A better argument would be the evidence the hash rate is going through the roof. The inertia of those miners won't be stopped by bad sentiment caused by even worse software venerabilities. You either believe in the idea of what crypto currency technologies bring us, or you don't. Invest accordingly and leave your fears at the door.
When the price of electricity is higher than the mining payout, the hash rate will again plummet. Even though hashing with ASICs is much more efficient than CPUs or GPUs, it still takes electricity. And if the transaction rate remains low, the payouts are going to be smaller and smaller.
Mining capacity should act like a utility company that spins up and down power stations in response to demand for electricity. Mining slows until transactions payments go up enough to make it worthwhile. (And that doesn't even cover the fact that mining is still a lottery for the pools).
Change in hash rate only gives you a loose indication of what the miners guesses the price development will be, assuming you know their cost base well enough (i.e. there are no ASIC developments that people have been keeping quiet).
Also, any strongly positive or negative correlations to the price movements in other financial instruments.
It hardly matters that Bitcoin spikes and crashes its price every so often if you can't move your other investments around to profit from it. For this reason, every hedge fund manager on the planet is constantly looking for strongly negatively correlated prices that no one else knows about yet, or paying quants to invent instruments that show such behavior so that they have the information advantage until the other guys' quants decipher the formulas.
The graph that shows price vs time for a single asset is not a great thing to look at in isolation if your goal is to make piles of money on it.
Something shared by almost every hyped and now defunct securities out there.
Upside potential should not be the only factor for selecting an investment.
Risk management => Downside risk, liquidity, momentum etc. are way more important than limited past performance.