Calculate the returns from buying/selling one unit of stocks and one unit of bonds (in the opposite direction, naturally) at the close a week before month end, and then taking the position off at the close on the last trading day of the month. In addition to the returns calculated by the spreadsheet, I have also tracked how you would have done with “market on close” orders in the SPY and TLT ETFs. Note that TLT is obviously more volatile than the overall government bond index. As you can see, the out of sample performance has generally been pretty good, albeit with a hefty increase in realized volatility since the coronacrisis. It clearly doesn’t work every month, but both the spreadsheet model and the SPY/TLT iteration have generated annualized returns of some 10% on an out-of-sample basis. Going further back, the trailing one year returns have rarely been negative over the past couple of decades. As you can see, the out of sample performance has generally been pretty good.