As I understand, Fisher-Yates algorithm shuffles data without replacement. I generated a couple of shuffled price curves with priceClose() and compared with original curve. I sorted and compared percentage returns (p0/p1-1) and price difference (p0-p1). They are different. I wonder what is shuffled then? I assumed the returns should be the same but shuffled? Mean of returns is different as well.

I want to run the Monte Carlo analysis: generate simulated price curves and run my algo on them to reject the null hypothesis. As I understand, Zorro provides 2 options out-of-the-box: Detrend=SHUFFLE (according to the updated manual it's randomize(SHUFFLE,...) ) and randomize(BOOTSTRAP). The difference is that the first one is shuffling without replacement (used for montecarlo in the Black Book example), the second one is shuffling with replacement (used for Whit Reality Check). Which method can be used for above-mentioned montecarlo analysis? Which one is preferable? What's the difference?