We're in agreement regarding spread trading, though you consider volatility of the asset (BTC) directly related and I view it as a separate factor. End result is the same.
10,000 US-NBT purchased and returned net $100 from spread for Nu. Say Bitcoin is at 1,200 USD at the time the NuBits enter circulation and 1,150 USD at the time they leave circulation via a BTC/USNBT value transfer pair.
10,000 / 1,200 = 8.3333
10,000 / 1,150 = 8.6957
8.3333 * 1,150 = 9,583
10,000 - 9,583 = 417
417 - 100 = 317
A customer buys 10,000 US-NBT for 8.3 BTC. Bitcoin falls 50 USD, then they return their 10,000 US-NBT and receive 8.7 BTC. That's a net loss for Nu of 317 USD.
NuBits' main purpose is to move that volatility exposure from customers to shareholders. Zero reserve would theoretically eliminate reserve assets and their volatility as a factor, though in practice there may be a lag between the backing pair and value transfer pair (such as NSR/BTC and BTC/NBT). Zero reserve requires a mature NSR market with investors who know what they're doing. Investors did it by proxy last year when they recovered the abandoned US-NBT peg. The money was provided. We want to automate that procedure with minimal friction.
For rational increase in value of NSR, the loss described from customers successfully hedging in NBT must be covered by equal failure to hedge, other income, or be resolved with zero reserve. The asset may also stabilize, but that seems unlikely.
Until then, using a reserve asset is the only way we know to do this. Using USD as reserve asset carries other risks not being decentralized.
Decentralized liquidity providers withdrawing funds in times of high volatility causes problems, but how critical? If Bitcoin stabilizes again, would decentralized liquidity providers return? That's an interesting topic I suggest breaking out of this thread and continue.