PvP rating inflation

Unknown

Not sure there's any mileage in this, but it's been kicking around my head a few days and abuelo reminded me about it, so thought I'd see if it's worth discussing.

Initial implementation of shields was pretty flaky. Even Demiurge's so-called "PvE" is 5+ day tournaments requiring adjustment of one's sleep schedule to place, so clearly they're into the competition thing. They didn't adequately anticipate or prevent people identifying that shielded opponents inject big points, and using them bagmanimously (thank you so much to whoever coined that) to collaborate for relatively easy runs at ratings of 2400 and beyond.

We quite like some rating inflation, as a rule. Free stuff is rarely bad. Demiurge isn't so enthused, and if I were selling said free stuff, neither would I be. But what, if anything, can we do to push average scores, and by extension hopefully our own peak scores, upwards?

Shields still help. Not nearly as much as they used to now you're all but removed from the game while they're up, but I've never thrown up a 3hr after a quick push without deflecting at least 150-200 in queued retaliations. Those are free points in the pool. Doing this early if you're not attempting to climb early and turtle out the rest amounts to a charitable donation of minimum 100HP to everyone in your MMR range, which I don't see flying for many.

Rating drop from losses and failed attacks (?) drops off at low rating. This is definitely true below 100 (where I've been hit for -2, surely less than the attacker was gaining), possibly higher. Does anyone know exactly where it cuts off? This means attacking someone below 100 earlier than you're much worrying about your rating is advantageous - they'll lose fewer points than you gain and, yes, will probably hit you hard in return, but you can recover that from them or someone else later. Alternating playing and tanking to try to get yourself hit below the threshold also injects points; not sure if some already do this or not. Will also probably reduce your MMR so you can faceroll some less competitive opponents along the way and feel big and healthy. Puny rosters. Least sure of this, but I think you could also deliberately farm point-injecting losses by attacking people above the threshold while you're below it and losing.

Don't know if anyone could actually be bothered with any of this, or if there's more obvious ways I'm missing.

kensterr

The way PVE is handled is as though everyone is in the same time zone.

Eddiemon

It is possible, but I think it requires too much co-operation.

At zero points each, you attack another player and there is a net gain of points. He loses none and you gain 25. At 100 points each he loses some for your gain, but it is still net positive. This continues up to 400? 700? Points. I think they moved the number and I have never studied it that hard.

So what you would need is a team of feeders below that threshold who only attack each other. They can keep building their points up to that number and effectively inject points into the eco-system. You would have another tier of players who only attacked those players from above. That tier would never fight each other, only feed on the lower tier who are generating new points.

You might then need to build that into a pyramid of tiers, where you have the group generating points, a tier above them who are feeding only on them, a tier above and so on. The different tiers should allow you to keep attacking for around ten points, and transferring those points up the pyramid.

But you need players to agree to stay within their tiers and attack the appropriate targets. The players doing all the point generation need to know that next time around they will be in the tiers and other players will be point batteries. And you need to get the system adopted widely enough that you can actually find appropriate targets and people aren't tearing down the higher tiers before they hit the rewards.

Unknown

I'll toss in a data point here for consideration.

Generally speaking, you can determine the cutoff point for the reduced loss scoring through racking up 100-200 points and then immediately retreating. In previous lightning rounds, if my score was over 100, then a retreat/loss would subtract 15% of my total score. If less than 100, the retreat/loss only deducted 10% of the total score. Thus, after scoring points in the event, the lowest possible score is 4 points, as (4 * 0.1 = 0.4), which rounds to zero, (4 - 0 = 4). For 5 points, (5 * 0.1 = 0.5 --> rounds to 1).

Interestingly, the deduction for low rating losses has significantly changed in the DD event. After scoring 1 win to register in the tourney, the lowest possible score I could achieve is 16 points. This would place the retreat/deduction at 3%, as (17 * 0.03 = 0.51 --> rounds to 1), whereas (16 * 0.03 = 0.48 --> rounds to 0).

It would be interesting to know where they moved (if at all), the line to where full points are lost on a defeat. Even if the line did not move, more points are being added to the available pool through the reductions in losses/retreats for very low-rating rosters.

Unknown

You're basically talking about a win trading scheme way back in the days of Starcraft. Roll a new account with rating of 1000, lose 10 games to some designated guy, and then repeat this to get a guy to say 1400 and then that guy loses to another designated guy, and so on.

The problem is this worked for Starcraft because you can pick your opponent. There's no reason to believe any other player would cooperate with this scheme. Anybody starting at a low point will see your feeder account as juicy targets to attack, and furthermore you also can't control who those guys lose to. Now, you can try to feed points into the general system by just constantly losing at around the 100 point range. In theory, if enough people do this, it'd be the same as shielding. That is, suppose you attack someone with 300 points at 100, you yield and give them 20 points and you lose 5, that is 15 points generated. It's not quite as good as a shield generating 30+ points per fight (usually the shielded guy is the highest point guy) but it'll still get you there, eventually.

Eddiemon

Lyrian wrote:

I'll toss in a data point here for consideration.

Generally speaking, you can determine the cutoff point for the reduced loss scoring through racking up 100-200 points and then immediately retreating. In previous lightning rounds, if my score was over 100, then a retreat/loss would subtract 15% of my total score. If less than 100, the retreat/loss only deducted 10% of the total score. Thus, after scoring points in the event, the lowest possible score is 4 points, as (4 * 0.1 = 0.4), which rounds to zero, (4 - 0 = 4). For 5 points, (5 * 0.1 = 0.5 --> rounds to 1).

Interestingly, the deduction for low rating losses has significantly changed in the DD event. After scoring 1 win to register in the tourney, the lowest possible score I could achieve is 16 points. This would place the retreat/deduction at 3%, as (17 * 0.03 = 0.51 --> rounds to 1), whereas (16 * 0.03 = 0.48 --> rounds to 0).

It would be interesting to know where they moved (if at all), the line to where full points are lost on a defeat. Even if the line did not move, more points are being added to the available pool through the reductions in losses/retreats for very low-rating rosters.

I like your methods, but from observation I believe the game actually stores the non whole element of your score and rounds it before display. So two 0.4 losses in a row may lead to a 1 point drop, or it might require 3 losses deepening on your starting score. 4.4 would take 3 losses to be displayed as 3.

Unknown

Phantron wrote:

Now, you can try to feed points into the general system by just constantly losing at around the 100 point range. In theory, if enough people do this, it'd be the same as shielding.

Yeah, I was trying to think of something that just inflates the point pool at large for anyone around your (fluctuating, anyway) MMR range, precisely because MMR hiding most people from one another and lack of any in-game comms would make explicit collusion extremely difficult (You got me queued? Right, shielding now...etc; does Ventrilo still exist?). Also not thinking of taking turns at feeding/pushing for entire events, more that multiple people would modify their play pattern for the first 30 hours or whatever to focus only on injecting as many free points as possible, then proceed as normal for the remainder. You'd probably need thousands at it to achieve a noticeable effect, so not even everyone who frequents this place would cut it, I just can't help but look for ways to make systems do what I want them to instead of vice versa.

Unknown

Eddiemon wrote:

Lyrian wrote:

I'll toss in a data point here for consideration.

Generally speaking, you can determine the cutoff point for the reduced loss scoring through racking up 100-200 points and then immediately retreating. In previous lightning rounds, if my score was over 100, then a retreat/loss would subtract 15% of my total score. If less than 100, the retreat/loss only deducted 10% of the total score. Thus, after scoring points in the event, the lowest possible score is 4 points, as (4 * 0.1 = 0.4), which rounds to zero, (4 - 0 = 4). For 5 points, (5 * 0.1 = 0.5 --> rounds to 1).

Interestingly, the deduction for low rating losses has significantly changed in the DD event. After scoring 1 win to register in the tourney, the lowest possible score I could achieve is 16 points. This would place the retreat/deduction at 3%, as (17 * 0.03 = 0.51 --> rounds to 1), whereas (16 * 0.03 = 0.48 --> rounds to 0).

It would be interesting to know where they moved (if at all), the line to where full points are lost on a defeat. Even if the line did not move, more points are being added to the available pool through the reductions in losses/retreats for very low-rating rosters.

I like your methods, but from observation I believe the game actually stores the non whole element of your score and rounds it before display. So two 0.4 losses in a row may lead to a 1 point drop, or it might require 3 losses deepening on your starting score. 4.4 would take 3 losses to be displayed as 3.

That outcome is plausible, although well beyond the point of diminishing returns. A strange anomaly when I was trying to reduce my DD tourney score from 17 to 16, was that the game often reported "Your team was defeated 17 (-1)", yet the point never was subtracted. I don't remember what the exact number of retreats/losses was, but it took a long time to finally reach 16 points. The game then reported, "Your team was defeated 16 (0)", thereafter. Achieving 15 points might be technically possible, if there is a miniscule non-zero deduction still occurring, but wracking up 20+ losses to achieve that 1 point further deduction is well beyond the required effort needed to obtain such a minimal deduction.

mischiefmaker

Definitely a good thought experiment, thanks for bringing it up! It seems to me that the dominant factor in determining the size of the overall point pool is the second one you mentioned; namely, the threshold at which point loss is equivalent to point gain. I think it's either 200 or 400, but I'm not sure, and it might vary per tournament (see "Hood-style scoring", where I'm pretty sure the threshold was 100). I also think that above the threshold, point gain is always the same as point loss, as opposed to the system we had once upon a time where you lost more than the winner gained if you were above 800.

If my second conjecture is correct, then that means that points, once injected into the system, are never lost; if it isn't correct, then trying to inject points is way harder because you have to inject more points into the system than there are points being lost at the top. As Eddiemon and Phantron point out, that is probably impossible to coordinate.

If we assume that points are never lost (which I think is correct, but this should be not too hard to test -- next lightning round, get two forumites who are both trying to place, near the end of the tournament, have one hit the other and report scores), then getting to higher progression rewards should in theory be simple:

1. Encourage as many people to enter as possible.
2. Encourage everyone to push their rating as high as possible, until they hit the point loss = point gain threshold.
3. Once you get above the threshold, tank as hard as you can so that you get as low as possible. Repeat until you make a run for progression rewards.

Also, if this is correct, then over time, if the number of active players is growing, then progression rewards will become easier to reach without us doing anything.

turul

"Working" point injection can be seen in the Elite tournament.

Waiting someone high-ranked too unshield - queue him - he reshields - attack.
However for regular tournaments this is too costy, and also no people even trying to get top progression rewards.

Instead of trying to inject points with the methods specified above, we should encourage developers to make shielded players visible with a smallish chance. (not from queue/retaliation)

Unknown

turul wrote:

"Working" point injection can be seen in the Elite tournament.

Waiting someone high-ranked too unshield - queue him - he reshields - attack.
However for regular tournaments this is too costy, and also no people even trying to get top progression rewards.

Instead of trying to inject points with the methods specified above, we should encourage developers to make shielded players visible with a smallish chance. (not from queue/retaliation)

I've been saying just make shielded guys only show up as first target, never on a skip. That way, it also encourages the 'fight your first opponent' thing that they try to encourage with the victory bonus/skip tax. Obviously you still wouldn't know who is actually shielded, but you'll know that you can never get a shielded opponent (clearly the best guy to attack) if you skip anyone at all.

turul

Phantron wrote:

turul wrote:

"Working" point injection can be seen in the Elite tournament.

Waiting someone high-ranked too unshield - queue him - he reshields - attack.
However for regular tournaments this is too costy, and also no people even trying to get top progression rewards.

Instead of trying to inject points with the methods specified above, we should encourage developers to make shielded players visible with a smallish chance. (not from queue/retaliation)

I've been saying just make shielded guys only show up as first target, never on a skip. That way, it also encourages the 'fight your first opponent' thing that they try to encourage with the victory bonus/skip tax. Obviously you still wouldn't know who is actually shielded, but you'll know that you can never get a shielded opponent (clearly the best guy to attack) if you skip anyone at all.

Good idea too

Yeggy

It seems to me that the tactics in this forum have proliferated a "bad" practice of non-playing. People try to play only in the last days of tournaments, trying to abuse the system by tanking and such. While this may have worked in the past, note that Demiurge is trying to change how the system works. One must adapt to the changes.

In my opinion, the best way to promote most people getting progression rewards (in PvP) is to simply play from the moment the tournament is released, gradually. If the tournament is 2 days time. Play only for progression rewards on day one, don't use shields. You will get retaliations, keep playing, you will populate the tournament with points along with others and keep climbing higher. The more people play as opposed to "non-play" with shields the easier it will be for everyone to get progression rewards. Then, on the last day, all bets are off, go for that nº1 spot, spend 5k HP on shields, do whatever makes you giggle.

If you want to be a super-duper awesome guy, start day 1 (or the first few hours) with a weaker team so people can attack you easier as well (but this is only if you're like, super cool, thanks bro/sis?). Or you can take a team that's awesome on offense and weak on defense, good hero synergy for player, bad synergy for the (DERP) AI. Low HP champs that can break through most trouble, but can easily be broken apart by picking off one hero. Try some crazy combos, have fun, play the game!

In any case, you should play however you want, but try not to encourage people to not play and only make pushes in the last few hours. If this works well for you, keep doing it. If you encourage others to do this as well, suddenly everyone is doing it and hell breaks lose. Noone is getting progression rewards.

All of this is in the interest of Progression Rewards in PvP, mind you. If you really need that nº1 spot, do what you think is best!

It may be demoralizing to get around 100 points in 30 minutes only to go away for an hour and see you dropped 150. If everyone keeps beating everyone else, everyone will get the progression easier. This is one issue I can see as not encouraging people to play. It's counter intuitive. If you're at low 400ish it is counter intuitive that losing a lot will make it easier for you to get progression rewards later. But! It does. If everyone keeps bashing each other.

I think that Demiurge implement most changes to PvP with the intent to increase the desirability to play a few matches every couple of hours, every day, not to play just at the end. Things don't go as you plan, people find loopholes, they find a way to abuse the system. It will keep happening, we like efficiency, we like dominant strategies. Do know that their intent is for us to play! If you just keep playing whenever you want/can, the system will work better.

*exhale*.

To finish I want to say that these are my opinions and not facts =P

turul

Yeggy wrote:

It seems to me that the tactics in this forum have proliferated a "bad" practice of non-playing. People try to play only in the last days of tournaments, trying to abuse the system by tanking and such. While this may have worked in the past, note that Demiurge is trying to change how the system works. One must adapt to the changes.

In my opinion, the best way to promote most people getting progression rewards (in PvP) is to simply play from the moment the tournament is released, gradually. If the tournament is 2 days time. Play only for progression rewards on day one, don't use shields. You will get retaliations, keep playing, you will populate the tournament with points along with others and keep climbing higher. The more people play as opposed to "non-play" with shields the easier it will be for everyone to get progression rewards. Then, on the last day, all bets are off, go for that nº1 spot, spend 5k HP on shields, do whatever makes you giggle.

If you want to be a super-duper awesome guy, start day 1 (or the first few hours) with a weaker team so people can attack you easier as well (but this is only if you're like, super cool, thanks bro/sis?). Or you can take a team that's awesome on offense and weak on defense, good hero synergy for player, bad synergy for the (DERP) AI. Low HP champs that can break through most trouble, but can easily be broken apart by picking off one hero. Try some crazy combos, have fun, play the game!

In any case, you should play however you want, but try not to encourage people to not play and only make pushes in the last few hours. If this works well for you, keep doing it. If you encourage others to do this as well, suddenly everyone is doing it and hell breaks lose. Noone is getting progression rewards.

All of this is in the interest of Progression Rewards in PvP, mind you. If you really need that nº1 spot, do what you think is best!

It may be demoralizing to get around 100 points in 30 minutes only to go away for an hour and see you dropped 150. If everyone keeps beating everyone else, everyone will get the progression easier. This is one issue I can see as not encouraging people to play. It's counter intuitive. If you're at low 400ish it is counter intuitive that losing a lot will make it easier for you to get progression rewards later. But! It does. If everyone keeps bashing each other.

I think that Demiurge implement most changes to PvP with the intent to increase the desirability to play a few matches every couple of hours, every day, not to play just at the end. Things don't go as you plan, people find loopholes, they find a way to abuse the system. It will keep happening, we like efficiency, we like dominant strategies. Do know that their intent is for us to play! If you just keep playing whenever you want/can, the system will work better.

*exhale*.

To finish I want to say that these are my opinions and not facts =P

The system should let any player reach the top progression reward with only the bottleneck of Health Packs, regardless of other players activity. But the current system just doesnt let anyone reach it*, regardless of player activity.

*except shield abusing, maybe