todd's unfair advantage (in General)
after the discussion in the most recent "overpowered robf" thread, i have come to the conclusion that the mythical todd's unfair advantage is now an integral part of the game!
one assumption in cb has been that at the upper echelons if you could fight more than the top dog, then you could overtake that position. if we truly have different tiers in the 6/20 ba range getting different rewards then this may not be an accurate assumption. i think it is possible to fight more than number one and still fall behind. the devil is in the details really and only jon knows the percentages.
the only ways around this are ncb/nub or by purchasing minions and jumping in the tiers. it would seem to me though that everyone in the 6/20 range should be getting the same rewards and paying the same amount for purchasable ba, otherwise the system is designed to be advantageous to those in the higher tiers.
furthermore, the current system was probably designed and was more fair under the top ten exemption. when the top ten was changed to the top forty percent though it most likely exacerbated the issue and needs to be adjusted accordingly if not made flat for all in that range.
I think this post by NS is very important. And I didn't know CB worked this way;
"Ranger, rewards are based on your BA cost. You get higher rewards than anyone else because you have a higher BA cost.
BA Cost is based on VPR. Since you have the highest VPR, you have the highest BA cost, and thus the highest growth rate."
If this is true (and I've no reason to suspect NS is wrong here) something needs to be changed, otherwise the Top charcater will always be the fastest growing charcater...
January 9 2008 11:06 AM EST
I believe NS is correct. Works that way for forging, so should work that way for base rewards. That magical number (BA cost) came out as most important factor in forging a while back -- if nothing else, it is a way to normalize forge formulae effectiveness.
Having the top BA cost leading to higher rewards has generally not been an issue... However, lots of gameplay elements have changed recently to make me wonder more about it:
Issue 1: Top characters cannot achieve a challenge bonus, so will not grow as much -- This is true, an up-and-comer choosing better targets can grow very quickly. But that growth is still asymptotic to what the top player is doing.
Issue 2: NUB and NCB have a bonus, so they can get up high and then surpass the top -- I think we have seen that even with purchase of minions, that does not seem to be the case, is it?
Issue 3: Up-and-comers will be "hungrier" and will be able to use more BA to get beyond the top -- Now that BA is easier to use fully each day, this is not really the case. It is actually harder for a rising player to get all BA used, while the top player is at 6/20 and can even get 8 full hours of sleep between maximum BA accruals.
Issue 4: Higher players have to invest in more expensive stats with the increasing training cost -- No longer the case; training curve is now flat.
Issue 5: Higher players will not be able to clan as successfully because they cannot fit enough PR into their clan limits -- This is _clearly_ not a limitation, judging by one of the most successful clans out there.
Issue 6: Higher players have to spend more on weapon x, exponentially, and since that helps with damage, they will suffer -- No longer the case, x curve is now flat.
Issue 7: Higher players need to invest more in +, both on weapons and on items, and that cost gets prohibitive -- STILL TRUE. This is a limitation for higher players.
Issue 8: Higher players will run into ceilings for weapon PR -- top ten exemption makes this a non-issue.
Issue 9: Higher players will run into issues fighting below themselves -- top ten exemption makes this a non-issue.
Issue 10: Higher players will run into Max Tattoo Level ceiling and so not be able to grow that important aspect of their team -- Not true. MTL is huge and practically unable to ever reach (or maybe that's a fact -- maybe it cannot be reached at the top levels?)
So, there are ten issues. Probably doesn't cover everything. The one issue that is still non-linear and therefore "tops out" at high levels is + on armor and + on weapons. That's it, as far as I can tell.
I am not expressing any additional opinions or solutions (if anything even needs to be solved). I never really thought through this before, so just wanted to take this opportunity to do so.
Very good post, sut.
I just wanted to point out, dudemus, that it didn't go from the Top Ten to the Top 40%. It's people with more than 60% of the top MPR, which is quite a different thing. MPR of all characters is not shaped like a bell curve, or maybe it is, and then Ranger's just 300 Standard Deviations off to the right side. People who come to 60% of that are still in the top 3%, at least.
January 9 2008 11:40 AM EST
Not exactly a bell curve, it's more likely skewed to the left. You are right about the number of standard deviation Ranger is from the rest of the pack though :)
January 9 2008 12:13 PM EST
I agree 100% the current system is broken.
Since NS's post about rewards, it is intuitively obvious even to the most casual observer.
The biggest issue I see with the high-level game is the lack strategy in using Net Worth. Coming up through the ranks, a huge part of the strategy is how you use your net worth, since net worth adds to your PR and PR sets your fight bonuses. Once you hit the high end of the 6/20 scale, where scores are always much lower than PRs, that element of the strategy gets completely tossed out the window.
I think the high-level game suffers for it. I'm not going to suggest a solution for this yet, but this topic seems to be the right place to mention this issue.
January 9 2008 12:20 PM EST
I'll second that; after a certain point NW becomes 'bigger is better' rather than a strategic use of items. Well, not completely, but certainly enough that it's noticeable.
January 9 2008 12:31 PM EST
The NW of Koy right now is very very high making all strategy I put into it academic.
January 9 2008 12:33 PM EST
hey.. ranger do you have any lose items on koy or is the nw just from the items you have equiped!?
January 9 2008 12:38 PM EST
Only lose items on Koy are about 100k exshots that are starting my 2nd million.
The main problem is that the system was intended to work this way:
1) The top players do not get any negative bonuses for fighting under their PR.
2) Those under the top players do not need this advantage since there are enough players over them to get positive bonuses.
The problem is that the second tier characters can not get any significant bonuses. Without ANY positive bonuses they can not catch up without the NCB. The evidence of this is that the scores of those in the second tier are severely depressed (actually the scores of the top tier are depressed to). I would have thought that the top tier scores would soar above their PR since they seem to be collaborating in a clan and not beating on each other, but I observe this is not true. I must assume that an occasional beating momentarily drags their scores down by a dramatic amount when someone beats them. It's dang hard to build up score when fighting below your score so I suspect they see long periods of slow score growth punctuated by short dramatic score loss.
It would be nice if Jon added a rule that if you fight a character with more PR than you, even if the character has a lower score than you, you at least get the 0% bonus.
January 9 2008 1:25 PM EST
When my score was 5M, I would get about 40 score points a win.
so i guess that leaves us awaiting a response from jon on this one then.
ranger and i can both purchase the same amout of ba per day, mine costs 1094 while his costs him 1162 per ba. does he get xp rewards at an increased amount over mine as well?
does anyone else get between 1094 and 1162 or am i only one tier out? if your ba does cost between those amounts please post it here.
1094 here and I have a larger mpr than you....
January 9 2008 3:49 PM EST
I am at 1139.
January 9 2008 3:52 PM EST
Hm, if it weren't for forging (where MPR is tied to NW increases), BA cost should only be based on two things:
-- BA accrual rate (to normalize for the various rates of BA generation)
-- Bonus (NUB or NCB bonus rate)
Leveling out the BA cost would at least alleviate this apparently top-benefitting phenomenon.
if ya put all of us so far on a graph and also add koy you can see what i mean pretty well. hubbel is in line with the rest of us and possibly growing faster, it is really hard to tell for sure though. koy's growth however is obviously better. what i do not know is how much ba everyone gets and how well they choose their targets.
when doing the history graph i was showing only mpr and choosing the period of 99 days. you can see some little retrain bumps and times when some used roe's, but the overall line is also pretty easy to pick up that way.
History graphs do not show the real picture. The only information that you need is in the average rewards thread from a while ago where I was listing the average rewards based on BA cost. Avoid's method of figuring out your average rewards still works great, and it shows that as your cost goes up, so too do your rewards.
well if that is true then this would be a call for change. that is an unfair advantage that forces the use of the roe or n*b as the only method of catching up, at least until you can get in the same tier as number one.
This is what I've been saying for months now.....
January 9 2008 5:07 PM EST
I have to admit NS, your comments about catching the top have been met in this corner with a slightly skeptic raised eyebrow. But that was before you brought up the BA price/rewards 'gap'. I'm firmly a believer now. Thanks for completely killing all hope of catching the top lol. ;)
i have been as well. i just didn't know what the specific culprit was and am still not sure how i missed linking the two in the thread you mention.
i have to apologize to battle royale though as much of my frustration with them in the threads regarding a forced break up or redistribution was wrongly focused on clan bonus. that was all i could come up with though as a difference but knew that shouldn't be enough of a difference.
January 9 2008 5:10 PM EST
/me whispers: "Rolling Bonus "
i would love a rolling bonus, in the meantime though, as that would take time, please make everyone in 6/20 have the same rewards or everyone in the no challenge bonus get it, whichever would be easier.
January 10 2008 9:34 AM EST
I don't get it... isn't the TOTAL possible reward per day the same anyway ?
I always believed that whenever your BA regen rate was 9 insted of 10, your rewards would be 10/9 of the rewards you had when you were at 10 BA regen rate, assuming you fought with similar sized (both absolute and relative) opponents in both cases.
In other words, if your BA regen is 8, you get 10/8 (+25%) rewards, and so on and so forth.
However, the NUMBER of BA you can spend daily is also proportionally smaller, so overall, you should get the same increase rate.
Am I wrong in assuming that, and it is only valid for 100% challenge bonus fights ?
In other words, do you need to always fight at 100% challenge bonus (or very close to it) to get a chance of catching up the top tiers ?
Or, at least, if not 100% challenge bonus, then BAregen*10[%] challenge bonus ?
those are the main divisions, there are smaller ones within those brackets though. we are all in the 6/20 range. within that range though our ba costs different amounts and if we are correct our fights give different rewards. however we all get the same amount of fights per day.
this is why it is not quite fair to everyone. we are all doing the same thing and are limited by the same constraints, it is just that some get more xp and cash for it. this is also why it is impossible to catch up to those in the other bracket barring adding a minion, using a roe or having some other type of non-normal bonus.
January 10 2008 10:08 AM EST
Oh. In that case, yeah, it's kind of broken then.
So, for instance, if somebody smack-dab in the middle of a 7 BA refresh gets aprox +42% rewards, somebody in the bottom regions barely gets above +33% while the one in the higher regions get almost +54% ?!?
That's... for lack of a better word... stupid.
January 10 2008 10:13 AM EST
Well, I don't think the disparity within any given BA accrual range is quite THAT high... Ranger's BA costs something like 1164, mine costs 1139 -- that would mean he gets around 2% more than me... Someone at 1094 vs. Ranger's 1164 would be missing out on about 6% or so (assuming the BA cost is some sort of linear value-determiner for rewards -- I should go read NS's thread to see what the relationship was...).
January 10 2008 10:14 AM EST
I can see a very simple solution however.
Not the most elegant, but simple nevertheless.
Make BA gains fractional.
That is, for instance, somebody in the lowest tiers of "6/20" BA regen actually gets "6.99/20" BA regen.
You can only spend "full" BA amounts, of course, but you can both buy more daily and get more daily.
January 10 2008 1:09 PM EST
That's an interesting concept... Simple and effective, except for the fact that Jonathan probably has everything driven my integer types (though that may not be that big of an issue in the TCL code, not sure how it is structured...)
January 10 2008 1:36 PM EST
Even if it's integers, that too can be bypassed by making all BA gains and caps x10/x100, and one "action" costing 10/100 BA... as fine a blend as you like it to be.
Heck, even allow "negative BA" for a single action (so, for instance, if you are at 1 BA and spend 10 on one action, you end up with -9, then have to get at least 11 more to make a move again).
January 10 2008 1:58 PM EST
Like mana regeneration! *smile*
Sickone, you are one creative kid.
So what you're suggesting is to make it fractional, like you said above, but to multiply everything involved by x100 in order to make it use only integers, correct? Like those in the lower levels of what is now the 7/20 BA regen rate would get 700/20 mins, and then as they go up, it would slowly move towards 600, until the next threshold? Or, alternatively, the 7/20 range would go from 799 down to 700, and then the 6/20 range would start at 699, with 600 then being the least you could regen (and what the top characters get)?
I do like it as a solution, but that would create a bunch of mini-thresholds within each currently-existing bracket. Like right now it's 2.5%, 15%, 40%, 60% (I think those are right... I got them from my memory), but at that point it would be "at 2.5% of the top MPR, you start getting 999 BA/20 minutes. Then, your obtained BA continues to go down by one every 1/8th of a percent of the top MPR that you increase, down to 900, which you should hit at about 14.875% of the top MPR. When you get to 15%, you will get 899 BA/20 minutes, and this amount will decrease by one BA every time your PR advances by 1/4% of the top MPR, down to 800." etc. etc. The page on BA would become the longest in the wiki.
I think the idea can be taken into consideration and worked on some more, but I think that it is not quite viable right now, because Jon would have to change code, multiplying everything by 100, and then throw all these extra BA rate ratios in there. I think if he started it off in such a way that there are only 5 brackets, he probably wanted it to be that way, for ease and space on the server, if nothing else. Still, a good contribution, and I'm not saying it should be entirely dismissed. I'm just troubleshooting.
do not say mana regen or ill say weapon durability!
"-- that would mean he gets around 2% more than me... "
anecdotally, i would guess the difference being greater than that. when i was tracking my growth against koy's, his numbers were quite a bit better than mine. when i would use my roe though, my growth was then just even with his or slightly better. at that time i think i was only getting a 10 to 12 percent clan bonus compared to his which was maxed most of that time.
If you want to see how much faster he grows than you, have him calculate his average rewards. Then, calculate your own. There's your difference. Multiply it out by 648 battles a day, add in XP time, and you'll see how things are progressing (ignoring clan bonuses).
ok, i finally had the time to figure mine out, i am putting it here so everyone can see it and correct me if i pulled a stupid. hopefully ranger will do the same, i used xanas as a test subject. i will run some more tests later and check the numbers with other characters to verify.
fight one cash reward: 867
fight two cash reward: 771
i then took both of those numbers and divided by 1.146 (14.6 percent clan bonus) to get my pre bonus base reward. this gave:
756 & 673
756/673=1.123 therefore i take the lower amount 673 and multiply it times 13/16. which gives:
547 which should be my average cash reward. to get my average xp reward i divide 547/1.655 to get my average xp which is:
i am including avoid's method below for others and to make it easier to check me math!
The first step is to attack a character twice, and record the cash rewards for each battle. If you get the same number twice, or if they are only a dollar or two off, you'll need a third battle.
Then Take the higher of your two values and divide it by the lower.
If you get 2.25,2,1.75 or 1.25, multiply the lower cash value by 13/8 to get your average cash reward.
If you get 1.8,1.6,1.4 or 1.2, multiply the lower cash value by 13/10 to get your average cash reward.
If you get 1.333 or 1.667, multiply the lower cash value by 13/12 to get your average cash reward.
If you get 1.286 or 1.143, multiply the lower cash value by 13/14 to get your average cash reward.
If you get 1.125, multiply the lower cash value by 13/16 to get your average cash reward.
If you get 1.5, there are two possibilities for what your average cash reward can be - low*13/8 or low*13/12. You can look at both and guess which is right, or you can fight another battle and try again, or you can find a nice quiet corner and cry.
Doesn't work for xp, only cash, but you can get your average xp rewards by dividing your average cash rewards by 1.655.
You can be even more accurate by doing it on Sunday. In that other thread I posted my rewards for my previous BA cost bracket. On Sunday, I'll post my rewards for my current BA Cost bracket (1,094)
i am in 1094 as well and just did it again using nemesis' character. with that one i came out 324 as average xp rewards, which is pretty darn close.
Remember that there is an unknown hidden bonus for fighting clan characters, though I guess since everyone only fights clan characters, that becomes moot..
well fighting a non clan character gave me 288 as average xp if i didn't mess up. i think i will wait until tomorrow and do more numbers then as you suggested.
Let me posit a theory, one that I have confirmed myself:
Your rewards on not based on your BA cost. Rather, they are based on the BA cost of the opponents that you fight. Actually, that isn't even accurate. It is based on the PR of your opponents. The higher your opponents PR, the bigger your reward. This is independent of any visible bonus. The effect is most visible when fighting low PR minions, but it is still measurable at higher PR.
Unfortunately, I have lost (read: already deleted) the data on which this is based, so I can provide no proof. I will leave it as an exercise for the reader.
January 12 2008 1:51 PM EST
Synge, correct, that is how challenge bonus is calculated.
We are talking about base rewards, ones where challenge bonus is zero percent.
That base reward does appear to be based on BA cost, just like forging "rewards" (i.e. forge NW yield).
Note how I said that this is independent of visible bonuses. Challenge bonus is a visible bonus. What I am talking about is not challenge bonus.
well using avoid's method today, i fought six different characters and got 6 pretty different results, they are in the same ballpark though but i am just not sure how accurate it is.
the six numbers were: 317,255,331,282,262, & 303.
if average those, i get 292 as my average rewards.
anyone else care to test this?
hmm, maybe it is accurate with at least six data points, i just logged 25 fight rewards in excel and averaged them and came out with 292.4 as well.
as another check i just trained, was left with 11 xp, fought 100 battles had 29008 xp afterwards. this comes out to an average of 290 xp per ba.
January 13 2008 11:16 AM EST
For cash rewards, I got:
NWO, high/low = 701/526 --> 570
Lega, high/low = 687/601 --> 558
Violent Femmes, high/low = 673/337 --> 548
Duixie Cousins, high/low = 701/526 --> 570
MrMunch, high/low = 694/309 --> 502
Nemesis, high/low = 576/330 --> 536
Average cash reward for those six opponents = 547.
My BA cost is $1139. Cash reward to cost ratio is 0.48.
all of my averages are xp btw.
i get your average xp as being 331 then sut. that would be 40 xp higher than mine which comes out at about 14 percent higher.
January 13 2008 11:26 AM EST
I stuck with cash rewards because it ratios properly against BA cost, in case anyone finds that number of interest. Is your average cash reward about half of BA cost?
Running 26 battles, average xp per minion was 78.9. Multiply that by 4 minions and it comes out to 315.7. Not too far off from the expected 330 xp reward that would be expected (dividing my 547 cash reward by 1.655).
"Is your average cash reward about half of BA cost? "
that would be affirmative, or at least darn close to my average.
using your other number, you get 8% more xp. so in reality it is probably somewhere between 8 and 14% higher xp than i get. i think you are in the next bracket up from me also.
January 13 2008 11:40 AM EST
You are 7/20? Ah, I thought you were 6/20. No matter -- as with the forging, by normalizing it against BA cost, we get the magic ratio. I need to find NS's other post to see if ~0.5 was what that came up with (cash rewards to BA cost). If so, then there is no doubt about it -- the higher the BA cost, the higher the reward, and that already builds in the 10/9/8/7/6 accrual rates.
by next bracket i meant that between my cost and yours there is not another level. i think that you are in the bracket below ranger. little brackets, not regen just ba cost, not big ones.
January 13 2008 1:35 PM EST
First 2 numbers are the money rewards. The 3rd is the average cash rewards per Avoid's method.
351 438 570.375
Vs Violent Femmes
589 757 546.928
438 788 569.4
vs Black Sophist
601 344 559
344 515 559
772 687 558.125
vs double trouble
678 452 490
473 631 512
For those that top of my fightlist: Average is 559$ which is 338xp per battle.
However one can see if I fight a bit lower, the rewards do go down a bit.
how much does your ba cost ranger?
January 13 2008 1:40 PM EST
I wonder what my average rewards would be if I were fighting people with 4M or more score.
January 13 2008 1:42 PM EST
It certainly looks to me that my rewards are the same as Sut's. Especially looking at the common opponents.
As one fights lower, one will certainly get lower rewards.
well with those figures you come out to about 16% higher rewards than me.
January 13 2008 1:43 PM EST
One has to look at common opponents to see if I really am getting higher rewards.
IE, if you look at Sut's example, he looks a bit smaller, but if you take common opponents, we are almost identical.
maybe we should all fight the same people, people i can beat, to rule out the fighting down gets lesser rewards issue? how low should we go to make sure i am fighting down also though?
January 13 2008 1:47 PM EST
Fight as high a score character you can and then post it.
Do it for about 4 characters. The 4 highest scores you can beat.
I will then fight the exact same ones and post my results.
ok, here goes with some of the top people on my fightlist. first & second numbers are the cash rewards, third number is my average cash using formula and third is then average xp using fomula. i did ten people and threw out the one that had the 1.5 multiplier for ease of use and included more than the four in case you get some 1.5 multipliers as well.
my total average xp for this set of opponents came out at 312. i am not sure why this is so far off from my first attempt, but i was getting people from all over my fightlist that time, instead of at the top only.
January 13 2008 2:27 PM EST
403 323 524.875
694 386 501.8
687 515 557.9
386 540 501.8
576 330 536.25
302 527 490.75
316 394 513.5
Average: 517$, which is 312 xp per battle.
January 13 2008 2:29 PM EST
It seems our rewards for common opponents are exactly the same.
I do better because I can beat higher score characters. Therefore better rewards about 16% as you figured.
well that surprises me and is not what i had thought at all. it is also going to change my fight list dramatically. i had been under the theory that it didn't matter endgame who you fought and was spreading out my clan fighting so as not to be to harsh on any one character.
i will now be cleaning my fightlist and farming the heck out of a select few it seems.
This result supports the theory that I had posited above. A simple test that you could run to convince yourself of that is one that I came across while running tourney characters a few months back.
Back then, I had a starting tourney character, so there was no challenge bonus. Bast has a character with 20 MPR and score, but the char had a big VB on it, so its PR was in the thousands. I got 4-5x as many rewards from this character as from a similar char with the same BA cost for myself and the same MPR and score.
A more complicated test would be doing the same thing for people at different PRs at high levels, but you have to have a good method for finding the true mean rewards for each normalized over the changing bonuses.
i always took the no challenge bonus to mean no challenge bonus. i wonder if we are dealing with challenge penalties then because it still seems to me like this is some sort of hidden challenge bonus.
Consider this model:
Each character has a "base" reward value based on its PR. Every time you win, you get this base value, modified by a random factor, then modified by the visible bonuses (clan, challenge, N*B).
"Remember that there is an unknown hidden bonus for fighting clan characters..." -- NightStrike
I just went through my fightlist, recorded average rewards from all my opponents for about 100 battles. I found Avoid's math to be pretty accurate for clan characters, but for my non-clan opponent, it gave me consistent numbers that were 1.5% low.
This "unknown hidden bonus" that you mention seems to be about 4.7%, based on my data. This was based on all opponents giving 100% challenge bonus and a 12% clan bonus. It may have some correlation to the clan bonus, rather than just being a flat 4.7%, so I think a lot more data with different variables needs to be collected.
This thread is closed to new posts.
However, you are welcome to reference it
from a new thread; link this with the html
<a href="/bboard/q-and-a-fetch-msg.tcl?msg_id=002Jsi">todd's unfair advantage</a>