I agree with Dale that you're not going to get anywhere near sharp
enough a filter by making something out of twinlead. Since the signals
are coming from different directions, it might be more practical to make
a null in the antenna pattern in the direction of the interfering
signal. This is most likely to work if the desired station is reasonably
strong, so it doesn't have to be in the direction of the antenna's
maximum response. A two element phased array can be designed to achieve
a deep and fairly narrow null, and you get a deep null off the end of a
very symmetrical dipole or the side of a Yagi. If the directions are
fairly close, an antenna with wider beam, like a 2-element phased
cardiod pattern array or a dipole, would give you more of the desired
signal when the undesired one is in the null, than a Yagi would.

You might have to use some sort of feedline decoupling (common mode
choke, also known as a "current" or "choke" balun) to prevent the
feedline from being part of the antenna and reducing the depth of the
pattern null.

Roy Lewallen, W7EL