For many applications (e.g. "social" coins, distributed exchanges, anonymity-focused coins) there are currently multiple, competing crypto coin or tokens.
I generally believe it is a bad idea to buy the smaller pair of any related/similar/forked crypto pair.
* They could quickly lose relevance if one of them "takes over" the respective market.
* They are generally those with less dev funding and easier to completely nuke themselves out (due to bad Press, bugs, failure to complete milestones).
* In extreme cases, low volume / impending delisting from important exchanges (note how my analysis only works that well for poloniex) completely overshadows these patterns.
One exception for still trading these:
If pairs have been observed to behave the same way for quite some while (bad: measured in time, better: measured in number of BTCUSD breakouts), then the decision is much easier.
I am watching the following sample coins/tokens to test my idea:
ETC (extensive script language), Bitshares (distributed exchange), Synero (social), ZCash (anonymity), RubyCoin/Lisk (alternate implementations, though possibly ETH already killed that unique selling point)