I think basing sells on average share value is a great alternative rule that someone might want to consider in individualizing the system.
In my own trading model, I don't track stats such as average price for each stock, because I only look at return at the portfolio level. I just rebalance each stock once a year.
The market doesn't know and is not affected by how much you have invested in the stock or what your average return is.
For example, if you bought 200 shares of a $10 stock, and it now went down to $5, you will have to add another 200 shares to get back to a constant value of $2000.
Now, you have $3000 invested in the position, but I would not track this, because it is reflected in the portfolio's return stats. Instead, I would focus on that now I have a $2000 position at $5. In terms of risk and potential reward going forward, I am equivalent to someone who just initiated a $2000 position at $5.
So, I don't think "If I rebalance, I have a sunken cost of $3000 in stock XYZ." Instead, I want to evaluate it going forward as a new position - "Do I want to invest $2000 today in stock XYZ at $5?"
Even if you make a sale below your average cost, you are making a LIFO profit.
