GEM Backtest Extended to 1950

A place to talk about speculative investing ideas for the optional Variable Portfolio

Moderator: Global Moderator

User avatar
Kbg
Executive Member
Executive Member
Posts: 1242
Joined: Fri May 23, 2014 4:18 pm

GEM Backtest Extended to 1950

Post by Kbg » Wed Oct 17, 2018 12:25 pm

A great post. It also has a very clear message that there are periods of under performance if your benchmark is the S&P 500. I think the message is clear...GEM under performs most of the time unless there are one of two conditions: 1) Extended foreign out performance 2) US Bear Market.

https://www.dualmomentum.net/2018/10/ex ... ities.html
User avatar
ochotona
Executive Member
Executive Member
Posts: 2476
Joined: Fri Jan 30, 2015 5:54 am

Re: GEM Backtest Extended to 1950

Post by ochotona » Wed Oct 17, 2018 4:25 pm

I will HAPPILY accept underperformance during bulls and much smaller drawdowns during bears. Ideally, I'd like my CAGR to the the same every year! What's not to like about his GEM growth of $10,000 chart, 1950-2018?

You can't have it both ways... the full unencumbered S&P 500 during bulls and the GEM or HBPP during bears.

GEM 60% and HBPP 40%. The ideal pairing for scaredy-cats who want some growth.
User avatar
Kbg
Executive Member
Executive Member
Posts: 1242
Joined: Fri May 23, 2014 4:18 pm

Re: GEM Backtest Extended to 1950

Post by Kbg » Wed Oct 17, 2018 5:26 pm

I think a better choice might be his Composite Dual Momentum plus a cash allocation...of course this assumes momentum remains a good investment mechanism. PP has the advantage of being as "untiming" as you can get.
User avatar
ochotona
Executive Member
Executive Member
Posts: 2476
Joined: Fri Jan 30, 2015 5:54 am

Re: GEM Backtest Extended to 1950

Post by ochotona » Thu Oct 18, 2018 1:15 pm

Another -6% on the S&P 500 by Halloween and GEM will be out of equities. Another ten full trading days. "Interesting times".
User avatar
InsuranceGuy
Executive Member
Executive Member
Posts: 187
Joined: Sun Mar 29, 2015 1:44 pm

Re: GEM Backtest Extended to 1950

Post by InsuranceGuy » Sat Oct 20, 2018 9:35 am

ochotona wrote:
Thu Oct 18, 2018 1:15 pm
Another -6% on the S&P 500 by Halloween and GEM will be out of equities. Another ten full trading days. "Interesting times".
If you use volatility as an additional filter it is likely you would be holding less than 100% come Halloween just using current levels.
User avatar
ochotona
Executive Member
Executive Member
Posts: 2476
Joined: Fri Jan 30, 2015 5:54 am

Re: GEM Backtest Extended to 1950

Post by ochotona » Sat Oct 20, 2018 9:52 am

InsuranceGuy wrote:
Sat Oct 20, 2018 9:35 am
ochotona wrote:
Thu Oct 18, 2018 1:15 pm
Another -6% on the S&P 500 by Halloween and GEM will be out of equities. Another ten full trading days. "Interesting times".
If you use volatility as an additional filter it is likely you would be holding less than 100% come Halloween just using current levels.
I think I'm doing that manually already. I folded up some positions what weren't working, they went below the 200 day MA, and I sold. I've been lightening up since February. Not driven by an algo, but selling down to my "sleep well at night" point.

What I'd really like Santa Claus to bring me is a back-tested trend-following algorithm that tries to minimize whipsaws and reduce the number of trades even below Antonacci's proclaimed 1.4 trades per year over the long term. I'd give up some CAGR and Sharpe Ratio for trade minimization.
User avatar
InsuranceGuy
Executive Member
Executive Member
Posts: 187
Joined: Sun Mar 29, 2015 1:44 pm

Re: GEM Backtest Extended to 1950

Post by InsuranceGuy » Sun Oct 21, 2018 4:57 pm

ochotona wrote:
Sat Oct 20, 2018 9:52 am
What I'd really like Santa Claus to bring me is a back-tested trend-following algorithm that tries to minimize whipsaws and reduce the number of trades even below Antonacci's proclaimed 1.4 trades per year over the long term. I'd give up some CAGR and Sharpe Ratio for trade minimization.
Maybe I need to sit on Santa's lap too...
User avatar
ochotona
Executive Member
Executive Member
Posts: 2476
Joined: Fri Jan 30, 2015 5:54 am

Re: GEM Backtest Extended to 1950

Post by ochotona » Sun Oct 21, 2018 7:05 pm

Such a method would probably be very non robust
User avatar
Kbg
Executive Member
Executive Member
Posts: 1242
Joined: Fri May 23, 2014 4:18 pm

Re: GEM Backtest Extended to 1950

Post by Kbg » Sun Oct 21, 2018 8:49 pm

I have a fairly robust back test capability and I can opine there are absolutely no freebies in this realm. It’s solely a matter of trade offs. If you want fewer annual trades then extend your look back period length. You can go out to about the two year point and still have the benefits of trend following. The trade off will be higher drawdowns. Whipsaw is also reduced at greater length periods as are tax events. Shorter periods drive up taxable events and tend to stay more in tune with the market and are particularly good when coming off bear bottoms...strikingly so. However, if the shorter look back length is out of phase with the markets action whipsaw can be brutal.

What look back periods work best going forward is absolutely unknowable and will be a function of market behavior. My personal approach is to assume I have no clue and I use two look back lengths and apply them separately to 50% of the portfolio. This way I’m guaranteed not to achieve the optimal possible but I will not get the worst possible either.
User avatar
ochotona
Executive Member
Executive Member
Posts: 2476
Joined: Fri Jan 30, 2015 5:54 am

Re: GEM Backtest Extended to 1950

Post by ochotona » Wed Oct 24, 2018 12:42 pm

I'm wondering if we will be selling on Nov 1... another 3% down from today's close and I'm out. I will keep my wife's stuff invested, her allocation is "coach potato" 50% equities, and a small portion of the family portfolio.
User avatar
Kbg
Executive Member
Executive Member
Posts: 1242
Joined: Fri May 23, 2014 4:18 pm

Re: GEM Backtest Extended to 1950

Post by Kbg » Wed Oct 24, 2018 10:47 pm

I have several accounts and rotate through one each week...can’t imagine I won’t be going to bonds on my shorter lookback and I’m curious to see as well if the longer lookback will go as well. If I can remember to post I will this weekend. At min I expect to be 50% bonds in one account by Monday.
User avatar
ochotona
Executive Member
Executive Member
Posts: 2476
Joined: Fri Jan 30, 2015 5:54 am

Re: GEM Backtest Extended to 1950

Post by ochotona » Sat Oct 27, 2018 9:53 am

S&P500 down 10% from the peak, I'm down 3.8%. Seriously, if I get "GEM stopped-out" of stocks at less than a 5% loss from the all-time portfolio high, that would be pretty nice, particularly if we stay in fixed-income for a few months at least.
Post Reply