Better than Elo, Glicko and Truskill

Elo, Glicko, and Trueskill ratings in essence they count wins-losses, as in average you earn 8 points for a win and you lose 8 points for a loss. This is wrong since e.g. a team or a player that has 20 wins in 20 games is better than a team or a player that has 40 wins and 20 losses in 60 games. And e.g., one player has 1200 wins and 1000 losses in 2200 games, and another player has 200 wins in 200 games. They both have wins-losses=200, but the second player is obviously much stronger than the first. Therefore the correct rating is not wins-losses but (wins-losses)/(number of games) which is equivalent to ((wins)+0.5(draws))/(number of games). But this is also wrong because a team or a player that has (1+0.5*0)/1=1 is almost surely much more weaker than a team or a player that has (80+0.5*0)/100=0.8. Therefore at least 2 modifications to the ((wins)+0.5(draws))/(number of games) formula are needed, one that corrects this mistake, and another one that takes in account how strong were the opponents that the team or player faced. I have found such modifications but I find no point to post them unless administration shows some interest, and I should get paid if they apply them instead of the Glicko formula.

Hmm.  The current formulae are not close to what you cite.   I am sure they take the strength or weakness of opponents into account and also judge how current a rating is.

Site CEO Erik wrote a detailed explanation of the rating system used here.

https://www.chess.com/article/view/chess-ratings---how-they-work

He talks about the Elo and Glicko systems. Perhaps you could benefit from reviewing it.

I don't believe that Mr. Glickman earns anything from his system directly.  The Wikipedia page says the Glicko System formulae are in the public domain. 

At any rate, it seems unlikely anyone will pay you unless you demonstrate the clear superiority of your system. Mr. Glickman published his formulae in peer reviewed academic research. That is probably the best approach.

Elo and Glicko should get paid if they did not get paid. And now that you mentioned that Glicko  did not get paid, I changed my mind and I will not reveal my solutions unless I get paid first. Actually, I revealed too much. My price is 20000 euros to reveal them, and if they need any help to apply them, then this means that they hire me with a salary of additional 3000 euros per month. 

Yeah, you're not going to get paid for a rating system. 

There is no point to post my solutions even for free if you do not realize that I already proved the superiority of  ((wins)-(losses))/(number of games)<=> ((wins)+0.5(draws))/(games) in comparison with Elo and the like. 

Proof of superiority of a system comes from publication of results. Put a bunches of old players and events through your method and see if it produces results in line with actual performance of players.

It's mathematics; stating it's better isn't the way it works,

There is no need to test it, it is concluded by reason. First you need to realize that Elo etc in essence they count wins-losses. And then to realize that ((wins)-(losses))/(number of games)=(W-L)/G is a far better rating than wins-losses. You objected that Elo etc do not in essence count wins-losses because they take in account how strong each opponent was. But ((wins)-(losses))/(number of games) also has a modification that takes in account how strong each opponent was, which modification is faithflul-corresponing to Elo's solution and it is NOT Elo's solution. OF COURSE it is not, as Elo etc in essence they count wins-losses!

P.S.  I just found out that perhaps there isn't a modification of (W-L)/G that takes in account how strong each opponent was, that is faithful to Elo.

When you say one will not get paid for a better rating system is like you are saying administration do not care to adopt a better rating system. Which is not the case, as then why did they adopt Glicko instead of Elo. Therefore, you should bring this thread to the attention of whoever decided Glicko to be adopted, and whether I will get paid or not is a matter between me and him IF he realizes the superiority of (W-L)/G<=>(W+0.5D)/G. Bookmakers will immediately realize it , and if they use Elo or the like for some sports, they will be very interested to apply my unknown solutions, so I guess I will write to them if nothing happens in chess room forums. They will immediatelly realize it also because they are familiar with estimating probabilities with goal average and Poisson. And to know that a team scored A goals and received B goals is a useless information, and only the goals PER GAME gives an answer. Similarly, W-L is a useless information and only (W-L)/G gives an answer, as it is like (A-B)/G. (A-B)/G might not give an answer and  A/G and B/G in a combination with the goals per game of the opponent is needed for Poisson, but you get my point.

Submit your claim in scientific research.  Get peers approval. Apply for patent and tell us when you get medal or $20,000.

I feel like I am not educated about rating systems to really say anything but I feel like your example doesn't fit well for a reason to replace Elo/Glicko ratings. When a player has 20 wins/20 games, and faces a player that has a 40 wins/60 games, of course the rating of the player with 20/20 has a higher rating but when facing the person with 40/60 they will most likely lose and that's when the point of these rating systems appear. The 20/20 player will lose soon enough and lose rating points and then after more games making a larger sample, it will even out and soon the player will be at his/her "correct" rating and go on to try to bring it up. I feel like that is the point of rating systems, to show how a player is doing in a win/draw/loss ratio. An average of 8 points change per game is still unlikely to newer players. 20 games isn't much and will likely even out quickly. If the 20/20 player keeps winning then it shows that he/she is not in the correct rating level and sooner or later the 20/20 player will be placed in its "correct" rating level. 

Again, a mathematically sound rating system, that has been published, peer reviewed, and tested extensively both experimentally and in practice, isn't going to be replaced by something that isn't any of those things.

I won't say there isn't a better rating system that could be used or developed. Just that the way they get implemented is through being proven they produce better results. If there is a proven better system, sites will use it. But I still don't think anyone will pay for it.

Statistics isn't something I'm great at, but Mark Glickman is and his system is statistically valid and you can easily read up on it and see what the formulas are doing (which is much more involved than how you're presenting it).

When and if administration reads and realizes the truth of my first post, then they will ask for my solutions. And if I reveal them to them, then the matter will be how to translate the Glicko rating each player gained, into my rating, otherwise all the games everyone played would be like never happened and each player should start over i.e. attaining his new rating by playing new games. Well, I found a translation (according to a solution that would be applied if my rating was used from the beginning), but if my solution is wrong enough, then (until someone finds and notifies a correct enough solution) the only solution is to start over and every player to be forced to face opponents from all ranges of ratings by random selection. Since, the (wins+0.5*losses)/(number of games) record of each player would be his correct rating (with a modification that corrects the e.g. (1+0.5*0)/1=1 to a more true rating) only if he had faced opponents from all ranges of ratings by random selection. And since he did not, a solution is needed that takes in account how strong were the opponents that he faced, and how strong were the opponents of the opponents that he faced, and so on!

Trueskill has been present since 2005, (16 years ago) and at that time OP might be peeing in his diapers.

And suddently why you talk about those? (From online chit chat from friends?)

https://imgur.com/McUT3Mo

Trueskill is patent of microsoft and it is created for for Free for All ranking games mixed team ability among 1v1 and 5v5 .

www.microsoft.com/en-us/research/project/trueskill-ranking-system/
chrome-extension://oemmndcbldboiebfnladdacbdfmadadm/www.researchgate.net/profile/Arman-Dehpanah/publication/343710894_The_Evaluation_of_Rating_Systems_in_Online_Free-for-

All_Games/links/5f3fd5f192851cd302107ec7/The-Evaluation-of-Rating-Systems-in-Online-Free-for-All-Games.pdf?origin=publication_detail

It is noting to do with chess, where it is 100% 1v1.

I just discovered that Elo and Glicko do not in essence count wins-losses, but:

Suppose a player chooses to play only against players rated 1500, and wins all games. If the rating was 1500+7((wins)-(losses)), the more games he plays the higher rating he gets. This is a wrong inflation, as his rating when he has played 10000 games should be very close to his rating when he has played 1000 games, and not (1500+7(10000-0))/(1500+7(1000-0))≈8.4 times higher. Does Elo create such an inflation? To answer this, you need the Elo formula

f(n)=f(n-1)+14(1-1/(10^-((f(n-1)-1500)/400)+1)), f(0)=1500<=>

f(n)=f(n-1)+14(1-1/(10^((1500-f(n-1))/400)+1)),f(0)=1500

where f(n) the rating after the current game and f(n-1) the rating after the current game. If f(10000) is higher enough than f(1000) then Elo has a wrong inflation. Unfortunately, calculators cannot calculate f(10000) and f(1000). So the question remains. However, one can see a similar formula

f(n)=f(n-1)+14(1-(1-0.5/(0.01(f(n-1)-1500)+1))), f(0)=1500

which

www.calcul.com/show/calculator/recursive?

can calculate it, and gives

f(1000) = 2588.1687720803807

f(10000) = 5143.33386212574

here THERE IS a wrong inflation. So, can anyone calculate whether Elo and/or Glicko have a wrong inflation? Whereas, the (wins+0.5draws)/(number of games) with the modifications of it that I discovered, surely do not have a wrong inflation, and beautifully rate the player with a range between 0 and 1 (instead of the nerve breaking 1500+-), and when one has rating b it means that his “expected score” against a player rated 0.5, approaches b when the number of games that he attained b, tends to infinity. E.g., at soccer one team in the last championship has (37+0.5*0)/37=1 and in the last game of the championship plays against a team which has rating (15+0.5*7)/37=0.5. It has not 100% probability that it will win, but from some data I gathered, it has 0.9105 expected score. If this strong team was known that it had (1000*0.5*0)/1000 (in games in parallel universes, or in our universe if it is a chess player), then the unknown data would say that it has ~0.99 expected score.

And yet you've been claiming that your rating system is better than either of those, when you don't even really understand how those ratings systems work: that seems like a pretty good indication that no-one should take your claims seriously. Which is why if you want it to be taken seriously, you need to subject it to serious peer review and analysis. Or see if you can successfully apply for a patent for your method.

Elo and Glicko rating must be replaced with the “expected score” against an average strength player e.g. that Sonas has estimated:

https://en.chessbase.com//post/sonas-overall-review-of-the-fide-rating-system-220813/37

which of course is the TRUE rating. But this has been already done!: e.g. Sonas table shows that the expected score for 100 points rating difference is 0.6, so the expected score of a player rated 1542+100=1642 against the average player of 1542, is 0.6. Similarly, it had been done before Sonas research, with the relation between the Elo rating difference, and the corresponding expected score, that Elo claimed. Yes, but the new better rating would get rid of the 1500+- rating and say how to find the “expected score” (wins+0.5draws)/(number of games)=(w+0.5d)/g against an average strength player, using only his past score=(w+0.5d)/g and the past score=(w+0.5d)/g of the opponents that he faced. THIS solution should be used instead of Elo and Glicko. Have I found it? Yes, and if my solution is not correct enough, then SOMEONE must find the correct enough one. Elo found a solution (infinitely better than the none solution) to locate this expected score against all players, but I most probably located e.g. wrong inflation in his solution. And the 1500+- is nerve breaking, why wasn’t it 0+-, it would be more obvious how correct it is.

As far as I know, and I don't know much, you can only profit from an execution, not from the immaterial concept itself.

Let's say I invented ambidextrous sports (I did, in my mind). Tennis. Right hand vs. right hand, left vs. left. Add up the score. Horse polo. First half of the time, right handed, second half, left handed.

I can only profit if I run such events. I cannot stop somebody from copying that idea and make their own event.

A closer example.

UTR is a rip off of Elo. It says so:

UTR Powered by Oracle is a modified elo rating system . . .

https://support.universaltennis.com/support/solutions/articles/9000151830-understanding-the-algorithm-complete-summary

UTR makes money by recording and calculating ratings, basically using Elo. It can't stop somebody else from doing the same.

You can only own an execution. You cannot own the concept.

There is Miss America and there is Miss USA. You can't make Miss USA pay royalty to Miss America.

Water freezes at 0° and boils at 100°. Celsius would be making bank.