In addition to the "noise" created by delaying a day or so to make an investment decision, we also have the issue of "intra-day noise" or the variations created by the price fluctuation within a normal day of trading. If I understand back-testing, prices are fixed using the end-of-day value while an alternative model would use the average between high and low prices of the day.

As Mr. Picerno points out, we should dismiss or at least question back-tests that come from a single sample. In addition to questioning single sample results, what about published results that are given to the nearest hundredth decimal place. Now that is laughable.

Lowell

http://itawealth.com ]]>

In addition to the "noise" created by delaying a day or so to make an investment decision, we also have the issue of "intra-day noise" or the variations created by the price fluctuation within a normal day of trading. If I understand back-testing, prices are fixed using the end-of-day value while an alternative model would use the average between high and low prices of the day.

As Mr. Picerno points out, we should dismiss or at least question back-tests that come from a single sample. In addition to questioning single sample results, what about published results that are given to the nearest hundredth decimal place. Now that is laughable.

Lowell

http://itawealth.com ]]>

While I do not use leverage, options, etc., one investor and author on my blog does use options with a portfolio called the Hawking. He also will use options with his Rutherford portfolio.

I have not been successful using inverse ETFs so I gave that up. This is not to say someone with more skill could not hedge a portfolios such as that described above.

Lowell

http://itawealth.com]]>

While I do not use leverage, options, etc., one investor and author on my blog does use options with a portfolio called the Hawking. He also will use options with his Rutherford portfolio.

I have not been successful using inverse ETFs so I gave that up. This is not to say someone with more skill could not hedge a portfolios such as that described above.

Lowell

http://itawealth.com]]>

A "noise" factor was introduced in the back-tests that continue to be investigated over at ITA. As investors, we rarely make all our portfolio adjustments on the specified date of the portfolio review. For example, if we set limit or Trailing Stop Loss Orders, those orders are likely to trade a few days after the review date. For this reason, a "noise" factor of -2 to 5 days is introduced in the Monte Carlo analysis. What this means is that for some runs the trades are made two days before scheduled and other times the trades are made up to five days after the scheduled review. These variations have a major impact on the final performance results.

These are just some of the reasons why I am not satisfied with one shot analysis runs.

Lowell

http://itawealth.com]]>

A "noise" factor was introduced in the back-tests that continue to be investigated over at ITA. As investors, we rarely make all our portfolio adjustments on the specified date of the portfolio review. For example, if we set limit or Trailing Stop Loss Orders, those orders are likely to trade a few days after the review date. For this reason, a "noise" factor of -2 to 5 days is introduced in the Monte Carlo analysis. What this means is that for some runs the trades are made two days before scheduled and other times the trades are made up to five days after the scheduled review. These variations have a major impact on the final performance results.

These are just some of the reasons why I am not satisfied with one shot analysis runs.

Lowell

http://itawealth.com]]>

http://tinyurl.com/pl5...

Lowell

http://itawealth.com]]>

http://tinyurl.com/pl5...

Lowell

http://itawealth.com]]>

Lowell]]>

Lowell]]>

If you are following the "Robust" series of articles (http://itawealth.com) you will see that a one time back-test can be misleading. However, the general trend is positive. That gives the investor the edge, even if the final results vary.

Lowell]]>

If you are following the "Robust" series of articles (http://itawealth.com) you will see that a one time back-test can be misleading. However, the general trend is positive. That gives the investor the edge, even if the final results vary.

Lowell]]>

Yes it is, but there are differences. The review period is different, but the major difference is the look-back period. Antonacci uses a 12-month look-back, but we found that is too long. Granted, we did not use 40 years of data for back-testing as our securities did not go back that far.

Lowell

http://itawealth.com]]>

Yes it is, but there are differences. The review period is different, but the major difference is the look-back period. Antonacci uses a 12-month look-back, but we found that is too long. Granted, we did not use 40 years of data for back-testing as our securities did not go back that far.

Lowell

http://itawealth.com]]>

Lowell

http://itawealth.com]]>

Lowell

http://itawealth.com]]>

The 33-day review period will catch most bear markets as bear markets take time to develop. The spreadsheet has other early warning signals such as the "Golden Cross." This is when the 13-Day Exponential Moving Average moves from above to below its 49-Day EMA.

I sent you the information on SS access.

Lowell

http://itawealth.com]]>

The 33-day review period will catch most bear markets as bear markets take time to develop. The spreadsheet has other early warning signals such as the "Golden Cross." This is when the 13-Day Exponential Moving Average moves from above to below its 49-Day EMA.

I sent you the information on SS access.

Lowell

http://itawealth.com]]>

http://bit.ly/1Nb3Mxd

Lowell]]>

http://bit.ly/1Nb3Mxd

Lowell]]>

How true as to correlations going to 1 in bear markets such as we experienced in 2008 and early 2009. When you see the graphs presented on the above approach you will see this model helped to avoid the cratering of the portfolio.

Most of the testing took place on a portfolio made up of 10 ETFs plus SHY. We used the Rutherford 10 for most of the back-testing. If you search "Rutherford 10" you will see a lot of references to this portfolio - a live portfolio managed by HedgeHunter.

Lowell]]>

How true as to correlations going to 1 in bear markets such as we experienced in 2008 and early 2009. When you see the graphs presented on the above approach you will see this model helped to avoid the cratering of the portfolio.

Most of the testing took place on a portfolio made up of 10 ETFs plus SHY. We used the Rutherford 10 for most of the back-testing. If you search "Rutherford 10" you will see a lot of references to this portfolio - a live portfolio managed by HedgeHunter.

Lowell]]>

Lowell]]>

Lowell]]>

There are different Exponential Moving Averages measured as well as the "Golden Cross." These are explained in more detail on the blog.

If you want more information I suggest you check out the 7.1.3 spreadsheet, soon to become 8.0.

Lowell]]>

There are different Exponential Moving Averages measured as well as the "Golden Cross." These are explained in more detail on the blog.

If you want more information I suggest you check out the 7.1.3 spreadsheet, soon to become 8.0.

Lowell]]>

I'll send you the information in a moment. To answer the Canada question, while I've not sought out Canadian ETFs, they must be available. Here is what to look for.

Find an ETF that covers the entire equity spectrum of Canada. Next, find an ETF that closely matches VEU as VEU not only includes developed, but emerging market equities as well. Third, look for an intermediate to long-term bond ETF. The last ingredient in this mixture is to find a low volatile treasury ETF. Again, look for something that closely tracks SHY.

Lowell

http://itawealth.com]]>

I'll send you the information in a moment. To answer the Canada question, while I've not sought out Canadian ETFs, they must be available. Here is what to look for.

Find an ETF that covers the entire equity spectrum of Canada. Next, find an ETF that closely matches VEU as VEU not only includes developed, but emerging market equities as well. Third, look for an intermediate to long-term bond ETF. The last ingredient in this mixture is to find a low volatile treasury ETF. Again, look for something that closely tracks SHY.

Lowell

http://itawealth.com]]>

Lowell

http://itawealth.com]]>

Lowell

http://itawealth.com]]>

The momentum model does provide for global diversification. Yes, it is a mechanical model and does not require specific insights.

Using SHY as the "circuit breaker" ETF keeps one out of major problems when bear markets strike.

Lowell]]>

The momentum model does provide for global diversification. Yes, it is a mechanical model and does not require specific insights.

Using SHY as the "circuit breaker" ETF keeps one out of major problems when bear markets strike.

Lowell]]>

Stocks will work but it is important that they carry low correlations. It is easier to find low correlated ETFs and there is much greater safety in putting 100% in an ETF such as VTI vs. investing the entire portfolio in one stock.

I can only speak to back-tests and out-of-sample tests run thus far. More reporting will soon be showing up on my blog.

Lowell]]>

Stocks will work but it is important that they carry low correlations. It is easier to find low correlated ETFs and there is much greater safety in putting 100% in an ETF such as VTI vs. investing the entire portfolio in one stock.

I can only speak to back-tests and out-of-sample tests run thus far. More reporting will soon be showing up on my blog.

Lowell]]>

Lowell]]>

Lowell]]>

A 20% weight is applied to volatility and in this example I use a mean-variation calculation over a 14 day period. The spreadsheet is set up to use semi-variation, but back tests indicate there is little advantage to using semi-variance.

Lowell

http://itawealth.com]]>

A 20% weight is applied to volatility and in this example I use a mean-variation calculation over a 14 day period. The spreadsheet is set up to use semi-variation, but back tests indicate there is little advantage to using semi-variance.

Lowell

http://itawealth.com]]>

Here are some early results using this model. To go back as far as 9/24/1996 one must use index funds so here are the four. VTSMX, VGTSX, VBMFX and VFISX as the cutoff security. I won't go into all the details other than to say that the momentum portfolio had an average return of 598% with an average draw-down of 11.4%. The maximum DD, using 100 iterations in a Monte Carlo test was 24%. As a benchmark, VTSMX returned 347% with a maximum DD of 47%. BTW, the momentum portfolio was reviewed every 33 days.

The protection is there if one uses this type of model.

Lowell

http://itawealth.com]]>

Here are some early results using this model. To go back as far as 9/24/1996 one must use index funds so here are the four. VTSMX, VGTSX, VBMFX and VFISX as the cutoff security. I won't go into all the details other than to say that the momentum portfolio had an average return of 598% with an average draw-down of 11.4%. The maximum DD, using 100 iterations in a Monte Carlo test was 24%. As a benchmark, VTSMX returned 347% with a maximum DD of 47%. BTW, the momentum portfolio was reviewed every 33 days.

The protection is there if one uses this type of model.

Lowell

http://itawealth.com]]>

What is missing in Antonacci's book is a method for ranking the sectors. This is necessary to take advantage of the relative momentum factor. The ranking of ETFs problem is solved so it is a matter of back-testing sectors. Unfortunately there is little historical data available.

Lowell

http://itawealth.com]]>

What is missing in Antonacci's book is a method for ranking the sectors. This is necessary to take advantage of the relative momentum factor. The ranking of ETFs problem is solved so it is a matter of back-testing sectors. Unfortunately there is little historical data available.

Lowell

http://itawealth.com]]>

In general, a back-test provides a general view as there are many uncertainties that cause back-tested results to differ from a "real" portfolio.

What I am looking for in a back-tested model is this - does it give an edge to the investor? Further, I am looking for a robust model rather than optimized model as the optimized model is unlikely to be repeated.

Lowell ]]>

In general, a back-test provides a general view as there are many uncertainties that cause back-tested results to differ from a "real" portfolio.

What I am looking for in a back-tested model is this - does it give an edge to the investor? Further, I am looking for a robust model rather than optimized model as the optimized model is unlikely to be repeated.

Lowell ]]>

Apologies. I was wrong in my instructions. Multiply the weight by the ranking and you will come up with the correct value. For example, .3*1 + .5*1 + .2*11 = 3.

Lowell]]>

Apologies. I was wrong in my instructions. Multiply the weight by the ranking and you will come up with the correct value. For example, .3*1 + .5*1 + .2*11 = 3.

Lowell]]>

Here is another link with a few more adjustments.

http://bit.ly/1FTTA7S

Lowell]]>

Here is another link with a few more adjustments.

http://bit.ly/1FTTA7S

Lowell]]>

One of the best places to go is StockCharts. You will need to set up your own EMAs and after you Update or save the changes, be sure to use or change to Linkable Version. Then you can click back to this saved URL and all your changes will remain. Just experiment. Here is a link I provided.

http://bit.ly/1L8h3Jo

Lowell

http://itawealth.com]]>

One of the best places to go is StockCharts. You will need to set up your own EMAs and after you Update or save the changes, be sure to use or change to Linkable Version. Then you can click back to this saved URL and all your changes will remain. Just experiment. Here is a link I provided.

http://bit.ly/1L8h3Jo

Lowell

http://itawealth.com]]>

I add that second risk reducer of not holding ETFs that are under-performing their 195-Day Exponential Moving Average. The Monte Carlo analysis did not take that into consideration.

Yes, the spreadsheet is available, but it is not free.

Lowell

http://itawealth.com]]>

I add that second risk reducer of not holding ETFs that are under-performing their 195-Day Exponential Moving Average. The Monte Carlo analysis did not take that into consideration.

Yes, the spreadsheet is available, but it is not free.

Lowell

http://itawealth.com]]>

In the article, I stuck with the single risk reducing model, selling ETFs when they under-perform SHY as those were the rules for the Monte Carlo calculation. However, one can add a second risk reducer by using a sell rule similar to the 200 SMA. You will note a column labeled 195. That is the 195-Day Exponential Moving Average and it is negative (RED) for GSG. In fact GSG is priced 8.9% below its 195-Day EMA. Personally, I would not be holding GSG at this time as I sell ETFs that are priced below their 195-Day EMA.

I use the 195 EMA as it is faster moving than the 200 Day SMA and I prefer to move faster than the 200 Day crowd.

Lowell

http://itawealth.com]]>

In the article, I stuck with the single risk reducing model, selling ETFs when they under-perform SHY as those were the rules for the Monte Carlo calculation. However, one can add a second risk reducer by using a sell rule similar to the 200 SMA. You will note a column labeled 195. That is the 195-Day Exponential Moving Average and it is negative (RED) for GSG. In fact GSG is priced 8.9% below its 195-Day EMA. Personally, I would not be holding GSG at this time as I sell ETFs that are priced below their 195-Day EMA.

I use the 195 EMA as it is faster moving than the 200 Day SMA and I prefer to move faster than the 200 Day crowd.

Lowell

http://itawealth.com]]>

The mean-variance is a standard deviation calculation. The spreadsheet also permits one to choose a semi-variance calculation.

The weight is calculated from the performance values for ROC1 and ROC2 times the weight assigned to each. In addition, volatility is also factored into the calculation with the percentage weight also entering the calculation.

Lowell

http://itawealth.com]]>

The mean-variance is a standard deviation calculation. The spreadsheet also permits one to choose a semi-variance calculation.

The weight is calculated from the performance values for ROC1 and ROC2 times the weight assigned to each. In addition, volatility is also factored into the calculation with the percentage weight also entering the calculation.

Lowell

http://itawealth.com]]>

If you are enjoying Mebane Faber's work, also consider reading Gary Antonacci's book, Dual Momentum.

Lowell

http://itawealth.com]]>

If you are enjoying Mebane Faber's work, also consider reading Gary Antonacci's book, Dual Momentum.

Lowell

http://itawealth.com]]>

You are correct and I'm in error. Antonacci does use aggregate bonds while using T-bills as the performance reference.

Lowell]]>

You are correct and I'm in error. Antonacci does use aggregate bonds while using T-bills as the performance reference.

Lowell]]>