I think any significant inaccuracies there may be more of a symptom of lack of inter-play, then a decision to overweight (or underweight) blowouts. If in a hypothetical higher league, games are often 1 point affairs, while in a lower league, there are often blowouts - within each of the leagues themselves, relative ranking should remain accurate. If a high scoring team in the lower league is 6 goals better than the lower scoring teams in that same league, they should have a ranking that is 6 goals better. And if a high scoring team in the higher league is only 3 goals better than the lower scoring teams in that same league, they should have a ranking that is 3 goals better.
Now if none of the higher league ever play any of the lower league in any recorded games or tournaments, it becomes a definition problem. Do you set the top league starting point 3 points higher to start? 6 points higher? 20 points higher? To establish the first reference point, it's probably a bit of finger held up in the wind. After only 5 or 6 games within that league, there is plenty of info to rate the individual teams against each other, even if there is zero history prior of any of them. But if there is no relative history of those teams playing with any of the teams in the other league, it can be stuck at whatever initial rating was assumed. And in this hypothetical construct - a very high scoring team in the lower league might be expected to show a rating higher than an average team in the higher league.
So it becomes a bit of a self fulfilling prophecy, where leagues that are completely separated from play from other leagues, can't objectively and mathematically be ranked against each other - simply because there aren't enough events to point to where teams from the leagues are compared (via actual games) with each other. However - every time that even a single team within a closed league does go outside to play a tournament or similar against outside teams, that result is then applicable back to the source league, to help balance and normalize the rankings of even teams that never go outside for opponents. It appeared that one good example of this were the teams in Alaska earlier this year. The ratings of some of them were uncomfortably high. They tend to only play each other (it's Alaska!) so there is very little inter-play outside. But looking through them all recently, it seems there were enough tournaments that pulled just enough of them out to play outside, that the ratings of the entire state appear much more normalized with what one might expect.
With all of this, it becomes a bit intuitive that national ratings numbers between a team in Dubuque, a team in Boston, and a team in Seattle, may be a bit off - if none of them every play each other, and they don't play anyone else who plays them, and they don't play anyone else who plays them who plays them, etc. It's like trying to compare AAA baseball here in the US to the 2nd level baseball league in Japan. You can try - but without direct play, it's going to be mostly conjecture no matter how you try to do it. But - the relative rankings in the case of this app are necessarily much more relevant when you're using them to compare against teams in your area or conference who you actually do play, or will play, or have played.
