Relative ELO vs. expected win/loss ratio?

    Professor Arpad Imre Elo did explain his rating system in great detail in his book "The Rating of Chess Players Past and Present". It is out of print but used copies are readily available online. Warning: lots of complicated math is involved.

In the name of science:

I copied result (first page = 20 games) from top 10 players at chess.com ranking-list (rapid).

After filtering all games with rating diff > 400 (174 games) I got the best expected result (lowest std dev) with the factor 353 (instead of original formula 400).

Expected result = 1 ( 1 + 10^((Ra-Rb)/353)))

I think I need more data in order to really say exactly what factor to use for the most accurate prediction on this level of play, but it seems like the 400 factor is rather good. Which maybe is what to expect when the ranking uses this formula, I guess it is sort of "self-tuning"?

Although it is not presented as The Truth, a mathematical justification is provided underneath it so that you can determine whether the logic persuades you.

https://9sleek.com/products/seo-link-building-service

Where is the flaw if thinking from this other angle:
Take 2 games with black and 2 games with white:
50% expectation means that most probable outcomes with 50% have 6 options:
Win & Loss
Loss & Win 
Tie & Loss
Tie & Win
Loss & Tie
Win & Tie
Tie & Tie
Tie & Tie
Total points: 2 0 0 2 1 0 1 2 0 1 2 1 1 1 1 1 = 16 from max of 32.
And 25% means that Wins and Ties come 2x less than before:
Because if opponent is 2x stronger then you manage to get Tie 2x less.
Win & Loss is not 2 & 0 but 1 & 0
Loss & Win etc
Tie & Loss 1/2 0
Tie & Win 1/2 2/2
Loss & Tie 0 0,5
Win & Tie 2/2 1/2
Tie & Tie 0,5 0,5
Tie & Tie 0,5 0,5
1 0 0,5 0 0,5 1 0 0,5 1 0,5 0,5 0,5 0,5 0,5 = 7 points of 32 max points.
Funny and seems wrong but why. Perhaps as the Wins wont turn into Losses, but may turn into Ties too.
In practies ties come much more often.
Total outcome is 25% of each combinations outcome is 25% is not same thing?