The 10/31/16 sell levels for systems 1 and 2 are 1162.17 on the R2000 and $19.37 on XLF. If not breached, system 1 will buy IWM at close, and system 2 will buy IWM and PTY 50/50. I will report the results in the comments.
I have previously expressed concern that the removal of equity real estate from the financial sector (and from XLF) on September 16 would mess up XLF as an indicator for the R2000. mREITs have not been removed; so that is good. My back-tests to 1987 on the system have been based on the Fidelity Select Financial Sector fund, and then on XLF when it was issued. Both of these funds included the real estate sector. Since I first started to follow investments in 1975, I have noticed that the real estate sector has been an excellent indicator for up-coming recessions, and therefore for bear markets. When the yield spread rises, that's good for banks and other financial institutions, but it's bad for real estate. Therefore, the financial sector, as previously defined, was an excellent indicator because both financial stocks and real estate should move together to signal a bull or bear market. When one moves opposite to the other it is time for caution. Real estate peaked in January 2007, incidentally, well before the financial stocks did.
So I have concocted my own index. As a measure of eREITs I am using ICF, the iShares eREIT. XLF will measure the financial sector as newly defined. My composite is calculated as New Index = .0119 times the ICF price times the XLF price. The .0119 is my estimate for adjusting the ICF price to the weight that real estate had in the former index. There are no dividend adjustments to the price of ICF. The daily highs and lows are simply based on the daily highs and lows of ICF and XLF, even though they are unlikely to occur at the same time of day. I am not sufficiently computer savvy to do running calculations of the new index during the day. I don't think it will be a major problem. Before computers, daily highs and lows of an index were calculated as if each component stock of an index reached their high at the same time, and as if the lows were all hit at the same time. So the reported highs and lows of the indexes were both over-estimated (more extreme) and did not actually occur. This will be the case with my new F index.
I will check the daily highs and lows and dispense with out-of-wack trades that typically occur in the first minute of trading before there is price discovery. That is likely for ICF, which does not have high trading volume, even though as an ETF it should stay close to its NAV. I have always done this for XLF, and discarded flash crash values as last occurred on 8/24/2015. These flash crash prices are still in the historical data, which must mess up many computerized back-tests.
Also, since I calculate the index after the close, buys and sells must be done the NEXT day. They will be done at the close of the next day. So I will report a new system, System 3, which will continue system 1 (it does not buy PTY), but follow the new index. And that's where it gets interesting. ICF peaked on August 1 at a close of $112.30. By September 16 it hit $103.46, rallied for a week to above $107, and then resumed its descent. It is currently at $97.15. My new F index hit a 55-day low, a sell on October 4. It has not come close to hitting a new 20-day high since.
Therefore, System 3 sold IWM on October 5 (the day after the signal) at the close for $123.97. I don't know what the 3.55 pm price was, but the R2000 didn't move much in the last half-hour on that day. It was an up day. Including the IWM dividend, the proceeds fetched $96,713, which is a big improvement over system 1. TLT closed at $134.64 on 10/5; so 718 shares were bought.
There is no point in providing advance buy or sell levels for my New Index. I will report when they are hit such that the signals can be executed the next day. I feel much more comfortable with this new system, although I will not re-buy TLT at this time. Rising yields on T-bonds are mostly inspired by rising yields in Europe, but also inflation is on the rise. If there is no increase in the CPI index for the next 3 months, YOY inflation will still come in at 2.07% when the December data is reported in January.
Meantime, the R2000 has risen in 25 of the last 29 Decembers, with an average increase of more than 3% in those 29 years. Furthermore, trading days 6-10 of December have declined on average, 17 out of 29 years. November has averaged 1% gains.
I will monitor the 3 systems for a couple of years, and see how they compare. I think that what the real estate sector tells us really matters.
Disclosure: I am/we are long AMJ PTY.