- 01/17/2015 at 2:09 pm #16031
Nice work on adding in correlations. Knowing your collective brainpower, you could optimize an allocation for a given level of risk- have you thought about doing that (among your hundreds of other to-do’s)?
Could you add another column showing drawdown? I think that would help put an allocation in perspective I think better than vol.
Great work all.01/17/2015 at 2:09 pm #16032
Aloha, and thanks for the feedback.
We have one strategy on the burner that does exactly that. Using minimum (alt. mean) variance optimization combined with volatility targeting. E.g. weightings are first computed by a solver algorithm to find minimum (or target) volatility at the portfolio level, and exposure is reduced if expected volatility is above the target volatility (or scaled up in low volatility environments).
This is inspired by the Adaptive Asset Allocation paper by Butler, Philbrick and Gordillo whom we are big fans of – another must read! We’ve been trading this already sucessfully, but need some more time to bring it into an easy-to-digest format.
Yes, drawdown is definitely the next to add. As this tool is running on monthly data, this would be monthly MaxDrawDown only, which for all strategies but the MYRS also would match realized drawdown. Wanted to wait for the integrated version of the tool, but can plug it in.01/17/2015 at 2:10 pm #16033
Very good Alex. Thanks for your detailed and prompt replies. I am likewise a big fan of Adaptive Allocation and respect David Varadi and his work.
Instead of solving for min. vol, which punishes upside vol, what about targeting drawdown or max Sortino as Patrick above referenced? Curious as to your thoughts. Showing Sharpe will certainly help people compare as you noted (in Patrick’s reply above) Sharpe is more widely used.
Being able to see your thoughts and your responsiveness is a big benefit.
Many black box or other less transparent strategies will be abandoned once the inevitable drawdown occurs as the user doesn’t know enough to stick with the strategy so your details help.01/18/2015 at 10:17 am #16090
Have just added the monthly drawdown chart above, please let me know if this addresses your needs. We can built portfolio examples with a target Calmar (CAR/MaxDD) or Sortino ratio, let me know some examples you would like to see here in the forum. Again, here these are ‘optimizations’ over the whole time period, in our background system we can build more sophisticated models like the ‘meta strategy’ we’ve been teasing now for several weeks, but still owe you :-)01/20/2015 at 6:27 pm #16154
Thanks LI. If you could please include the Sortino ratio, max drawdown and duration (i.e. drawdown + recovery). I think the portfolio builder should take desired returns/metrics as inputs and display the weighting of strategies as outputs. Separately, it is difficult to appreciate the impact of the metrics regardless if they are inputs or outputs. CAGR is clear but at what volatility do people panic and abandon strategies even if the volatility is within the bounds of the strategy but not acceptable psychologically.
Daniel02/03/2015 at 1:44 pm #16777
I have a question for the forum. I am from Denmark and hold my portfolio in USD at the moment. As you probably know speculators are trying to push the Kroner higher. http://www.bloomberg.com/news/articles/2015-01-19/denmark-strikes-back-at-speculators-as-peg-defenses-burnished
Please lets have a discussion how to protect the portfolio thats in USD ETF’s.
Do you hedge the currency and how do you keep the capital intact.02/03/2015 at 6:04 pm #16779
A question on the portfolio builder. Does it take into account the new variable hedging strategies? It seems to be based on the pre-2014 non-hedged strategies. I am trying to set up a portfolio but I’m stuck on whether to add a bond rotation strategy to hedge the portfolio as the current strategies such as UIS, MYRS and GMRS are already hedged as of this year. Your thoughts please.
email@example.com/04/2015 at 2:08 pm #16811
A suggestion. The UIS strategy is presented in a non-leveraged and a leveraged version. It would be helpful to show the leveraged version in the Portfolio Builder so we could see what effect the leveraged strategy will have on the portfolios we build.
firstname.lastname@example.org/05/2015 at 1:52 am #16825
Depending on your broker you can hedge your USD ETF exposure by buying an equal nominal amount of DKK/USD in the forex market (via margin) or use futures if those are available to you.02/05/2015 at 3:28 am #16826
Thanks. I am using Interactive Brokers. What futures can you recommend.?02/05/2015 at 5:13 pm #16847
I am not familiar with available futures as far as DKK is concerned. If you do find a solution please share for the benefit of other users. If anyone else has more info feel free to join the conversation.02/05/2015 at 5:22 pm #16848
I assume you mean the BUG leveraged?
Thanks for pointing it out. Will do.02/05/2015 at 5:27 pm #16849
No, I mean the UIS strategy. In the monthly stategy update, the UIS strategy is posted as a SPY- EDV strategy. As an alternative, a leveraged short version is also listed. Since this strategy is traded in an IRA I have substituted a SPXL-TMF leveraged strategy instead.02/07/2015 at 6:42 pm #17086
re your question on the timeseries in the portfolio builder, sorry for the late response. The data used is based on the former strategies, and only since late last year the new adaptive strategy signals have dropped in. However, as the portfolio builder is rather a ‘forward-looking’ instrument, I agree that we should use the backtested data from the adaptive strategies.
For consistency to our long-term followers we’ve so far stuck to our principle to ‘fix the past’, e.g. not change any signal or performance data even if the strategy had evolved. With the new portfolio builder (just approved final specs and design and kicked-off built yesterday, Yuppie!!) we will use the adaptive strategy signals. Finally, you will see the difference not to be significant, we’ve done some test already.02/07/2015 at 6:54 pm #17087
Michael, if there is a general interest in a leveraged or short ETF version of the UIS we could certainly built this, publish the backtest and include signals in the strategy posts also. I’m a bit hesitant to include further options in the portfolio builder though. It’s already quite crowded with the strategies that have the most followers, so adding further alternatives might create confusion in my eyes.
In addition to the long promised and finally kicked-off next version of the online tool, I’d rather propose to make an excel tool available for the advanced users, so you can also do your own optimizations and tracking. We could include more timeseries like the requested there, build in some solver optimizations for example, and share somewhere for open collaboration. For some stuff excel is just more flexible, also like the things I still owe Daniel from the post above.
Deal? Also happy to be convinced otherwise..
You must be logged in to reply to this topic.