Technical Analysis

Hi Peter I am new here but have over 35 years of trading and trade system development.  I would submit that technical analysis can generate alpha in the right circumstances.   Have you looked at using either a neural network or some other back testing of  your strategies ?  

I'm not a content creator in this space, I'm just a user. I will say, though, I do have exerience with neural nets outside of finance, and they are no panacea. It's very easy to get yourself into an overfitting situation, where you get a perfect backtest but the model doesn't perform very well out-of-sample.

Hi Peter I am new here but have over 35 years of trading and trade system development.  I would submit that technical analysis can generate alpha in the right circumstances.   Have you looked at using either a neural network or some other back testing of  your strategies ?  

I'm not a content creator in this space, I'm just a user. I will say, though, I do have exerience with neural nets outside of finance, and they are no panacea. It's very easy to get yourself into an overfitting situation, where you get a perfect backtest but the model doesn't perform very well out-of-sample.

Hi Peter I am new here but have over 35 years of trading and trade system development.  I would submit that technical analysis can generate alpha in the right circumstances.   Have you looked at using either a neural network or some other back testing of  your strategies ?  

Howdy, James.

I have been following this thread since it first appeared.

The theories swapped here have spurred me to get better informed on technical analysis. So far, in about 2 months, I have read 2 books and downloaded 3 OL courses on TA and completed the assignments and passed all the tests. I am nowhere near competent at a novice level.

Your interest in applying TA to identify sector momentum trends interests me. I hope you get a lot on responses to your request for insights.

I have applied my meager TA level 1 knowledge to screen several dozen ETFs I am interested in, looking at trends, support/resistance levels, and searching for "clues" that might indicate a breakout pattern, writing down my observations and predictions, then checking my predictions to see how good I am at seeing pattens in TA "Rorschach" in the chats I have access to at several brokerages. Basic stuff. I am not ready to solo. yet. I don't trust that my "student" training will guarantee that I can keep the shiny side up on landing.

I have a stake in your success. My basic investment strategy is to identify low ER (low cost), high AUM (high liquidity)  sector ETFs.

I have found 2 sources of data I have found to be trustworthy, all at the same broker. (1) A Trader Outlook that updates sector performance trends weekly, and (2) a Monthly Outlook that addresses 5-8 trends/issues that are intended to link past data (usually in line chart format) to their analysis of current market trends.

One recurring chart is a diagram of the performance of all 11 GICS sector indices updated monthly. The designer has chosen a very clever format. She displays the monthly performance data in a "quilt" format. Each sector is color-coded and then stack-ranked like a Totem pole for the month. Then she displays the last 12 months side-by-side so users can easily observe how the 11 sectors change relative to each other month-to-month. Each cell has its performance data. It is really a matrix that has been sorted dynamically MOM. It deserves to be in the Data Hall of Fame because all the information it provides in a compact format.

Since this diagram is proprietary, I cannot copy it into this thread, but here is a link to the most recent version that discusses how many sectors' performance have inverted from 4Q22 to 1Q23.

https://www.schwab.com/learn/story/caveat-emptor-important-market-shifts-underway?cmp=em-QYD

MOM changes appear to be completely random. No trends are observable in either the monthly quilts or the YOY quilts.

These data lead me to estimate that whatever momentum trends may exist among the  sectors MOM and YOY are random because of the wide variations across time periods.

This is all I can contribute.

I hope someone has data that disproves my observation. After all, Karl Popper's Falsification Principle suggests that for a theory to be considered scientific, it must be able to be tested and conceivably be proven false. The hypothesis that "all swans are white" can be falsified by observing a black swan.

Regards.

Gary Antonacci did a study of industry sector relative momentum, and he found that it did not generate excess returns. Gary is the author of Dual Momentum. 

All of the momentum strategies I see published on Allocate Smartly are not sector strategies... rather, they submit even larger asset classes to various relative and absolute momentum tests. The fastest and "twitchiest" strategies use 1, 3, 6, and 12 month momentum, some use 3, 6, 12, some use 6, some use 12. They almost universally trade one time per month, last trading day of the month... so it's not swing trading. They don't ever use a technical indicator that a computer cannot utilize in a hard quantitative manner. Any technical that requires human interpretation or guessing isn't useful for their purposes. Everything is back-tested to 1972 or so.

Some examples of these large asset classes are: US large caps, US small caps, US value, US momentum, US REIT, Emerging Market, EAFE, Japan, Europe, 2 year T-Notes, 10 T-Notes, 30 T-Bonds, high yield bonds, investment grade corporates, T-Bills, gold, commodities. 

Howdy, James.

I have been following this thread since it first appeared.

The theories swapped here have spurred me to get better informed on technical analysis. So far, in about 2 months, I have read 2 books and downloaded 3 OL courses on TA and completed the assignments and passed all the tests. I am nowhere near competent at a novice level.

Your interest in applying TA to identify sector momentum trends interests me. I hope you get a lot on responses to your request for insights.

I have applied my meager TA level 1 knowledge to screen several dozen ETFs I am interested in, looking at trends, support/resistance levels, and searching for "clues" that might indicate a breakout pattern, writing down my observations and predictions, then checking my predictions to see how good I am at seeing pattens in TA "Rorschach" in the chats I have access to at several brokerages. Basic stuff. I am not ready to solo. yet. I don't trust that my "student" training will guarantee that I can keep the shiny side up on landing.

I have a stake in your success. My basic investment strategy is to identify low ER (low cost), high AUM (high liquidity)  sector ETFs.

I have found 2 sources of data I have found to be trustworthy, all at the same broker. (1) A Trader Outlook that updates sector performance trends weekly, and (2) a Monthly Outlook that addresses 5-8 trends/issues that are intended to link past data (usually in line chart format) to their analysis of current market trends.

One recurring chart is a diagram of the performance of all 11 GICS sector indices updated monthly. The designer has chosen a very clever format. She displays the monthly performance data in a "quilt" format. Each sector is color-coded and then stack-ranked like a Totem pole for the month. Then she displays the last 12 months side-by-side so users can easily observe how the 11 sectors change relative to each other month-to-month. Each cell has its performance data. It is really a matrix that has been sorted dynamically MOM. It deserves to be in the Data Hall of Fame because all the information it provides in a compact format.

Since this diagram is proprietary, I cannot copy it into this thread, but here is a link to the most recent version that discusses how many sectors' performance have inverted from 4Q22 to 1Q23.

https://www.schwab.com/learn/story/caveat-emptor-important-market-shifts-underway?cmp=em-QYD

MOM changes appear to be completely random. No trends are observable in either the monthly quilts or the YOY quilts.

These data lead me to estimate that whatever momentum trends may exist among the  sectors MOM and YOY are random because of the wide variations across time periods.

This is all I can contribute.

I hope someone has data that disproves my observation. After all, Karl Popper's Falsification Principle suggests that for a theory to be considered scientific, it must be able to be tested and conceivably be proven false. The hypothesis that "all swans are white" can be falsified by observing a black swan.

Regards.

Gary Antonacci did a study of industry sector relative momentum, and he found that it did not generate excess returns. Gary is the author of Dual Momentum. 

All of the momentum strategies I see published on Allocate Smartly are not sector strategies... rather, they submit even larger asset classes to various relative and absolute momentum tests. The fastest and "twitchiest" strategies use 1, 3, 6, and 12 month momentum, some use 3, 6, 12, some use 6, some use 12. They almost universally trade one time per month, last trading day of the month... so it's not swing trading. They don't ever use a technical indicator that a computer cannot utilize in a hard quantitative manner. Any technical that requires human interpretation or guessing isn't useful for their purposes. Everything is back-tested to 1972 or so.

Some examples of these large asset classes are: US large caps, US small caps, US value, US momentum, US REIT, Emerging Market, EAFE, Japan, Europe, 2 year T-Notes, 10 T-Notes, 30 T-Bonds, high yield bonds, investment grade corporates, T-Bills, gold, commodities. 

Howdy, James.

I have been following this thread since it first appeared.

The theories swapped here have spurred me to get better informed on technical analysis. So far, in about 2 months, I have read 2 books and downloaded 3 OL courses on TA and completed the assignments and passed all the tests. I am nowhere near competent at a novice level.

Your interest in applying TA to identify sector momentum trends interests me. I hope you get a lot on responses to your request for insights.

I have applied my meager TA level 1 knowledge to screen several dozen ETFs I am interested in, looking at trends, support/resistance levels, and searching for "clues" that might indicate a breakout pattern, writing down my observations and predictions, then checking my predictions to see how good I am at seeing pattens in TA "Rorschach" in the chats I have access to at several brokerages. Basic stuff. I am not ready to solo. yet. I don't trust that my "student" training will guarantee that I can keep the shiny side up on landing.

I have a stake in your success. My basic investment strategy is to identify low ER (low cost), high AUM (high liquidity)  sector ETFs.

I have found 2 sources of data I have found to be trustworthy, all at the same broker. (1) A Trader Outlook that updates sector performance trends weekly, and (2) a Monthly Outlook that addresses 5-8 trends/issues that are intended to link past data (usually in line chart format) to their analysis of current market trends.

One recurring chart is a diagram of the performance of all 11 GICS sector indices updated monthly. The designer has chosen a very clever format. She displays the monthly performance data in a "quilt" format. Each sector is color-coded and then stack-ranked like a Totem pole for the month. Then she displays the last 12 months side-by-side so users can easily observe how the 11 sectors change relative to each other month-to-month. Each cell has its performance data. It is really a matrix that has been sorted dynamically MOM. It deserves to be in the Data Hall of Fame because all the information it provides in a compact format.

Since this diagram is proprietary, I cannot copy it into this thread, but here is a link to the most recent version that discusses how many sectors' performance have inverted from 4Q22 to 1Q23.

https://www.schwab.com/learn/story/caveat-emptor-important-market-shifts-underway?cmp=em-QYD

MOM changes appear to be completely random. No trends are observable in either the monthly quilts or the YOY quilts.

These data lead me to estimate that whatever momentum trends may exist among the  sectors MOM and YOY are random because of the wide variations across time periods.

This is all I can contribute.

I hope someone has data that disproves my observation. After all, Karl Popper's Falsification Principle suggests that for a theory to be considered scientific, it must be able to be tested and conceivably be proven false. The hypothesis that "all swans are white" can be falsified by observing a black swan.

Regards.