up.

He means that he’s currently doing well as opposed to when he started playing. I was the same way with Halo Reach only I never got good until Reach’s final days. And I guess I was kinda the same way with Halo 4 since I only play on weekends and it was a new game, so it took me a few months to get my K/D out of the .8s. Now I’m stuck in the .9s, gradually inching myself to get above a 1.0 :stuck_out_tongue:

@OP, I think your last 50-100 games is a bit more appropriate being included in your K/D. That way, a larger span of matches are covered and keep your status fairly distanced over time, but current.

POSSIBLE PROGRESSION SYSTEM IDEAS FOR FUTURE HALO GAMES.

social.

1.each playlist has ranks that are 10% more xp each level to go up making progression endless but eventually very slow. eg captain 30wins captain grade 2 33wins. similar to h3 ranks. eg general grade 2 could keep going ( general grade 5,6,7 or other names inheritor, master chief grade 1,2 and keep going. each playlist should have its own xp tally, and a total tally for the service record.

ranked

  1. once a skill rank has been earned it cannot go down, but make people grind for that rank. eg skill 20 needs a 50% win ratio over 20 games to gain skill 21. then 52% win ratio over 25 games for skill 22. each level taking longer than the last with higher win ratios needed. if you fail you do not rank down but start your 25 games again and aim for 52%. the target should be visable to you. by the time someone has a 49 skill (eg 80% win ratio over 100games.) the number of games and win ratios needed would only be achievable to the best players that keep winning. and if you get disconnected, afks, lag, have a bad day etc and you fail to hit the mark you just try again.

you may not like this suggestion but going down ranks in my opinion is the worst thing that can happen in a game. i went down 1 rank in h3 with 20 kills 0 deaths and stopped playing for 3 months, i was so pissed off. playing allday and trying your best only to go backwards does not keep people playing. this will help prevent people giving up making playlist populations higher but still keep lower skilled players from becoming higher than their actual trueskill.

these are just my thoughts on the subject that i think will benefit all parts of the community. if a skill 50 took 2000 games to earn for even the best players that rarely lose games, selling accounts would not be worth it as it would take to long.

  1. K/D doesn’t represent much in a team-based game and has little to do with the win.

It doesn’t matter if you win my an inch or by a mile, winning is winning and K/D doesn’t represent who won as a team.

  1. Players should be “ranked” on both team and individual merits.

The main slayer on a team that always wins will always have a better K/D than someone who supports the main slayer on that same team via strategic moves that create distraction or objective points.

If we look to the 2 systems we’ve had, the 1-50 scale can be used for team-based CSR while the mineral/metal can be used for the individual-based CSR.

From there we can see who plays with the best and/or wins consistently in spite of K/D while the secondary/personal rank can track those personal stats and relay them as well.

An Onyx-50 would be a main-slay always winner, a Bronze-50 is someone who always holds 4th (or almost always) but is also almost always winning (and against other 50’s not just others).
An Onyx-40 would be someone who consistently holds 1st or 2nd place on the team but they tend to run with randoms and/or their team just doesn’t win those important games against higher ranks.

Either way there is a way to recognise the team effort that most Halo gametypes are about and to recognise personal efforts without removing either.

> > Now, I know what everyone is thinking, “Why, just use a kill/death ratio” because, kdr’s represent who that player has been since they started playing, take me for example, Halo 4 was my first halo, so I had no past experience, which is why <mark>I have a 0.84 K/D when most of the time I have a positive score in games</mark>, all the kill’s I got when I started still add up. So I was thinking: Why not make some way/measurement/calculation to figure out current skill, like maybe the measurement only uses your last 10 or so games. But I wanted some way to actually measure the players capabilities. not a clue how, but that’s where you guys come in. I could really use help, We could talk back in forth I waypoint, anyone want to help? Even if you don’t want to continuously help, even contributing just one idea would be appreciated!
> > thanks
>
> That doesn’t add up.

Your kidding me right. He already said when he began that he was not good. This annoys the heck out of me because same thing happen to me in Reach. Now that i went back to reach every stupid and loser inheritor that i beat just says your kd sucks then leaves even when in game i could have gone like 20 kills and 5 deaths. K/D is a stupid way of seeing how good a player is and just because he is good now but still has a bad k/d doesnt mean that it doesnt add up.

> Now, I know what everyone is thinking, “Why, just use a kill/death ratio” because, kdr’s represent who that player has been since they started playing, take me for example, Halo 4 was my first halo, so I had no past experience, which is why I have a 0.84 K/D when most of the time I have a positive score in games, all the kill’s I got when I started still add up. So I was thinking: Why not make some way/measurement/calculation to figure out current skill, like maybe the measurement only uses your last 10 or so games. But I wanted some way to actually measure the players capabilities. not a clue how, but that’s where you guys come in. I could really use help, We could talk back in forth I waypoint, anyone want to help? Even if you don’t want to continuously help, even contributing just one idea would be appreciated!
> thanks

Well here’s a little problem with trying to evaluate skill in a video game, its a complex behavioral trait that we have a hard enough time defining. Quantifying it in any reasonable way (ie. one that could actually stand up to some test and be taken as an approximation of the “real” value of skill) is practically impossible without tremendous advances in the field of psychology.

I’ll use an example here. Say I’m facing you in a game of CTF. How well we do is context dependent, ie. my performance is going to be strongly affected by who else is in the game, what the settings are, and what the map is. OK, for the sake of argument say you factor all those elements into some skill-based rank (I say for the sake of argument because right here you have a practically impossible situation.) How valid is that number? In a standard match we only played for 10 to 20 minutes. The per-match “sample size” is, in all likelihood, woefully insufficient to factor out all the random/chaotic interactions (ie. everything besides differences in ability) that may be influencing the outcome of the game. Then the question becomes “how many games is sufficient?” And mind you, those additional “trials” would need to use the SAME parameters (ie. same map, gametype, and players) in order to get at how any action in that game relates to player skill. That leaves us with the depressing end to this attempt to improve skill based ranks of having to stretch a single match to multi-hour ordeal just to pull meaningful data.

K/D, W/L, it really doesn’t matter simply because ALL the measures yet devised are irredeemably crap. The ideal situation is to simply cut it all out and let players judge each other’s worth without a computer feeding them bad information (or at the very least a twisted premise) using the fantastic human capacity to make such social evaluations with very little data. You might then provide individual statistics (ex. flags captures, kills, assists, ect.) as “data points” for those ad-hoc judgments, but you should never present them as an evaluation in and of themselves.

> K/D, W/L, it really doesn’t matter simply because ALL the measures yet devised are irredeemably crap. The ideal situation is to simply cut it all out and let players judge each other’s worth without a computer feeding them bad information (or at the very least a twisted premise) using the fantastic human capacity to make such social evaluations with very little data. You might then provide individual statistics (ex. flags captures, kills, assists, ect.) as “data points” for those ad-hoc judgments, but you should never present them as an evaluation in and of themselves.

Mind you, subjective human based evaluation is unreliable. Mathematics, on the other hand, is consistent and powerful. When you disregard all the details about optimal tactical choices, coordination, and other complex player actions and only concentrate on the end result, you actually find that skill is a very simple concept, from a mathematical perspective.

Relative skill is your probability of winning. It’s a fairly intuitive idea when you give it some consideration. It’s obvious that if the skill difference between players is large, the better player will win almost always. If the difference is only minor, small fluctuations take over and the probability of winning approaches 50/50. Then, of course, you don’t assume that if a player has 75% chance of winning the other player that they would win precisely 75 matches out of a 100, you obviously have some kind of a standard distribution.

And that’s very much how it works in most games where an accurate system for determining player skills is needed. It’s by far the most accurate way of determining player skills because the amount of trials (hundreds or even thousands per player) assures that all anomalies (such as player getting carried by their team) are smoothed out.

> And that’s very much how it works in most games where an accurate system for determining player skills is needed. It’s by far the most accurate way of determining player skills because the amount of trials (hundreds or even thousands per player) assures that all anomalies (such as player getting carried by their team) are smoothed out.

I do not believe this is saying what you want. To clarify;

We don’t know a player’s skillsets by their win/loses. It is very true that as long as we concentrate only on W/L, all other stats don’t matter, but to determine a player’s skillset and level of those skillsets via W/L is unattainable.
A player can be lousy with the sticks but great at anticipation and so make up for their lacking abilities in an arena where getting to a powerweapon first can make or break a game. And through this ability to be first when need be, show domination over others by the strategic elements of the game.
Skilled at Halo??? Well skilled enough to win consistently in a “real” match but if that player ever had to do a “skills” competition, they could likely not be top 3 or even 5 in any one “skill.”

It is why I state it is unfair to judge a player’s K/D as a measure of their “skill.” 2 top level players are going to keep each other near 1.0 if all they fought was each other. But put those 2 against thousands, even millions of others, and they’ll prove themselves amongst the best, as though without equal (but each other).

It’s why it doesn’t matter if we win by an inch or by a mile. The skill of the game doesn’t matter, in the end it’s truly the skill of ourselves versus others. It’s the nature of tea.

>

Obviously, no capabilities at specific skill sets can be figured out. That’s a much more elaborate system. It’s only the combined end result of all those different skills that you can dig out. I was saying “player skills” not in the context of different skill sets, but as in “the skill levels of all players”.

And yes, K/D is a bad tool for judging player skill because it doesn’t have any absolute values. It’s more of a display of the quality of the matchmaking system rather than the individual’s level of skill.

K/D doesn’t work all the time. I just played a match on Halo 3 where my entire team quit and it was me vs. 5 other players. I didn’t want to be punished for quitting. So, obviously, my K/D spread was pretty bad.