YSR is obviously better. The two big issues it has are:
1. Sometimes it confuses teams and puts results from Team A for Team B and vice versa.
2. Its ranking isn't actually based on who beat who. It is based on goal differential between teams. So, if your team beat a team ranked 300, which has a score of 20, by 19 points, YSR would think your team should have 39 points. (YSR goes on averages, so this would actually occur if you play the bottom 10 teams and beat them by an average of 19 points). So, it can kind of reward teams for racking up points in a less competitive schedule.
Ultimately, if a team plays an equal number of teams better than it and teams worse than it, it will even out, but that is not always the case. Also, the second point above highlights a lesser issue, which is that YSR can be fairly accurate as far as general ranking, but not as to the point-spread predictions. A team with two points higher will, at the younger ages, usually beat the lower team by 4-5 points.
As an aside, I would let the site's webmaster know that they could save a lot of time by not using U-## designations instead of just birth year. They could have saved a lot of time and effort by using the birth year as the only identifier because currently they have to move everything around when u15 becomes u16. (They have been transitioning recently, and it is screwy.)