The following Tranche Recommendations use the following assumptions.

- Ten (10) ETFs plus SHY are used to create this data. The ETFs are identical to those used in the “Rutherford 10”.
- The number of Offset Portfolios is set to eight (8). Twelve (12) is the maximum number possible.
- Periods between Offsets = 2.
- ROC1 = 30% and ROC2 = 50% with Volatility set to 20%. These are the percentages recommended from numerous back-tests.

**Tranche Recommendations:** If the offset portfolios were set to one instead of eight, the portfolio would be divided into thirds where 1/3 goes to SHY, 1/3 to TLT, and 1/3 to PCY. With the offsets set to eight (8) we buy 1000 shares of SHY, 80 shares of TLT, and 150 shares of PCY. Those orders would be placed *today*, not spread out over some future period. When possible, I prefer to buy in round lots so I would set an order for 100 shares of TLT, 200 shares of PCY, and then reduce the number of shares of SHY to a round lot depending on available cash.

At this point there are two options. We can wait another 33 days until the next review period or we can watch the changes on a daily basis. I prefer to wait so as to avoid unnecessary portfolio churning and possible whipsaws. All three suggested ETFs are “conservative” so I would watch to see if any equity ETFs moved from below to above SHY in performance. Say VNQ made a positive move and showed up performing better than SHY. At this point I would likely sell off sufficient shares of SHY and buy 100 shares of VNQ. For this to happen, I want to see a positive absolute acceleration and a positive “Golden Cross” from VNQ. Right now, all conditions to buy VNQ are in place with exception of outperforming SHY. VNQ needs to move up a few slots in order to become a favored ETF.

This is an additional help session in the use of the Tranche 1.6 spreadsheet.

Richard Dougherty says

Lowell,

I noticed that you and HedgeHunter have different parameter settings for ROC1, ROC2 and Volatility. Your tranche work sheet shows the weight and look back period to be the following: ROC1=60 trading days and 30% weight, ROC2=100 trading days and 50% weight, Volatility+ 14 trading days and 20%.

In HedgeHunter’s Rutherford post dated October 5, he uses the following weight and look back periods: ROC1=63 trading days and 50% weight, ROC2=126 trading days and 30%, Volatility= 14 trading days and 20%. I know you and HedgeHunter have done a lot of work optimizing these parameters. My question is will it make a difference which set of weights and look back periods are used ?

Thanks,

Richard

Lowell Herr says

Richard,

It looks like we may need further discussion as to what is a “robust” look-back period and what weights should be assigned to those periods. Before going further, be sure to use calendar days when working with the Kipling 8.0 and trading days when updating the Tranche 1.6 spreadsheet.

In many of the back-tests run this summer, if I recall correctly, the 30% weight for ROC1 or the shorter term look back and 50% for ROC2 ended with better returns and lower volatility. I’ll need confirmation from HedgeHunter, Ernie, and Herb on that point.

In looking back over the studies, I think this is where I came up with the ROC1 = 30% and ROC2 = 50% weights.

“An important thing to note is that the heat map is a summary of data for ROC1 weighted at 50% and ROC2 weighted at 30%. Therefore, choosing an ROC1 value of 100 days and an ROC2 value of 60 days tells us that the longer term momentum look-back period should be weighted heavier than the shorter term momentum period in the Ranking spreadsheet (current default is 50% x ROC1, 30% x ROC2). Note that the default settings are not bad – still generating return/risk values in the top two deciles – but reversing these weightings would appear to provide more robustness (less impact of being slightly off in optimal look-back periods).”

Again, we need more discussion on this point.

Lowell

Herbert Haynes says

Richard,

This is explained very well in HedgeHunter’s post entitled “How ‘Robust’ is the ITA Wealth Momentum Strategy? – Part 2” and the heat map that appears there. A definition of “robust” is included in that paper.

This and other heat maps showed two broad areas of “good” performance and HedgeHunter simply chose a point near the center of one of those areas. This is the area to the left and below the diagonal and the point chosen was ROC1=100 and ROC2=60. But in these back-tests, ROC1 had a weight of 50% and ROC2 had a weight of 30%. In order to keep ROC1 as the shorter look-back period, the two period lengths were reversed, along with their weights, this giving ROC1=60 trading days with a weight of 30% and ROC2=100 trading days with a weight of 50%. The weight of volatility was left at 20%, but the recommended look-back period was 13 or 14 trading days.

The other broad area of good performance was to the right and above the diagonal. If a point in that area had been chosen as the “best,” it would have coincided very nearly with the original period lengths and weights of ROC1 and ROC2 that HedgeHunter had chosen long ago and that were used in the 7.1.3 spreadsheet and its predecessors.

Recently, HedgeHunter has chosen to revert back to his original recommendations, while Lowell is using the back-testing recommendations. However, Lowell has suggested that the weights of ROC1 and ROC2 should be reversed (at least for the time being) in order to take advantage of the anticipated improvement in the market, thus making Lowell’s recommendations very nearly match those of HedgeHunter.

Regardless of which of these sets of look-backs and weights is being used, all can be considered to be valid and “robust.” It is up to the individual investor to determine which set she/he prefers to use.

Herb

Richard Dougherty says

Herb,

Thanks very much for a thorough explanation.

Richard