Another good read, and this strategy seems really promising! I am curious on your thoughts to try to squeeze out more returns with TQQQ or 3x ETFs. Drawdowns will be rougher, but maybe tolerable with a strat like this. In general, I would like to understand your opinion better around trying 2 and 3x leveraged products.
Have you tried more regime filters based on VIX related ETFs? Is that generally something you try?
Hi, Jay. 3x ETFs can be an option, too. It boils down to your risk tolerance and other relevant situations like your investment horizon. To me, 2x ones look like a sweet spot. Volatility is known to be more predictable than return and that's why I try to put more weight to risk measures than return measures.
There are practically infinite variety of regime filters and canaries you can mix and match, so please feel free to experiment and share the result if you find anything interesting :)
Thanks for your response, I know some regime filters and canaries, but definitely still trying to learn about interesting data based filters (vix, yield spreads). An overview of those strategies or approaches could be an interesting topic to dig into more.
Anything with a MAR of around 1 over a time period of 18 years is interesting, especially when you have zero negative years.
Actually, Hi-Yield Spreads also belong to my Macro Indicators of choice.
What I personally like to do is look at the relationship of $HYIOAS to the SPX. (On Stockcharts.com, just enter $SPX:$HYIOAS).
I think what I do -- check the 50/200-day moving average crossovers -- isn't so very different from your basic approach. It seems to me that those crossovers are quite good risk-off and risk-on indicators.
As always, this is not flawless. It didn't work in March 2020, and also gave bum signals in 2002 (too early risk-on) and in 1998 (noisy).
$NDX:$HYIOAS looks slightly superior, in fact, which might confirm your results with the QQQ variants. (Unfortunately, I haven't backtested any of this, it's strictly observational). This is more on the mark in 2019. And, it had excellent timing in 2011.
I totally agree that achieving a MAR or Sharpe of 1 or above while showing a two-digit return over a 17~18 years is not an easy feat.
Thanks for sharing the ratio idea! IIGC, you're saying doing the crossover check over the ratio time series, aren't you? There is currently no way to do such a check in QuantMage, but it's definitely an interesting idea.
"you're saying doing the crossover check over the ratio time series, aren't you?"
- exactly.
I'd upload a screenshot of the chart but I think that's not possible on Substack.
When I post my next TA overview, I'll be including that one. In February, I expect.
By the way, it's easy to download historical price data from Stockcharts, so if you wanted to integrate that into Quantmage, I assume it would be possible.
Oh my, with a further look back, SPY : High yield spreads loses a lot of performance. The post-bubble years in particular look poorly. Back to the old drawing board...
But I like HYIOS:IEF! Excellent catch, much improved on the SPY pair.
Yeah, I noted the behavior, too. It can happen when both A and B are true at once when the logic is "Once A Then enter C Until B Then exit to D." I updated the spell to check (A and not B) instead of just A as the entry condition :)
I was able to manually backtest and it seems that you get conflicting signals in from April of 2020 to July 2020. Which signal should take precedent buy or sell? Also maybe you can post what the chart entry and exits looks like so I can see if I have plotted everything properly.
I got it to work on Quantmage thanks! Not sure why its wrong on thinkorswim. Thanks for the help. BTW if there anyway to get alerts on Quantmage when it enters or exits a position? I cant seem to find that setting?
There is no separate alert feature. You can either automate the trading or do a paper trading, which will send you trade notification emails right after each day's trades. Both require the upper tier plan, though.
Another good read, and this strategy seems really promising! I am curious on your thoughts to try to squeeze out more returns with TQQQ or 3x ETFs. Drawdowns will be rougher, but maybe tolerable with a strat like this. In general, I would like to understand your opinion better around trying 2 and 3x leveraged products.
Have you tried more regime filters based on VIX related ETFs? Is that generally something you try?
Hi, Jay. 3x ETFs can be an option, too. It boils down to your risk tolerance and other relevant situations like your investment horizon. To me, 2x ones look like a sweet spot. Volatility is known to be more predictable than return and that's why I try to put more weight to risk measures than return measures.
There are practically infinite variety of regime filters and canaries you can mix and match, so please feel free to experiment and share the result if you find anything interesting :)
Thanks for your response, I know some regime filters and canaries, but definitely still trying to learn about interesting data based filters (vix, yield spreads). An overview of those strategies or approaches could be an interesting topic to dig into more.
very good work, thanks JJ!
Anything with a MAR of around 1 over a time period of 18 years is interesting, especially when you have zero negative years.
Actually, Hi-Yield Spreads also belong to my Macro Indicators of choice.
What I personally like to do is look at the relationship of $HYIOAS to the SPX. (On Stockcharts.com, just enter $SPX:$HYIOAS).
I think what I do -- check the 50/200-day moving average crossovers -- isn't so very different from your basic approach. It seems to me that those crossovers are quite good risk-off and risk-on indicators.
As always, this is not flawless. It didn't work in March 2020, and also gave bum signals in 2002 (too early risk-on) and in 1998 (noisy).
$NDX:$HYIOAS looks slightly superior, in fact, which might confirm your results with the QQQ variants. (Unfortunately, I haven't backtested any of this, it's strictly observational). This is more on the mark in 2019. And, it had excellent timing in 2011.
Just my grains of salt. Have a great new year!
I totally agree that achieving a MAR or Sharpe of 1 or above while showing a two-digit return over a 17~18 years is not an easy feat.
Thanks for sharing the ratio idea! IIGC, you're saying doing the crossover check over the ratio time series, aren't you? There is currently no way to do such a check in QuantMage, but it's definitely an interesting idea.
Have a prosperous new year, too!
"you're saying doing the crossover check over the ratio time series, aren't you?"
- exactly.
I'd upload a screenshot of the chart but I think that's not possible on Substack.
When I post my next TA overview, I'll be including that one. In February, I expect.
By the way, it's easy to download historical price data from Stockcharts, so if you wanted to integrate that into Quantmage, I assume it would be possible.
I just added a feature to support price ratios:
https://quantmage.canny.io/changelog/nuanced-estimation-of-relative-momentum-with-price-ratios
With it, I implemented your idea:
https://quantmage.app/grimoire/f761789a29adbe91b855ef697b49a7b7
In my experiments, using IEI, a medium-term bonds ETF, instead of SPY gave me a more promising result:
https://quantmage.app/grimoire/3e8fc96369f0461cff139ab13f4d225d
the price ratio feature is fantastic!
Oh my, with a further look back, SPY : High yield spreads loses a lot of performance. The post-bubble years in particular look poorly. Back to the old drawing board...
But I like HYIOS:IEF! Excellent catch, much improved on the SPY pair.
Hi, I'm Korean. :)
I found you through Mr. Martin.
I've been reading all your posts lately. Thank you.
Looking at this strategy, I'm doing UUP one day and XLK one day.
What should I do in this case?
* I am using Google Translate, so there may be some awkward sentences. I hope you understand.
Yeah, I noted the behavior, too. It can happen when both A and B are true at once when the logic is "Once A Then enter C Until B Then exit to D." I updated the spell to check (A and not B) instead of just A as the entry condition :)
Thank you. :)
You inspired me a lot. Haha
Are you Korean? Judging from the fact that you used your Korean name, I think you lived in Korea and then moved to the US.
네, 한국 출신입니다. 반갑습니다.
I was able to manually backtest and it seems that you get conflicting signals in from April of 2020 to July 2020. Which signal should take precedent buy or sell? Also maybe you can post what the chart entry and exits looks like so I can see if I have plotted everything properly.
You can easily check that in QuantMage as explained in https://quantmage.app/library/guides/branching_history/.
I got it to work on Quantmage thanks! Not sure why its wrong on thinkorswim. Thanks for the help. BTW if there anyway to get alerts on Quantmage when it enters or exits a position? I cant seem to find that setting?
There is no separate alert feature. You can either automate the trading or do a paper trading, which will send you trade notification emails right after each day's trades. Both require the upper tier plan, though.
Also are you completely out of the market all of 2014? I may be backtesting it wrong
Great idea. Is there any ticker on any platform such as Thinkorswim, Tradingview, etc.... that we can set alerts on to follow this?
TradingView has https://www.tradingview.com/symbols/FRED-BAMLH0A0HYM2/. No idea about ThinkOrSwim, though.
Thanks when I try to put alerts on tradingview it says they are not allowed on economic indicators.