Jump to content
EleTD.com
Sign in to follow this  
ArthurDent

Speed vs Income scoring discrepancy

Recommended Posts

Quote

We are going to fix that next patch by getting rid of the networth bonus in classic mode.

 

I dont think thats the best solution.

The best "pure speed" score on the leaderboard is around 380k away from the best "pure slow" score. If you remove the networth bonus at the end, the slow build might lose around 100k.

With the new increased overall gold in 1.5 the speed build should only be very slightly in the negative or even in the positive networth bonus if played well and therefore the score should be around the same.

So the overall gap is still huge.

 

It would also nerf "random" alot since the 10% more networth at the end are worthless.

 

How about changing the end-speed-bonus/penalty instead?

 

For example (those numbers and timestamps are just made up on the fly and need tweaking of course, but you defenitely need some sort of an exponential growth to make it balanced):

 

Penalty: Keep it the same for minute 45-46, then add 10% to the penalty of 45-46 for minute 46-47, then add 11% to the penalty of 46-47 for minute 47-48, then 12% and so on.

 

Same for the reward from minute 45 downwards, just faster scaling. (this is obviously the tricky part to balance, since speedbuilds endscore relies alot on that bonus)

 

I think this way you could create an extremely good balance between fast and slow build scoring.

Maybe you can even use some data from your statspage for a "speed(time to end lvl 55)/score(after frogs)" graph to get pointed in the right direction.

Share this post


Link to post

Oooh it's way more than 100k. See, how a lot of those top builds work is using a dumb abuse trick (that unfortunately, too many players like to just remove it) that gets them stupid amounts of networth. How the networth bonus at the end of wave 55 works is:

  • Your Networth / Base Networth
  • Base Networth = the combined bounty values of all creeps

And then it's just a straight multiplier. On Insane, it's roughly 100k networth. With the buy/sell abuse trick, players are getting anywhere between 200k and 300k, meaning a x2 to x3 multiplier. The base score for Insane, without any modifiers, is about 200k. On all pick, even if you run a slow build, you can get maybe a x1.5... at best. On random, if that's going up to a x3, that's still a 200k score difference.

 

Additionally, more networth = naturally stronger defense = more frogs killed = more score anyways, so it's a rather redundant measure on top of frog score. It's not redundant on Express cause there is no boss wave afterwards, plus trying to interest abuse in Express... there's not enough time for it to truly accumulate into broken levels of brokenness.

 

 

That said, going based on the penalty system you suggest would be a little... iffy. Main reason being, yes, that would work....... for the top 1% of players. A very large chunk of the playerbase clears slow naturally, not because of abuse, but because they're not masters of the game yet, and whether they're playing for score or not, suddenly seeing their score of 250k suddenly drop down to 100k for reasons they don't understand would be a very big ".... huh?"

 

There is another option we're considering that'd be somewhere inbetween what we have now and straight out cutting out the Networth Score Bonus. Someone suggested that instead of the Networth Score being a multiplier, it's a flat amount. 1 Networth = +1 Score. For anyone using All Pick, they probably won't notice it at all, but for anyone on Random with their Buy/Sell Abuse shenanigans, that'd nerf 'em a fair 100k to 150k score, and even more due to Speed being the only multiplier left. It's probably not quite enough, so maybe even 2 Networth = +1 Score.

 

 

To be honest, the No Networth Score Bonus hasn't been tested yet. We're still working on it, if it proves to not be enough, we'll try out other alternatives, ie tweaking the Speed Bonus, trying the Non-Multiplier Networth Score Bonus idea, maybe even trying a steeper penalty (though with an altered formula that doesn't screw over the majority of the playerbase).

 

 

The last resort we've got would be to directly nerf Buy/Sell Abuse by going with 99% sell. The reason we're not doing yet is because although it is a huge issue with regards to the leaderboard, it would effectively nerf everyone that isn't doing Buy/Sell Abuse. Other alternatives have been proposed, but ultimately, they all still nerf everyone else, so... they're on the backburner, it is an option, we're just not relying on it yet and seeing if we can alter something like score that doesn't actually potentially break the balance of the game.

Share this post


Link to post
Quote

Oooh it's way more than 100k. See, how a lot of those top builds work is using a dumb abuse trick (that unfortunately, too many players like to just remove it) that gets them stupid amounts of networth. How the networth bonus at the end of wave 55 works is:

  • Your Networth / Base Networth
  • Base Networth = the combined bounty values of all creeps

And then it's just a straight multiplier. On Insane, it's roughly 100k networth. With the buy/sell abuse trick, players are getting anywhere between 200k and 300k, meaning a x2 to x3 multiplier. The base score for Insane, without any modifiers, is about 200k. On all pick, even if you run a slow build, you can get maybe a x1.5... at best. On random, if that's going up to a x3, that's still a 200k score difference.

About Random mode:

 

I hope this doesnt sound too rude, but it seems to me your networth numbers on random builds are only estimations and far from accurate.

 

I was comparing slow vs. speed builds, sucessful random builds are usually pretty fast builds (and defenitely no slow builds if you compare the times at the leaderboard), so I think the up to 300k networth is a very huge overestimation for the current top random scores. Just look at the amount of frogs they killed (and the 2 players at position 3 and 4 have pretty good elements). If those two had more than 200k networth, they would have killed way more frogs for sure.

 

200k+ networth might be possible if you get the right elements in a decent order to get early wealth towers, but it would defenitely be very, very unlikely. (and wealth towers with slow attackspeed and high range are quite impractical to rebuild every 15 seconds)

 

Nevertheless you are defenitely right about the posibility to get alot of networth with that trick. Maybe increase the time towers need to be built and upgraded by a percentage (especially higher tier towers, arrow and cannon towers should stay the same imo) or add a 2 second cooldown on upgrading after each upgrade/construction (so it wouldnt effect allpick mode that much if you upgrade a tower during a round) to prevent consecutive upgrades.

This way people could still use the trick, but the usability would be nerfed alot for stronger towers which require multiple upgrades.

 

Back to the slow vs. speed (allpick mode) discussion:

The best slow build right now has an endscore before frogs of around 262k points. If you estimate a value of 200k networth, you would lose 131k score with the removal of networth bonus. If you estimate 250k networth, the score would be 158k less. So it wouldnt close the gap between fast and slow builds enough.

I dont say the removal of networth bonus is generally a bad idea, but I dont think it would do the job alone.

 

Quote

Additionally, more networth = naturally stronger defense = more frogs killed = more score anyways, so it's a rather redundant measure on top of frog score.

Quote

There is another option we're considering that'd be somewhere inbetween what we have now and straight out cutting out the Networth Score Bonus. Someone suggested that instead of the Networth Score being a multiplier, it's a flat amount. 1 Networth = +1 Score. For anyone using All Pick, they probably won't notice it at all, but for anyone on Random with their Buy/Sell Abuse shenanigans, that'd nerf 'em a fair 100k to 150k score, and even more due to Speed being the only multiplier left. It's probably not quite enough, so maybe even 2 Networth = +1 Score.

I personally do like the networth bonus as an additional reward for efficient play in builds which are not aimed at pure income/slow killing, removing the multiplier and making it additive would weaken this,but its defenitely an interesting idea.

 

Quote

It's not redundant on Express cause there is no boss wave afterwards, plus trying to interest abuse in Express... there's not enough time for it to truly accumulate into broken levels of brokenness.

I am not sure if I understood this the right way. Are you saying networth bonus is important for Express?

According to the Express leaderboard, it seems to be purely about speed right now.

 

Quote

That said, going based on the penalty system you suggest would be a little... iffy. Main reason being, yes, that would work....... for the top 1% of players. A very large chunk of the playerbase clears slow naturally, not because of abuse, but because they're not masters of the game yet, and whether they're playing for score or not, suddenly seeing their score of 250k suddenly drop down to 100k for reasons they don't understand would be a very big ".... huh?"

 

I dont agree here.

 

If the average gametime of new/casual players is very slow, the removal of networth bonus wouldnt help that either. I might not have enough insight how the endgame bonuses of new/casual players are composed right now but I think think they rather get a positive networth bonus than a positive speed bonus right now with the huge bounty increases in 1.5.

You could for example move the point of origin for speed bonus to 50 minutes to give casual players a better reward and still tweak the numbers in my suggestion in a way to prevent someone who can finish the game in 35 minutes to get a too high score. You just need someone who is quite good at math to find a formular for it, but there is one for sure.

Maybe it wasnt clear in my suggestion, but the important part was not about penaltying slow play (since the majority of points comes from frogs anyways if you compete for the leaderboard), it was about rewarding faster play in a way it can compete in highscore with slow/income play.

 

I know the main intention of my suggestion is mostly about balancing the scores for people who play way too much ele TD like me, but I think you can still find a decent balance for casual players with it aswell.

 

Anyways, I am sure you will find a good solution to this, as you managed to improve the game with every patch so far.

 

P.S.: This post became way too long, but since I didnt manage to improve my highscore without slow/income builds lately, I waste my time here instead.
 

Share this post


Link to post

Ah, regarding Express, I'm not saying a Networth Bonus is entirely necessary, just that removing or altering it won't really have that much of an effect, so might as well keep it for the heck of it.

 

Regarding my estimations on the networths, I was actually pulling from Utoko's numbers from his own testing. As for increasing build times.. it wouldn't really fix it on the lower end, but that'd definitely fix it later especially when you're able to buy/sell a frigin' Pure Tower (or worse, two of them). There is a mode I actually fully support 100% random and buy/sell on because it's practically impossible to abuse - Random, Chaos, Rush, Express. Problem is, as ideal as that gameset is for all the tricks of the game required just to beat it, it's a rather niche mode that not many play and totally caters to the 1%. Anyhow, between increasing build times or 99% sell... shibby, that's a hard choice, but either way, I think I'd stick both of them as last resorts.

 

Okay, Speed Bonus formula tweak. Technically, we don't need one fancy formula for it, we can split it into multiple formulas.

  • If your time is equal to or less than 50 minutes, your score is multiplied by (50 / Your Time)
  • If your time is greater than 50 minutes, your score is multiplied by 1 + (5 * [(1 - (50 / Your Time)])

Aka, for each minute over that threshold, you lose 5%. Well, bit of an arbitrary number thar, but it's a fair point to start out.
 

 

And yeah, we'll try out the additive Networth Bonus rather than completely axing, but only half, so 2 Networth = 1 Score.

 

 

1 hour ago, ArthurDent said:

Anyways, I am sure you will find a good solution to this, as you managed to improve the game with every patch so far.

 

Funny story, every patch prior to Patch 1.5 kept exaggerating a particular problem till it finally hit me "oops, I screwed with the core balance of the game without realizing it, we need some really big changes to fix it". We do try to look at what the biggest issues are each patch and try to fix them, but sometimes, there's some underlying problems that we don't even consider until it grows and grows and eventually punches us right in the face. We actually did know about this Networth/Speed Bonus issue prior to Patch 1.5, but because that patch essentially changed the game, we didn't wanna screw with the scoring system until afterwards.

 

Thanks fer the input, and keep it comin'!

Share this post


Link to post
44 minutes ago, WindStrike said:
  • If your time is equal to or less than 50 minutes, your score is multiplied by (50 / Your Time)
  • If your time is greater than 50 minutes, your score is multiplied by 1 + (5 * [(1 - (50 / Your Time)])

Aka, for each minute over that threshold, you lose 5%. Well, bit of an arbitrary number thar, but it's a fair point to start out.

 

Sounds like a good idea, but the second formula got some mistake since it would increase the score by roughly 10% if you need 51 minutes.

 

48 minutes ago, WindStrike said:

And yeah, we'll try out the additive Networth Bonus rather than completely axing, but only half, so 2 Networth = 1 Score.

 

Please add it before the speed bonus multiplier :D

Share this post


Link to post

Meh, I suck at math, but thankfully, the programmers have more brains than I do at mathing up some formulas. I still don't know when we're releasing this patch; we're hoping this week, but there's a potential obstacle with that, and not just work on the co-op version (which is close to done), but there's no way we're releasing this patch at the same time as Dota 2 Patch 6.87, or else all our hype up and whatnot for co-op will be for nothing. Should they end up releasing this week, we'll probably push back to the start of next week, which would give us plenty of time to test out both the Networth & Speed changes. If we're able to release within the next few days though, Networth change would happen for sure, not positive on the Speed change though.

 

EDIT: Oh, it's supposed to be a minus, derp:

  • If your time is greater than 50 minutes, your score is multiplied by 1 - (5 * [(1 - (50 / Your Time)])

Share this post


Link to post
13 hours ago, ArthurDent said:

I was comparing slow vs. speed builds, sucessful random builds are usually pretty fast builds (and defenitely no slow builds if you compare the times at the leaderboard), so I think the up to 300k networth is a very huge overestimation for the current top random scores. Just look at the amount of frogs they killed (and the 2 players at position 3 and 4 have pretty good elements). If those two had more than 200k networth, they would have killed way more frogs for sure.

 

200k+ networth might be possible if you get the right elements in a decent order to get early wealth towers, but it would defenitely be very, very unlikely. (and wealth towers with slow attackspeed and high range are quite impractical to rebuild every 15 seconds)

 

 

I agree with your overall points but the random numbers are correct. 


I had several 700.000k+ scores but it doesn't show up because my ap score is higher


you can get 250k+ with most element combinations(no gold tower needed). If you play slow and keep sell/rebuy basically the whole game. 
Of course your prefrog score is then pretty low and you need to have the right towers for the frogwaves then. 

Here is one example with 320.000:

FC05FC7F19635AE4165508C161D9031A4A75E299


 

Share this post


Link to post

before I start you should know that I don't watch other people play this game often because I prefer to figure out strats my self and I feel by copying people I am taking away from the game and my own skill, with that said..

 

This is the biggest issue I have with this game right now is this abuse of buying and selling which is jumping people into retardedly high scores that are impossible without abusing, honestly can someone just flat out say how this works in regards to the game then build a discussion around how to remove it from possibly happening again in the future.

 

I am guessing (I have personally never seen it done..) it has something to do with selling all/most of your towers just before an income run finishes so you can get the extra income you wouldn't normally get without doing this and as the game gets further the money becomes exponentially higher given you can do it quickly enough and not leak from doing it.

 

if this is in fact the case why not simply limit the number of towers that are sell able at a time (or per round, but remove it once the icefrog rounds begin), make towers that have been upgraded beyond their base form permanent (cannot sell) or create some smart system that detects it and simply blocks the player from abusing it period/removes their chance to be entered onto the leaderboard.

Share this post


Link to post
1 hour ago, DBX_5 said:

This is the biggest issue I have with this game right now is this abuse of buying and selling which is jumping people into retardedly high scores that are impossible without abusing

First you say you don't watch the games and then you claim that these high scores are impossible without abusing these buy/sell tactics...

I wouldn't call it the biggest issue when the best scores have nothing to do with the buy/sell tactic.
97/250 high scores random


In fact it is not even easy to do in that extreme manner. If you mess up once you lose easy 15 lives in one wave.
 

and like I said even if you get the high net-worth it is very unlikely that you get a very good element combination. In most games you can kill with 250.000 gold the same amount of frogs as the other guy in -ap with 110.000 networth.

 

So with the change of the networth bonus I don't think this will be a issue anymore.


Even in all pick 220.000 gold is possible so the gap isn't even that high.




The Speed vs slow Income scoring discrepancy is the important and hard thing to balance.


 

 

Share this post


Link to post

Yeah, I looked into some things a bit more, as much as I hate the buy/sell tactic, while it originally did cause the highest scores on the leaderboard, various other tactics related to shenaniganing with slow-kill builds is the primary issue. The new Networth Score Bonus should theoretically help it, but it won't be enough. Something I'm trying is just tweaking the current End Speed formula, like doubling or tripling its effect, though unfortunately, I'm not a master of the slow-kill build, sooo I don't know if that's actually fixing it or not. Once the new Networth Score Bonus is in, I'll upload that to ETD Test along with trying double effect on the End Speed Bonus.

 

Heck, there's another option, which is to alter the wave speed bonus, such that it seriously screws over builds intended to wait on killing the last creep until the very end, though I'd have to make sure that it doesn't completely invalid something like a standard double pass build using the right lane.

Share this post


Link to post

I did my highscore without this "abuse" (still top 10), except maybe early game on accident, when i had to use other basic towers for the waves.

@ Limiting towers you are able to sell per round: Bad idea, as sometimes you really want to spam a lot of arrow towers fast to hinder some wave from leaking and get em away after the wave. Or some builds used a specific tower on level 2 for a couple of waves, which they did not need later any more. Doing what you suggest would screw a lot of possible paths, and i do not like that.

@ the topic:why not make selling a tower lock the income ticker for a certain period of time? The higher value the tower has, the longer the lock is.

Share this post


Link to post

Huh, sell locking the income timer..... I actually don't have an issue with that. I'd probably just make it equal to their total build times.... except pures, those can enjoy a nice 15 second delay, lol. Yeah, I like that idea, though twould like to hear more input on it. Since it's not as immediate a concern, it probably won't make Patch 1.6, but it's something to consider for a future patch.

Share this post


Link to post

The thing is: it should stack if you do it for more towers. In example, let's say selling an arrow tower locks it for 0,1 seconds, selling two should lock it for 0,2 and so on.

 

Share this post


Link to post

Imho that seems to me like a complicated implementation of the 98% sell value. It has the same effect.  

 

but it gets way more complicated because the lock times would need to be different when you have a different amount of money saved up. 

If you don't implement a scaling then you cant get a balance between the early and lategame.

 

Quote

0,1 seconds, selling two should lock it for 0,2 

If you use flat numbers you couldn't even use arrow towers anymore to prevent leaking in the midgame for example without losing way more money.



Can't really see any advantage of something like that which needs a lot of testing and balancing and in the end it would have the same effect as 98% or 99% sell value.

 


 

Share this post


Link to post
Sign in to follow this  

×
×
  • Create New...