The standard way of completion with *nix shells and Emacs is to insert a prefix and then hitting TAB (or another completion key). Because this behavior has become second nature to a lot of Emacs users, Ido offers, in addition to the default substring matching method (see above), also the prefix matching method. The kind of matching is the only difference to the description of the substring matching above.
You can toggle prefix matching with C-p
For example, if you have two buffers 123456 and 123 then hitting 2 does not match because 2 is not a prefix in any of the buffer names.