Index | Recent Threads | Who's Online | Search

Posts: 15    Pages: 2    1 2 Next
Last Post
New Thread
This topic has been viewed 2205 times and has 14 replies
mv388158
Thoughts On Logical-Invest.com

Hello P123 users I am not trying to advertise for logical Invest but I was wondering if anyone has ever tried implementing any of their strategies and how P123 compliments the QuanTtrader software. I personally think if P123 developed something similar it would be game changing. In a nutshell it is dynamic allocation of stocks/Etfs based on risk. One of the biggest gaps is the trading turnover and slippage are not accounted for in Quantrader but thanks to the tools we have at P123 we can easily account for them. I could run the SIM every month and take the positions with the allocated weights and run them in P123 but for a 5 year sim that is 60 months every time. Since I need to run a couple hundred Sims to optimize that is not very feasible. I have not implemented any real money with QuanTtrader yet so I would love anyone’s thoughts on this good or bad.
If you are curious to what it does here is the link.
https://logical-invest.com/app/dashboard

Regards,
Mark V.

May 15, 2019 3:07:15 PM       
yuvaltaylor
Re: Thoughts On Logical-Invest.com

I have not used LogicalInvest. A similar service is supplied by Allocate Smartly, https://allocatesmartly.com/. A P123 user, Georg Vrba, also offers TAA strategies on his website, https://imarketsignals.com/. So there are a number of players in the tactical allocation space.

I myself don't think out-of-sample returns for tactical allocation strategies will be better than static allocation strategies with periodic rebalancing.

What does QuanTrader do that P123 doesn't?

Yuval Taylor
Product Manager, Portfolio123
invest(igations)
Any opinions or recommendations in this message are not opinions or recommendations of Portfolio123 Securities LLC.

May 15, 2019 9:17:56 PM       
hemmerling
Re: Thoughts On Logical-Invest.com

Just my two cents but I would stay away. The backtests looked amazing. I signed up for my free trial and looked at the strategies. The NASDAQ strategy had 30%+ annual returns in highly liquid tickers. But if you look closer - there are only 4 or 5 stocks max. This made me think it is super-curve-fit. And forward performance since they went live is really bad. Same thing when I tested the ML ranks of Kavout - they had these machine curve-fit portfolios of 70% annually on bizarre stuff like obesity tickers. Live performance always did horrible. When you blend a big bag of highly curve-fit strategies, the backtest looks incredible as low correlation super-optimized ridiculously small portfolios will do.

The forward returns are almost guaranteed to disappoint. I think this is data-mining at its worst (or best depending on how you look at it). Either they will keep adding new strategies with no live performance and removing older ones that don't work, or they will struggle to bring in new customers. I could be wrong though.

May 16, 2019 10:29:07 AM       
mv388158
Re: Thoughts On Logical-Invest.com

Yuval and Kurtis thank you for the input,

I really cannot do the strategy justice if you download the quanttrader software and play with it for a day you will be able to recreate every strategy on the website and fully understand what it does. The process of optimization is key and you can easily curve fit. I will try to explain.
P123 is not able to dynamically allocate the % invested over a given time period based on the volatility of the asset. The following article explains this concept with the most basic strategy possible. P123 cannot do this.

https://seekingalpha.com/article/2714185-the-...nvestment-strategy?page=2

This strategy last year had a 6% drawdown versus the S&P which had a 15%. The returns were the same. It’s a great way to Hedge. The 20 year simulation doubles the S&P with much better sharp.

I agree with Kurtis that there is probably a lot of curve fitting in the models but the basic spy-tlt model has merit. The optimizer shows steady returns across any lookback period and modified sharp ratio. All the other models the returns are all over the place so yes they are probably going to fail.

The reason I am so interested in this is I think it is another great addition for hedging and market timing. All the models I created using quanttrader hedged Nov and Dec 2018 and then got back in Jan or Feb 2019. This would be a great addition to P123.

I’m not holding my breath judging by the lack of response to this forum most people think it’s not a worthy effort and I do think building a Dynamic optimizer like Quanttrader would be a significant effort with questionable value but I am going to use the strategy for market timing. I will let everyone know my results.

Regards,
Mark V.

May 16, 2019 8:00:37 PM       
wwasilev
Re: Thoughts On Logical-Invest.com

I think the idea in the seekingalpha article is worth exploring. Perhaps a R-lang based dynamic book solution is feasible.

May 18, 2019 11:50:28 AM       
Edit 1 times, last edit by wwasilev at May 18, 2019 11:59:43 AM
ClayAndresen
Re: Thoughts On Logical-Invest.com

Any update on what you found using Logical-Invest? I just started their free trail and would appreciate any input.

Feb 18, 2021 6:24:40 PM       
mmasand
Re: Thoughts On Logical-Invest.com

Yuval - Thanks for sharing your thoughts. I have been using AllocateSmartly for the liquidity portion of my portfolio and was curious about your comment on tactical asset allocation not doing better than static allocation with rebalance. Is it based on some testing you did or broad understanding? If latter, can you share if your opinion is related to things like curve-fitting or transaction costs, etc.

Feb 18, 2021 8:12:27 PM       
Jrinne
Re: Thoughts On Logical-Invest.com

Just my two cents but I would stay away. The backtests looked amazing. I signed up for my free trial and looked at the strategies. The NASDAQ strategy had 30%+ annual returns in highly liquid tickers. But if you look closer - there are only 4 or 5 stocks max. This made me think it is super-curve-fit. And forward performance since they went live is really bad. Same thing when I tested the ML ranks of Kavout - they had these machine curve-fit portfolios of 70% annually on bizarre stuff like obesity tickers. Live performance always did horrible. When you blend a big bag of highly curve-fit strategies, the backtest looks incredible as low correlation super-optimized ridiculously small portfolios will do.

The forward returns are almost guaranteed to disappoint. I think this is data-mining at its worst (or best depending on how you look at it). Either they will keep adding new strategies with no live performance and removing older ones that don't work, or they will struggle to bring in new customers. I could be wrong though.

I could not agree more that overfitting is possible and occurs pretty much everywhere—including at Logical-Invest.com.

Kurtis,

You have written about this general topic at Seeking Alpha: The Hidden Danger Of Adaptive Asset Allocation

From the article:

"Adaptive asset allocation can lower volatility risk and even improve returns."

I think that one of the concerns in your article is that this lower volatility is not enough to safely allow for the use of leverage:

"If you think that low volatility allows you to boost returns by using leverage - read on."

Summarizing you said the following. I do not think this was taken out of context and I made sure to include your caveat about the use of leverage. The emphasis is mine. Your short article was publish 3 years ago on Valentines day (Feb. 14) in 1998. Please expand on it and let us know how your opinions may have changed:

"The point I am trying to make is that Adaptive Asset Allocation is great. Use it to create a smoother ride and boost your Sharpe Ratio…and possibly total returns. But if you think that blending various asset classes takes away the risk that they could all fall at once and you begin to employ leverage - watch out!"

It does not seem strange to me that someone might want to create a diversified portfolio using some sort of rational method. I am not aware of a professional fund that does not attempt to at least diversify--often using something they learned while getting their degrees. Admittedly some of the pros have lots of options, including well, options and derivatives in general as well as being able to short including the use of pairs-trading in some cases.

I am not saying anyone has to do it and Kurtis has written a short (possibly incomplete) article about only one method. For those who do make an effort to diversify and/or hedge their portfolios: What methods do you use? Mark and Kurtis, thank you for sharing some of your research on this topic already.

Jim

From time to time you will encounter Luddites, who are beyond redemption.
--de Prado, Marcos López on the topic of machine learning for financial applications

Feb 19, 2021 6:22:34 AM       
Edit 22 times, last edit by Jrinne at Feb 19, 2021 8:08:10 AM
yuvaltaylor
Re: Thoughts On Logical-Invest.com

Yuval - Thanks for sharing your thoughts. I have been using AllocateSmartly for the liquidity portion of my portfolio and was curious about your comment on tactical asset allocation not doing better than static allocation with rebalance. Is it based on some testing you did or broad understanding? If latter, can you share if your opinion is related to things like curve-fitting or transaction costs, etc.


I thought this article did a very good job of explaining why I'm skeptical. https://www.ipe.com/tactical-asset-allocation...-timing-/10006908.article

I've also written a bit about this issue. See https://backland.typepad.com/investigations/2...d-trading-a-dialogue.html and https://blog.portfolio123.com/change-partners-some-thoughts-on-market-regimes/

Yuval Taylor
Product Manager, Portfolio123
invest(igations)
Any opinions or recommendations in this message are not opinions or recommendations of Portfolio123 Securities LLC.

Feb 19, 2021 9:12:18 AM       
Edit 1 times, last edit by yuvaltaylor at Feb 19, 2021 9:19:39 AM
mmasand
Re: Thoughts On Logical-Invest.com

Thanks Yuval

Feb 19, 2021 9:53:46 AM       
Posts: 15    Pages: 2    1 2 Next
 Last Post