In the worst case (1 tx per channel) LN would actually double the # of tx.

I am really looking forward to LN being really tested, but the thing that worries me is that the security model relies on txing before a certain block. If we see a large amount of the traffic move from on chain to LN then we could get ourselves into a situation where there is a run on the chain to close channels. If there is a large market maker that attempts to steal from a channel that is being used to route many others we could see a mass closing of channels which could push the fees high enough to make it cost prohibitive.

LN isn't a panacea it actually changes the security model.

It's not deus ex machina, because it's not unexpected. Lightning has a working alpha on Bitcoin, and Ethereum has something similar with Raiden, plus the Plasma paper and early code, a recently-published spec for the first version of sharding with a 100x throughput improvement, various special-purpose off-chain solutions like FunFair's fate channels, TrueBit for verifiable computation off chain, a simple multicore hack, and simulation code for full proof of stake.

I'm not saying there won't be a market correction, since they happen on a regular basis. Just a few months ago Ethereum dropped from $420 to $130.

Yes I am a long term believer in the technology, my gut just says the next correction that comes will be a bloodbath. Accordingly, I'd prefer to wait until that occurs before getting more involved.

If I'm wrong and "lose money" (ie don't get the gains) I'm okay with that too, since I've thought about this a fair amount and I'm confident my position is the most reasonable at this time.

Yeah that's probably a good idea.