PSR mechanic critique

[Kai_]

I observed a situation yesterday where there were 8 players roughly around the 1620 level, one 1450 player, and one 1800 player.

When the 1800 player was with the 1450 player, the match was about 60:40 in their favour. The 1800 player's gain/loss was +0/-0

When the 1800 player was against the 1450 player, the match was ~90:~10 in the 1800's favour. The 1800 player gain/loss was +5/-2

Sorry what?
Algorithm-win

[Jotain]

Kal, try to calculate averge psr for those games. I got the feeling that other game had higher psr people playing.

[[tiME]xHomicide]

Example: assume we have 10 players who have all played 100 games, not specifically with each other. Assume 1 of those players had 1 completely erroneous game, and that there is no other errors currently in the system. I will say that a game, in which the player in question was the MVP of the entire game(both teams) and was on the losing team, is a 100% erroneous game. This would both optimally confuse %wins and PSR.

Assume that those 10 players play a game together, and the game added no additional error into the system. %error for %wins decreases overall since the same amount of error is averaged over a larger game set. i.e. (100*9+99)/100 < (101*9+100)/101. However, since the player with 1% of error accounts for 1/10th of the calculation for each persons PSR change, 0.1% of error gets added to each persons PSR change. thus, even though this new game had no error in it, the PSR scores of this game added extra error into the system anyways. i.e. after that game 9 players have a very slight amount of error in their PSR(meanwhile 0%error in their %win), and 1 player has roughly 1% of error in their PSR(and likewise 1%error in their %win).

This merely showed how 1 flawed game out of 1000 impacted the next un-flawed game. Now imagine a system with a constant amount of error. e.g. for every 10 games played there is 1 games worth of error (a.k.a. 10%error). %error in %wins will remain at a constant 10%error, while %error in PSR will grow exponentially, so even if you assume %error for PSR starts below 10% because the PSR numbers start-out accurate, it will eventually grow larger than %wins %error. That also assumes PSR numbers start-out accurate, which they clearly don't since everyone starts at an arbitrary 1500.

Wrong.

The error does not remain the same in the 1 player. His error actually DECREASES. He will be rewarded MORE psr than he otherwise would have received.

Furthermore, this entire argument can be disregarded by simply acknowledging the fact that a player's ability changes over time. Thus, in order to properly calculate error in a system like this you have to make the assumption that the older a game is the more error it contains. You have to take this into account for EVERY SINGLE game. Its also fairly safe to assume the error% of older games grows exponentially.

A game that is 1000 games old tells significantly less about a player's current skill than the last game he played.

Third, 1500 is in fact NOT average. The only way 1500 could remain average is if the total PSR remained constant, changing only with the creation and deletion of accounts. I believe there is both a mathematical proof and historical events out there that prove total PSR is not constant for me. Since the total PSR in the system can grow, and PSR is supposed to represent a unit of skill, when the total PSR grows it would suggest the HoN community as a whole got more skilled, which makes a whole lot of sense, BUT it means that the PSR value that represents average is constantly changing with the fluctuations in total PSR.

Having a system in which the average skill level is a constant number has advantages over a system that doesn't when viewing it from a singular person's perspective. If I have 1632 PSR, am I good or not? I don't know unless someone tells me what average is, and no one has, so PSR is rather pointless other than comparing to other people you meet on a micro scale.

Also note that, since S2 proclaims 1500 to be average and starts new accounts at 1500 even though it's not, if PSR has grown over time, which it has, then accounts created now are started at a lower skill value than accounts created in the past. This again makes sense on a global aggregate scale, but punishes new individuals on a micro scale.

The average is not computationally significant. Knowing the average has no purpose beyond stroking a player's egos.

I'm all for ego stroking...S2 should provide this statistic. S2 should calculate how many players have lower PSR than each player for maximum ego-strokage. "I'm in the 98 percentile."

Forth, Ludo-logically speaking, a scale that has essentially a 500 point radius, and starts at 1500, isn't very positively re-enforcing. It implies that you can't even come close to doubling your skill level, which is not encouraging, which in turn is against a major tenet of good game design.

This doesn't even make sense. Even a 5 year old child can comprehend a non-linear relationship. Everyone that plays hon understands that someone with 2000psr will DESTROY someone with 1500psr, even if the "number" is only 30% larger.

[Ryuukun]

Isn't PSR actually very simple?
Like you win, you get PSR, you lose, you lose PSR?
I don't see the big math there.

[Emin3m]

chane psr by tsr ? i think tsr can calculate a player skill better after 100 game played. or maybe just implent the tsr balance ?

psr doesnt mean anything at a certain point. uper 1700 it does have a signification but below 1600, its doesnt mean anything cuz some1 with 1100 could be better than the other with 1500 who only pub with his friends and have no skill on his own xcept the team work.

my idea implent tsr balance instead of a mods....

[[S2]Vahn]

Keep the discussion on topic, suggestions are not on topic.

[saucebrune]

And if you want to use any other stats than raw wins, those can always be manipulated... see TSR and wards placed where so many people who'd have had 0.1 wards placed per game will now have 10 wards per game for their rest of their life due to ward of revelation abuse/spam at the end of games.

Ward of rev dont count...

TSR tells as much about someones skills as PSR. Play support heroes -> bad k/d ratio -> bad TSR -> you get a codex scout telling how bad your TSR is.

lol you dont know anything ...support get awsome TSR since they get a lot of assist and wards...you can ez be 5.5 TSR while playing only support

Sku1l has 5.4 TSR with a 0.8 KDR but has 2.6 Wards per game

[CobaltBlue]

Wrong.

The error does not remain the same in the 1 player. His error actually DECREASES. He will be rewarded MORE psr than he otherwise would have received.

Take a look at my explanation of %error. You are arguing difference in PSR here, which is not what I was referring to with %error. Also, you assume his team won? If his team lost he loses extra PSR because of the error.

A game that is 1000 games old tells significantly less about a player's current skill than the last game he played.

This is a good point, but not about this conversation, since neither PSR nor %win use time as a relevant factor. I think if someone were to revamp the PSR system though it would be ideal to weight newer games at a higher value. Excellent observation.

This doesn't even make sense. Even a 5 year old child can comprehend a non-linear relationship.

Can a 5 year old tell me whether a 1500 and a 1560 are supposed to be better or worse than a 1520 and a 1530? no. You can't either. If it is a made-up stat for public consumption like PSR, it should be linear since that is the most obvious and straight-forward way a stat could work.

[Phopis]

@CobaltBlue

I agree that PSR is very flawed, and I never thought about the problems of a value that uses itself in its own calculation.

I’d also like to say that statistical math generally seems counter-intuitive to me, but I don’t see how the %error in PSR can grow exponentially when the game is weighted using PSR. Instead I would say that because PSR gets calculated using PSR, it causes erroneous games to propagate. Playing in one erroneous game increases the chance that the next game you play will be erroneous, but in the opposite way.

For example, if I play a game and play really well and get lucky (the opposing team has lag or something), my PSR becomes inflated or my displayed PSR is higher than my “true” PSR. In the next game I play, my team’s % chance to win will be overrated. This will increase the chance that I lose the game and therefore PSR.

I would say the problem with PSR is that it promotes a somewhat chaotic change in itself. This is balanced by the fact that the further you get from you “true” PSR, the more “pressure” there is pushing you back toward it. An acceptable analogy would be if everyone has an elastic band attached to them and their “true” PSR, and they’re all running around pushing, and bumping into each other.

[Trolldish]

What I don't understand is why must there be a scaling factor for PSR change for higher rated players vs. lower rated players? For example, a 1900 rated player could play against all 1600-1700 rated players and have a +0/-0 on the line, sometimes even despite an unbalanced game. On the other hand, his teammate rated at 1600 would be at something like +2/-10 for the game, which seems very imbalanced. The 1900 rated player has absolutely nothing to gain or lose, and therefore doesn't have much incentive to try as hard as the lower rated player, even though IMO higher rated players should play with more responsibility and show for their rating. I've felt no incentive to try as hard before when I had nothing to gain or lose, so I can definitely see how other players could feel this way.

In my opinion, the rating change per game shouldn't scale at all this way, it should instead scale based on the number of games played for each player. A player who just started out should have more to gain or lose, while a player who has played a ton of games should have less to gain or lose (to a certain extent, say +3/-3 at least).

[[tiME]xHomicide]

Take a look at my explanation of %error. You are arguing difference in PSR here, which is not what I was referring to with %error. Also, you assume his team won? If his team lost he loses extra PSR because of the error.

Wrong, again. On both accounts.

His team would have a lower average PSR and therefore his team's predicted win% would be lower. Thus, he would be rewarded more points for a win than a loss.

This is a good point, but not about this conversation, since neither PSR nor %win use time as a relevant factor. I think if someone were to revamp the PSR system though it would be ideal to weight newer games at a higher value. Excellent observation.

Consider 2 players: Player A and Player B.

Player A has won 50/100 games resulting in a PSR of 1600.

Player B has won 500/1000 games resulting in a PSR of 1600.

Now, consider if both players played on the same team and won 1 more game. They will both receive the same PSR gain, however, Player A's win% will increase to 51% while Player B's win% will only increase to 50.1%.

PSR gain/loss is independent of the number of games played.

Can a 5 year old tell me whether a 1500 and a 1560 are supposed to be better or worse than a 1520 and a 1530? no. You can't either. If it is a made-up stat for public consumption like PSR, it should be linear since that is the most obvious and straight-forward way a stat could work.

The relationship between PSR and skill is linear. At least in concept.

The only thing about PSR that is non-linear is the rate at which PSR is gained/loss. Which makes sense, because the rate at which players improve is non-linear.

[[UmUb]CSDragon]

The average is not computationally significant. Knowing the average has no purpose beyond stroking a player's egos.

I'm all for ego stroking...S2 should provide this statistic. S2 should calculate how many players have lower PSR than each player for maximum ego-strokage. "I'm in the 98 percentile."

Isn't that already in the game with the ladder? Just take your position number and divide by the total number of players. You would read it as being in the "top x percent"

As of 1 AM (PST) Nov 21 2010, there are 230,183 active accounts total.

And the worst psr in the game is -5.

[MrWinston]

TSR balancing > PSR Balancing. THen again this being said if you PSR balance most games will end up with roughly 50% tsr balance anyway.

[[scrm]Blet`]

tsr balancing is utter crap. I have 1.8 kd, im still way better than some 1400 scrub with 40% win and 100% scout with 3.0 kd. PSR IS NOT BROKEN. There are flaws. But the better you play, the more you win, the more psr you get.

Also : YES dewards count to wards placed. Ive tried it on my new account.

[Sku1l]

@Scopae

Your TSR is 6.43 and that is pretty good .

In your example of a 1400 scout with 3kdr , you forget to say that he probably have probably 0 ward placed and 0.5 D:A ; witch result to an approximately 4.5 TSR.

I'm still against the TSR balance because

1. Its easier to get TSR with Hard carry and Gankers
2. you can get a great TSR out of Unfair games
3. You can get TSR increase in games that you lose .
4. It can be cheated (buy wards of reveal at end of game , or any stat padding behaviour)
5. Its not always representative of how well you played