Shifting the mu value ($\mu$) has no effect on the rating system since it's just an arbitrary number. Shifting it will only shift the mean. The system starts with an assumption of $\sigma = \frac{\mu}{3}$. To take the default values as an example, at $\mu = 25$ and $\sigma = \frac{25}{3}$ implies the system has ascertained the actual skill of the player could be anywhere within 3ドル$ standard deviations of the mean. In other words, the system has no idea about the skill of the player. Don't compare it to Elo, since it does not have a variance parameter.

So one of the issues we have is that when we want to create a balanced team of players, the new players are severely overrated causing the teams to be unbalanced. Is this a problem that has come up?