1. It would have to solve the double-spending problem in a decentralized way (so that it has no single point of failure).
2. It would have to have desirable economic properties, like having a constant number of monetary units, and solving the problem of distributing the initial coins in a reasonably fair way.
2. It would have to show significant evidence of a dedicated and growing developer base (both for the core client and the endless applications).
4. It would have to be undervalued compared to its potential.
...and I actually have invested all of my savings into bitcoin.