Your modification seems ok.
Note: BayesElo uses
pdra=xpect(dr+de)-xpect(dr-de)
for the drawratio where de=draw_elo.
So it does not need any kind of "switched" parameter.
It might be interesting to compare this formula to yours. I think it will
be pretty similar (no time right now to check).
Question about Ordo
Moderator: Ras