Why are top ranked players always telling us how broken the system is? 

I can sum up the frustrations.  See, when don't you have a point loss cap, it makes it very hard to put a 100,000 point spread  between ranks and lock in your rank.  There is alot of posturing that skill or rewards aren't great enough, but that's basically smoke and mirrors. 

IF you regulate, or cap, the point loss....you build protection into the top ranks and allow them to put up astronomic point totals over time.  This has the secondary benefit to regular top players of widening ranking gaps by 30000,40000,50000,100000 points, and a tertiary effect of freezing the rankings altogether.  This was demonstrated in 2008 ranks as the year progressed there was almost no movement within the top 15 players and there was NO movement of new players into the top 15 (unless inactivity knocked out a 30 day no-show).

Graphically, a Capped point loss system it looks like this:

Conversely, if you DON'T regulate the point cap loss, and make it increase exponentially to thier point total...well then you get a diminishing returns curve that makes it very hard for the Top 20 or so players to have runaway point totals.  What I mean by this can be summed up by the following graph.  Lower point totals make for more contested rankings and difficulty for the top players to "freeze" their position by building up huge point buffers.

Paying attentinon to the x scale shows how no point loss caps effectively keep the max points lower.

In the end, I favor the NO CAP to point loss, because I feel like it makes for a more liquid ranking system.  Although it does favor the Skill tree more than the time played.  Any arguement that a point cap favors skill is wrong.  When you have NO point cap it forces good players to make the captures to get the big points...because they can't make thier living on killing/fragging alone.  It will not change the way the top 20 players play because they will quickly realize "playing it safe" is a quick way to knock yourself out of the top.  Take Kbuz for example...if he had not been leading the charge in Kd ratio and capping points..then he would not have gained ANY points...he would have suffered huge losses because if he just sat back and sniped..or camped out killing ...he would have been making only a few points off kills...and suffering huge losses when he did die. 

Sorry, this ended up being much longer than I intended it to be.

I recognize why the system in place is used, and it's even explained in the community HLstatsX help files, but the basis of my argument is not whether or not the stats are fair to the community and players in general, but whether point values gained and lost for kill, deaths, and objectives are truly reflective of Team Fortress 2's gameplay. I was under the impression that stats were an objective measure of a player's skill and not his relative rank to other players. I believe you posted those graphs in the last discussion on this topic, Giller, a few months to a year ago when a stats reset was also being discussed. Unfortunately, I don't believe your reasoning is quite as applicable to the current situation on the server.

With the old stats system, a minimum of 1000 kills was required to receive 'real' points for kills when I started playing on the Tempest servers about a month after the game was released. Along with that, there was only one 24/7 server running along with two rotation servers. At the time, the average person reaching 1000 kills had around 15,000 points under his belt and very few people had fewer than 10,000 points and 1000 kills. At that time, Tempests 1 and 3, the rotation servers, were the only true ways to gain points since Tempest 2 was typically full of players and thus not much flag capturing was being done. The introduction of additional 24/7 servers was the biggest blow to the objectivity of the stats. Because players in the typical 24/7 game seldom achieve the game's objective, they simply kill each other and gain relatively little 'skill' in the stats system. Keep in mind, however, that both flag and point captures were worth 50 points a piece back with the old system and later worth 25 additional points to each member of the team achieving the objective.

Under the original system, even with the addition of additional servers, nearly every player getting 'real' points for kills had 15,000 points or more. Now, however, 3500-5000 points is the typical starting point for 'real' points because of 24/7 servers. The game has changed. The worst part about this, though, is that point values for objectives have decreased dramatically, too. Point capturing is only worth 20 points (previously 25 under the new system, 50 with the old). The risk/reward in point terms makes going after these objectives foolish for someone who solely wants to rise in the standings. I doubt this applies to any Tempest regulars, but the general idea is still in everyone's minds. Even though the objective of the game should take priority, I'm certain quite a few people, myself included, hesitate to rush in and reset the cart timer knowing that they will get killed for -100 points by a sentry gun right next to it in a hopeful attempt to do what is best for the game in the game's eyes, not the stats's.

With the current stats, I have captured more flags and points than any other Tempest player. In the last 28 days, point and flag captures have made up over 45% of my points gained. I believe this is the highest percentage out of regular players for the time being. At the same time, though, I've also lost what I approximate to be 2400-2500 points due to deaths. That's 125 points captured--almost one third of my total points captured--only 12 of the Tempest's 52,660 players have 125 or more captures.

If I went onto the zombie server, for instance, I could potentially (and easily!) lose every single point I have in one death. You read that correctly. I could, in theory, lose over 600,000 points, but I don't have quite that many, but suffice to say that I would be losing points in the thousands frequently. There are a handful of players on the zombie server that, due to the nature of the mod, have the required 500 kills, but have points in the double and low triple-digits. Taking those extremes out of the picture, a death to someone with a paltry 3000 points will result in a minimum of 100 points lost on my account. Taking those extremes back into the picture, what if the player with 10 points joined a round on Tempest 6 and killed, say, the top 4 players during the round. Boom. The top 4 players on the server would be in dead last in the space of an instant. Is this fair?

Sure, this keeps things 'balanced' in some sort of warped-logical way, but I disagree that this is a good thing. I am typically a fairly aggressive player--I push for the objective. Actually doing what the game wants you to do is worth a measly 20 points in a land full of the shadows of -100s popping up around every corner.

What HLstatsX measures, with this skewed points system, is not what it calls 'skill,' but rather how many kills you can get before accomplishing or failing the objective. Dominations? 15 points, with a likely 3 more points for the kills needed to get it. That's 18 points right there, practically the same value as a point capture, but you, in all likelihood, did not help anyone reach that objective. I fail to see how a system that does not justly reward for completing a game's objectives and punishes extremely harshly for common occurrences can be called balanced. If the stats system is to be a measure of skill, it should go by the game's definition and reward more greatly for goals accomplished than it punishes for actually attempting to reach those goals.

First: there is a points loss cap. It's 500 points.

Second, I personally think the current point system is working better than the earlier ones we've tried. The problem I always had with the old system was that new players benefitted heavily by killing long-time players, ie. gaining 50 points for a random shot. HLStatsX was really broken, since it merged a time-based points system and treated it as though it were skill-based when it came to adding/subtracting points. IE it figured someone with 10,000 points was 10X better than someone with 1,000 points, when that simply wasn't the case. Now that doesn't happen - there's flexibility to the points, and a noob with 3,000 points may only get 7 points for killing someone with 65,000. I have seen the occasional player get over 20 points for killing me, so there is some math going on in the background. Now imagine if those gained caps were removed (which is the only other option at this point). Would you honestly feel better if a first-time scout got 500 points for killing you, than if you still lost 500 but he only got 7?

Right now, it's a rather interesting time-based points awarding system, merged with a skill-based system to KEEP those points. Essentially it's supposed to keep you mostly EVEN in points if you maintain a 2:1 kill/death ratio. Obviously this isn't completely the case, since I'm running under that ratio and still growing, but admittedly at a slow rate. Which means if you whore out the capture point maps, you're probably going to see more points lost per death than if you play 2 Fort. Which actually helps balance out the maps. Even playing medic isn't a way to gain uber points, since you end up with a pretty low K:D ratio. Which again, seems like a good balancing factor to prevent particular play styles from gaining massive points (and don't tell me that playing medic wasn't the key on the old system).

Finally, the version we have installed is not yet perfect; it doesn't award points for some events like it's supposed to. There are fixes, but I haven't been able to upgrade the system (no access).