I know next to nothing about the science of inventory management, but I have a feeling that you’re interested in numbers that people aren’t normally interested in. While it does make sense that you want to get some idea on what your yield is on the money you have tied up in inventory, someone analyzing a business cares much more about the yield on the total investment. That is, if you sometimes have the cash from a sale sitting around unused, that’s money isn’t working for you in the same way that inventory sitting on store shelves isn’t.
If you order new product once a week to get you to a 21-day projected supply, then you will probably always have capital equal to the cost of 21 days worth of stock tied up somehow. It’s possible that you put your income from sales into the overnight money-market and thus make a small return off your sales early in the week, but at the profit margins and turnover rates you’re citing that return probably isn’t relevant. So long as you order once a week to the same level, you will always have that level worth of capital invested in that product, regardless of how the sales of that product are spread throughout the week. Storage costs could complicate this slightly, but again, what does it matter the exact day an item is sold if you only make decisions on reordering once a week regardless? Thus the expected weekly yield on inventory is 100% * (expected profit per item * expected number sold per week)/(expected cost per item * expected number sold in 21 days). Since the 21-day supply is 3 times the number sold per week, this will simplify to (profit per item)/(3 * cost of item). This won’t be what your past return was, but what you expect to make in the future, which sounds like what you actually are interested in.
It sounds somewhat simplistic, but is based on ordering up to a 3 week supply once a week and not using your capital for anything else in the meantime; the only period that matters is the week. It’s possible I’m forgetting something important, but from what I see the only thing I’m not taking into account in intraweekly reinvestment which it sounds like is not going on. If you’re actually reinvesting the income throughout the week as sales happen then it complicates matters far beyond what anyone here is likely willing to do. Such a calculation would likely need to take into account the weekly cyclical purchase patterns to determine how much you would be reinvesting on average.