View unanswered posts | View active topics It is currently Tue Feb 21, 2017 2:24 am



Welcome
Welcome to XP-Waste!

You are currently viewing our boards as a guest, which gives you limited access to view most discussions and access our other features. By joining our free community, you will have access to post topics, communicate privately with other members (PM), respond to polls, upload content, and access many other special features. In addition, registered members also see less advertisements. Registration is fast, simple, and absolutely free, so please, join our community today!


Reply to topic  [ 9 posts ] 
 Very exciting discovery - GTO-like boxing stats 
Author Message

Joined: Tue May 28, 2013 5:54 pm
Posts: 5
RSN: My Arena
Post Very exciting discovery - GTO-like boxing stats
I believe to have found the loss leader stats for the typical 1 vs 1 boxing stake for any CB level.

If you don't already know what GTO stands for, it stands for Game Theory Optimal. The GTO strategy is one that at least can not be exploited for any given set of strategies your opponent may use. For example, in Rock Paper Scissors, the GTO strategy is to choose at random (1/3 rock, 1/3 paper, 1/3 scissors). This is because any deviation from this strategy may be exploited.

Anyways, I have found the stat set for level 57 CB (W,X,Y,Z) that when paired against ANY other stat set of level 57 CB or lower, is at LEAST a 50% to win. Now, how I got here -

With a calculator that uses markov chains, I am capable of calculating the mathematically correct odds (within an extremely small margin of error) of winning vs any two stat sets. From here, I set a base level 57 stat set of (50 50 50 50), and try all other reasonable stat sets against that one to try to find the maximally exploitative level 57 stat set vs 50 50 50 50 (The stat set who wins the highest % against [50 50 50 50]. I then take this maximally exploitative stat set, and then try to find the maximally exploitative stat set against the maximally exploitative stat set of (50 50 50 50).

Now, at first I wasn't sure if overlap between stat sets was high enough to actually find a meaningful loss leader stat set. (Loss leader meaning the level 57 stat set with the least amount of cumulative loss against all other level 57 stat sets). But I was surprised. Not only after 3 maximally exploitative runs from the 50 50 50 50 did I find a stat set who was a favorite against any other 57 stat set (no loss), but the winning margins of that stat set were HUGE.

Against the stats (50 50 50 50), the maximally exploitative level 57 stats were a 72.96225% favorite over 50 50 50 50. However, the "GTO" stats were 72.87959% against 50 50 50 50. Not much overlap at all!
Against stats like 55 att 37 str 55 def 55 hp,(also 57 cb) which are more common among stakers, these stats hold up a surprising 59.06%. In fact, these stats are around a 60% favorite against most advantage stakers boxing stats right now.
These stats also do extremely well vs higher level stat sets. They seem to get the most out of matches against both higher and lower stat sets.

I plan on doing this for every CB level in boxes. I'm currently working on DDS / Whip stakes.


Thu May 30, 2013 10:53 pm
Profile
Tech Admin
User avatar

Joined: Thu Jul 28, 2011 12:32 am
Posts: 1652
RSN: Calisme
Post Re: Very exciting discovery - GTO-like boxing stats
What were the stats?

_________________
Image


Fri May 31, 2013 2:47 am
Profile
DGS Reviewer
DGS Reviewer

Joined: Sun Jun 19, 2011 4:54 pm
Posts: 736
RSN: 0ur Moon
Post Re: Very exciting discovery - GTO-like boxing stats
PMmed you.

_________________
Image


Fri May 31, 2013 8:51 am
Profile
Researcher
Researcher

Joined: Sat Apr 16, 2011 9:40 am
Posts: 1216
RSN: Meredith
Post Re: Very exciting discovery - GTO-like boxing stats
I get that if this was public, it would potentially ruin your income, but this really isn't the place to leave important details classified. You need to make enough information available so that someone can repeat the experiment to ensure its validity.

_________________


Fri May 31, 2013 11:07 am
Profile

Joined: Tue May 28, 2013 5:54 pm
Posts: 5
RSN: My Arena
Post Re: Very exciting discovery - GTO-like boxing stats
Before I release the stats, I want to try to make the account. If I find myself too lazy / not getting enough action, I'll release them. My post was just to inform that these stats exist / look for anyone else who has found them too.


Fri May 31, 2013 1:10 pm
Profile
DGS Reviewer
DGS Reviewer

Joined: Sun Jun 19, 2011 4:54 pm
Posts: 736
RSN: 0ur Moon
Post Re: Very exciting discovery - GTO-like boxing stats
I pmmed you asking for details about your trials. Can you respond to it?

_________________
Image


Fri May 31, 2013 1:23 pm
Profile

Joined: Sat Aug 27, 2011 3:41 am
Posts: 64
Location: California
RSN: Bryan Habana
Post Re: Very exciting discovery - GTO-like boxing stats
GTO HAS, AND ALWAYS WILL, STAND FOR GT OPEN

miss dem old days

_________________
Image


Wed Jun 05, 2013 5:59 am
Profile

Joined: Wed Jun 04, 2014 1:58 pm
Posts: 9
RSN: Classified
Post Re: Very exciting discovery - GTO-like boxing stats
I have to wonder what calculator you are using to calculate these statistics.

I am currently making a calculator since the most used one, on oldschool, is not optimal. It does not take into account the difference in max hits, for example 47 strength is exactly the same as 53 strength, but the calculator does not agree.

So, when you say 50 50 50 50, what you mean is 50 47 50 50, since the +3 strength levels make no difference. Therefore any calculations based on that theory, that 50>47 are wrong, and will mess up further calculations.


Wed Jun 04, 2014 2:06 pm
Profile

Joined: Tue May 28, 2013 5:54 pm
Posts: 5
RSN: My Arena
Post Re: Very exciting discovery - GTO-like boxing stats
ThinkingMan123 wrote:
I have to wonder what calculator you are using to calculate these statistics.

I am currently making a calculator since the most used one, on oldschool, is not optimal. It does not take into account the difference in max hits, for example 47 strength is exactly the same as 53 strength, but the calculator does not agree.

So, when you say 50 50 50 50, what you mean is 50 47 50 50, since the +3 strength levels make no difference. Therefore any calculations based on that theory, that 50>47 are wrong, and will mess up further calculations.


It's been a while since I last had interest in this calculator. I made a near optimal boxer (because optimal stats are a pain to get), and actually turned a nice profit. Nevertheless, I couldn't find enough stakes at my level and lost interest.

Yes, you are correct that 50 47 50 50 = 50 50 50 50. The calculator I have has the accurate max hit formula, and so it judges strength levels accurately. If any other osrs staking calculator doesn't correctly handle something as simple as the strength levels, like 53=/47 then they almost surely have a ton of other things wrong with it.

The optimal stat set for a combat of 55 is [45 atk 27 str 62 def 68hp] This stat set was calculated before random pid, (we assumed perfect atk/def switching), but it really shouldn't matter that much.


Fri Jun 13, 2014 7:15 am
Profile
Display posts from previous:  Sort by  
Reply to topic   [ 9 posts ] 

Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
suspicion-preferred