This topic contains 96 replies, has 0 voices, and was last updated by  BOY 9 years, 1 month ago.

  • Author
    Posts
  • #53374

    Raden
    Participant

    Hi BigBear,
    Yes, that is correct. A pure ratio system will be the total opposite of a pure time-spent based system. By having both in the calculation it more or less takes the middle path. But i know I am slightly biased towards an accuracy orientated system. That was the whole idea in the first place, to make a new skill rating biased towards accuracy but with time-spent based parameters also in the picture.

    I did not take into account Game n Round wins because it is also partially time-spent based too. And as PeanutsRevenge suggested, if a Round Wins / Rounds Played ratio were included then I guess it would be fine.

    Referring to the equation, you will notice that each parameter is multiplied by a certain number, this is a way I factorized each parameter based on their significance to achieve a more correct interpretation of skill. So if you say that I’m basing it too heavily on Weapons Used, then the reduction of the multiplier for Weapons Used will reduce its effect on the equation overall. Maybe instead of multiplied by 10, it can be multiplied by 6 or 7, but not lesser than that because it will render this parameter insignificant if we do so.

    As for resigns, my equation will discourage resigning although not much because its multiplier is only valued at 5. This has nothing to do with skill determination actually. To me resigning just denies another player a kill. Just like in Poker, if you close your cards and don’t proceed with the bet, then you stand to lose your ante. So in Scorched’s, a player should be subjected to some sort of penalty for resigning. Lets just say a player makes 3 kills in a round and then resigns, he has gained 3 kills, but he has denied his surviving opponents the chance for an extra kill. Again, if you think it is unfair to penalize the resigner too much, then you can always reduce the multiplier value to lets say a value of 3.

    And finally, as for Suicide kills, ok, I agree. You cannot impose a penalty for suicide by random funky deaths because the player is punished enough by the subtraction of 1 kill. However, I must point out that new players do tend to kill themselves more regularly due to inexperience (aka lack of skill). Experienced players who do kill themselves accidentally by forgetting to increase power after charging up batteries or by shooting too low and not checking their shot clearance show their lapse of skill, not their lack of it. I will amend my equation so as to not include Suicide Kills in it.

    Thanks for the positive feedback BigBear.

    #53375

    BOY
    Participant

    If kill ratio (kills per shot) is used heavily to determine rank, players are in fact penalized each time they shoot. The smart player will not risk a hard shot with a low chance of success. The stats should never penalize a player just for shooting, only for MISSING. hmmmmmm.
    #1. (KILLS/SHOTS)

    Survival is important, this is obvious, but if round wins are too heavy in deciding the winner, the smart player doesn’t even shoot, he dirts or motors away to get the win whenever he can. Also it is clearly obvious that many rounds are won completely by accident. Therefore round wins should not be a factor in skill. What is important is survival by elimination of the oponent. The attribute of a skilled player is one that knows who to shoot and when and how and has a high amount of shots per round. (such as maybe 12 :P)
    #2. (SHOTS/ROUND)

    Note that #1 penalizes players for shooting and #2 rewards players for shooting, neither of which really makes any sense because it is kills that matter. Simply multiply them and the result is KILLS PER ROUND.

    *SHOCK*

    Examine the stats of the past. You will find that the best players have a high kill per round ratio. Many also have a high kill per shot ratio as well, and that is a good stat as well. But the persistent player, and the winning player does not always have the highest kill ratio because he shoots more, he lasts longer and consistently takes longer and harder shots as the round goes on.

    each new round each player takes a risk and each round is like its own minigame. Shouldn’t kills per game be the standard for Skill? Uhmmm. YES!

    You might think kill ratio is the key stat, but just remember a long surviving player with a good Kill/Shot ratio will also have….. what? A HIGH K/Rnd ratio! 😀

    In order of importance, skill should be weighted by the following

    (Kills + Game Wins*10 or even 20) divided by rounds

    I’ve proposed it before and I will again. Take the skill score and make an modified skill score using the option above or something close. It would make more sense if the server wins were determined ONLY by kills (which would be the right way) but I can only hope for so much 😉

    #53376

    Thrax
    Participant

    You wanted details on exactly what went into the current skill calc?

    Here is the code used.

    If you can follow it, it updates both the killer and the victims skill each time,
    accounting for the weapon used.

    	// Update both players skill points
    std::list skillRows =
    runSelectQuery("SELECT a.skill, b.skill FROM "
    "scorched3d_stats as a, scorched3d_stats as b "
    "WHERE a.playerid = %i AND b.playerid = %i AND a.prefixid = %i "
    "AND a.seriesid = %i AND b.prefixid = %i AND b.seriesid = %i;",
    playerId_[firedTank->getUniqueId()],
    playerId_[deadTank->getUniqueId()],
    prefixid_,
    seriesid_,
    prefixid_,
    seriesid_);
    if (!skillRows.empty())
    {
    std::list::iterator itor;
    for (itor = skillRows.begin();
    itor != skillRows.end();
    itor++)
    {
    StatsLoggerDatabase::RowResult &rowResult = (*itor);

    int firedSkill = atoi(rowResult.columns[0].c_str());
    int deadSkill = atoi(rowResult.columns[1].c_str());

    float weaponMult = (float(weapon->getArmsLevel()) / 10.0f) + 1.0f;

    int skillDiff =
    int((20.0f * weaponMult) / (1.0f + powf(10.0f, (float(firedSkill - deadSkill) / 1000.0f))));

    runQuery("UPDATE scorched3d_stats SET skill=skill+%i "
    "WHERE playerid = %i AND prefixid = %i AND seriesid = %i;",
    skillDiff,
    playerId_[firedTank->getUniqueId()],
    prefixid_,
    seriesid_);

    runQuery("UPDATE scorched3d_stats SET skill=skill-%i "
    "WHERE playerid = %i AND prefixid = %i AND seriesid = %i;",
    skillDiff,
    playerId_[deadTank->getUniqueId()],
    prefixid_,
    seriesid_);
    }
    }

    Feel free to suggest better ways to work it.
    Of course, if you can’t decypher it.. We’ll have to find another way to explain.

    #53377

    BigBear
    Participant

    @ BOY
    Using (kills + wins * weightFactor)/rnds would indeed give a better picture of who is a good player than the current ranking system IMO. However, I have a feeling that many would feel robbed if the top ten spots at the end of a series were to be occupied by 10 accounts with 1 “perfect” game on record. That would be quite possible with a pure ratio system. As much as I like the sheer simplicity of the formula and the way it relates to what I would consider to be skill, I just don’t see how the above mentioned pitfall could be avoided.

    #53378

    Raden
    Participant

    @boy wrote:

    If kill ratio (kills per shot) is used heavily to determine rank, players are in fact penalized each time they shoot. The smart player will not risk a hard shot with a low chance of success. The stats should never penalize a player just for shooting, only for MISSING. hmmmmmm.
    #1. (KILLS/SHOTS)

    But BOY,…..it is not Kills/Shot that is being said, but Shots/Kill. And by taking more shots to make a kill, wouldn’t it mean they ARE being penalized for missing? 🙂 Anyway we can always increase its multiplier to make it less significant overall because it is part of the denominator in the equation I posted.

    And your reason for not including Round Wins is the same reason I had for not including it in the calculation of skill in the first place. But what if it’s a ratio of Round Wins/Rounds Played?

    Thrax…. 😳 I don’t understand codes 😛

    #53379

    Raden
    Participant

    Friends,
    Prior to submitting my previous reply to BOY and Thrax, I had the whole night to think this over and also after a reasonably lengthy discussion (10 mins 😆 ) with someone close to me, I have come to the decision to withdraw myself from discussing this topic any further. It seems very inappropriate of me to ask for a change in the current system. I will leave it to you guys to continue this discussion and hopefully something positive and something accepted by the majority will be the final outcome.
    I apologize to the game developers and to ALL players if I ever seemed disrespectful or ungrateful in my words and intentions.

    #53380

    naka
    Participant

    @boy wrote:

    If kill ratio (kills per shot) is used heavily to determine rank, players are in fact penalized each time they shoot. The smart player will not risk a hard shot with a low chance of success. The stats should never penalize a player just for shooting, only for MISSING. hmmmmmm.
    #1. (KILLS/SHOTS)

    Absolutely, to win the game you must shoot, regardless of how bad is the position and how despair the shoot is. Unfortunately shooting to win the game will decrease your kill ratio.

    I think, the kill ratio is the best indication so far of the players skill, but it is not so important.

    #53381

    PeanutsRevenge
    Participant

    BOY.. The problem with going with rounds, is in the later rounds people tend to have shields.

    If a player has joined late in the game and has only missiles it will take two perfect shots minimum to get the kill, whereas another player with some financial backing has some baby shogs, it would only require one ‘pretty good’ shot to get the kill.

    However, there’s always going to be people that’ll lose out, if this group REALLY care about rank, they could easily not join too early.

    Raden. You have been the major contributor to this thread and thus are ‘NOT ALLOWED’ to withdraw :@

    I was thinking that people can come up with their ‘best’ formulas (there will be no ‘perfect one size fits all’ method.
    Then when there are several different suggestions, a poll could be made for the community to vote for which they’d prefer which can then be viewed by all active admins to ‘hopefully’ be implimented by a developer.

    #53382

    Hyde911
    Participant

    My idea is to calculate rank it this way:
    rank= kill ratio * number of kills * arms level

    #53383

    BOY
    Participant

    @peanutsrevenge wrote:

    BOY.. The problem with going with rounds, is in the later rounds people tend to have shields.

    If a player has joined late in the game and has only missiles it will take two perfect shots minimum to get the kill, whereas another player with some financial backing has some baby shogs, it would only require one ‘pretty good’ shot to get the kill.

    However, there’s always going to be people that’ll lose out, if this group REALLY care about rank, they could easily not join too early.

    PR, your comments so far have been the most sensible…. but… 😛

    there really is no problem with rounds at all.

    Okay, when one wants to get the game win, yes, you must try to join the game early if that is what you mean. But there is very little reason to assume that joining late will hold down a players skill rank when using kill/round. There area always oportunities to get the kills. About the only time you cannot is if you join a game of just a few very good players who all have money. In this case, the skill of the other players works in your favor because the kills you do get are worth more in the current skill rank.

    when the game begins it is very often that two players begin, and other join later. This is simply the nature of the game. The kill ratio of a good player who joins late will be poor, plinking away, but I would bet that same good player will maintain a high kill/round ratio. Even though he can’t win in the final shots due to being poor, he can still show his skill. Don’t penalize a player who shoots it out to the end.

    kill/round !!!! players are not penalized for shooting, and not penalized for failing to win a round due to being poor, or anything like that. It is simple. You join, you shoot, you get kills, the more efficient you are at kills, the higher your skill.

    The financial backing….. THe free market drains the players with the most money pretty fast. dont forget that shogs and rollers and all weapon and everything baught is simply leveled out by the economy (assuming it is right).

    My idea is to calculate rank it this way:
    rank= kill ratio * number of kills * arms level

    so what hyde is saying is (KILLS-squared) * (avg arms lev.) / (SHOTS). As I’ve said above, a player who joins late, must take harder shots in many cases and players shouldn’t be penalized for trying a shot, ANY shot. Only penalized for no kills. Which brings me back to kills/round 😀 The best overall indicator of skill.

    Just try to achieve more than 2 kills per round after 50 kills. You will fail unless you’re cheating somehow.

    #53384

    PeanutsRevenge
    Participant

    Okay, I have gone with game wins as my main decider.
    I know this punishes those that join a game late or leave early, but there is no way of making a ‘completely fair’ calculation.
    This keeps it simple and I would say that when it comes to winning, it’s game wins that count the most.
    This does allow people to avoid playing skilled players and try to only play newbs, but their just sad anyway, so should that happen, lets just point and laugh 😀

    This takes away arms level complexities (does it really matter if someone uses funkies? if it wins them the game because somehow they are able to afford them, so what).

    Hence my formula is

    ((GW / RP) / SpK) * 10000
    (GW = Game wins, RP = Rounds Played & SpK = Shots per Kill)

    This gives the following examples from the previous stats series, which I think is reasonably accurate for actual players ability (altho Raden is alot lower than I would have put him).

    PLAYER – – – – – SKILL
    VIHOR


    362.65
    XTC – – – – – – – – 251.89
    VITOS


    188.69
    OUTER – – – – – 174.28
    NAKA


    136.17
    HWB – – – – – – – 120.17
    PEANUT


    117.76
    BIG BEAR – – – 113.06
    RADEN


    91.49
    SHELL – – – – – – – 87.59
    POM


    65.67
    TNICK – – – – – – – – 65.49
    ROMMEL


    36.53

    I also though about maybe having these formulas (whoevers got picked) only kick in after 100 kills for example, until that point the ranking continues as is. This would encourage new players to continue playing as they see their rank increasing, but also weed out the possability of people only playing a couple of games and getting a new I.D as they’d had two brilliant games.

    #53385

    BOY
    Participant

    So rearanged it is ….. KILLS * GW / SHOTS / RNDS.

    or GW/RND * KILL RATIO 😀 Good theory, I do like it. Although as I said, shots are not a good way to account for skills. They aren’t useless, but anyway, it has always been an accepted benchmark of a good player so I could live with it. I just don’t want players who don’t shoot on purpose to preserve thier rating, THAT is the drawback.

    Also the skills in your chart are not proportionally right. Vihor is clearly the best, but… is he THAT much better than Rommel or tnick? I doubt it. Try running some tables with Kills added to GW instead of multiplied maybe. It doesn’t compound the differences in stats.

    #53386

    PeanutsRevenge
    Participant

    I’m trying to keep kills out of the equation..
    Unless I tallied em to rounds played maybe.

    If its too much about kills, Vihor wouldnt have had a look in due to the complete lack of play in that series.

    #53387

    PeanutsRevenge
    Participant

    Okay, upon further analysis of my theory, I have deduced it was slightly flawed.

    I have now brought in kills per round and also taken game wins per game (worked out from rounds played /10).
    I tried having a factor of 5 added to the number of game wins per game.

    I have included the spreadsheet with my working, it’s rather messy as there are the 3 different ways I was going, but this way it shows a few different stats which I cba to paste here!

    #53388

    HWB
    Participant

    Has anybody thought if we don’t have it as amount of kill as the rank then maybe the servers get dry. I am thinking that people say they don’t play for rank but a lot players can’t keep minds off the top 10.

    I mean, there are many players better than me (cough, i don’t use anything but sight = me a noob)…oh as i was saying….many player are better than me but I play against them now for the kills I might get when I have my eye on the prize. Going with any other skill rating, I would rather stay out of matches which I won’t get less that 10 kills a game or say 2 kills in a 1 vs. 1 mcb game.

    Does this make sense. You may call me a kill whore/but not a stats whore. 😆

    But I have liked the topic already. I just don’t think it should change. We make better players when the want a high rank with the skills. It takes alot of kills to be in a top position and practice can make perfect.

Viewing 15 posts - 46 through 60 (of 98 total)

You must be logged in to reply to this topic.