Post New

Here's a couple of wild-and-crazy ideas using leveraged index ETFs that I recently heard.

1) "If you want to try something fun' when the QQQ ETF is in a tight range on a 1 min chart, place a buy order of both TQQQ and SQQQ (they are inverse ETF's of each other so that when one goes down the other goes up) and place the buy order at the recent top of the range (or swing high). When the market starts heading either down or up you win (because the other buy order never triggers)."

"...The above strategy has not only an extremely high success rate but usually the move typically grosses more than at least half of my other automated trading that does automated trailing stop losses."

No further details, no mention of when to close the trade if it fills. End of day?

-------------------
2) Look at a leveraged index as a long-term buy&hold. Yeah, the volatility is scary, buckle your seatbelt.

but...' look at the long-term charts of portfolio backtest. At PORTFOLIO VISUALIZER.

Huge volatility, huge max drawdown. TQQQ lost -79% when QQQ lost -35%. But TQQQ went down from $1,792,400 to $416,391 while QQQ dropped from $98,613 to $67,337. Notice that the bottom of $416,391 (TQQQ) is higher than the top of $98,613 (QQQ).

Might be worthwhile to buy a soupçon of TQQQ and just let it ride.


PV link: https://www.portfoliovisualizer.com/backtest-portf...

"Notice that the bottom of $416,391 (TQQQ) is higher than the top of $98,613 (QQQ)"

WOW ..... Just WOW ...... Speechless!!!!

Those are amazing numbers, but 2011 - 2023 was a strong market. How would this strategy have done from 2000 - 2009?

Those are amazing numbers, but 2011 - 2023 was a strong market. How would this strategy have done from 2000 - 2009?

Good question. I guess you could perhaps model it by downloading the daily historical prices for QQQ or SPY and calculating the daily results like the 2X or 3X ETF does. Remembering to include the Expense Ratio. Verify by checking agaist the actual results when the ETF was available.

And here I had no plans for the next couple of days. ;-)

UOPIX, which is 2x NASDAQ, can be backtested to 1999.

Go ahead, but it's not pretty.

I have looked at a trading strategy that is long a leveraged NASDAQ fund only when BCC=3!6!7. Difficult to backtest now that GTR1 is broken. But when it wasn't, UOPIX if BCC=3!6!7, else cash, could be backtested to the late 1997. It produced a CAGR of about 22, MDD -36, UI 8, Sharpe .8, and (surprisingly) an AT of 3.

Not an insane strategy for the strong of stomach. TQQQ would of course add to the excitement.

Baltassar

2011 - 2023 was a strong market. How would this strategy have done from 2000 - 2009?

As Henry Wadsworth Longfellow said, 'when she was good she was very very good. But when she was bad she was horrid.'

Better make sure your life insurance covers jumping out of a skyscraper.


Growth of $1000
Actual, week close-to-close, 2/12/10 to 6/16/23
QQQ:: $8408
TQQQ: $94,324

Computed from yahoo daily data:
QQQ: $8425
3X:: $102,641

Computed vs actual is close. Chart of TQQQ and 3X the lines are largely on top of one another.
Therefore the computation is close to being real life.

Computation applied to the actual QQQ data, 3/10/1999 to 2/10/2010, daily close to close:
Growth of $1000:
QQQ: $842
3x:: $13 (not a typo)

The good(?) news is that 3X would then grow to $1450 by 6/16/23.

I had always assumed that when they said "daily return" it meant open-to-close. But close-to-close fits the actual data.

Thank you for taking the time to check it.
I guess those good at technical analysis could do well with this strategy.

Trying some timing schemes, 1/1/2000 to 8/11/2023
Evaluating weekly, Relative Strength and SMA
SMA: does not work well.

RS: Best was a 30 week lookback.
TQQQ: $1,000 grows to $22,000 = CAGR 13.98%
QQQ: $1,000 grows to $3,865 = CAGR 5.89%

B&H:
TQQQ: You don't want to know.
QQQ: You don't want to know.

But, from Jan 2000 B&H of TQQQ was 20% CAGR, timed with RS was 14% CAGR
B&H of QQQ was 5.9%, timed was 8.0%

I do not have any idea of how you can tell when you should own TQQQ and when you shouldn't.

I do not have any idea of how you can tell when you should own TQQQ and when you shouldn't.

How about using the Nasdaq New Highs-New Lows? By selling TQQQ when that indicator goes negative should help to avoid 2000, 2001, 2002, 2008, etc. Buying when NH-NL starts going up again should work... most of the times.

How about using the Nasdaq New Highs-New Lows?

Where can I find a table of historical Nasdaq New Highs-New Lows? I found a few places with a chart, but to backtest I need the actual numbers.

Ray, I have a spreadsheet that has daily values of NHNL from the GTR1 backtester from 12/30/1926 to 3/10/2023 (the date the GTR1 backtester stopped updating). Would be happy to send to you, and anyone else who would like it!

I have a spreadsheet that has daily values of NHNL from the GTR1 backtester from 12/30/1926 to 3/10/2023

Maybe put it on a google docs spreadsheet and post the link? Then anybody with the link can download it. And/or just send it to me.

I can get the recent NH-NL from barcharts, for the last 2 years. For free. Could get much further back with $10/mo subscription.

I like free. :-)

Thanks.

Here's a link to the NH-NL spreadsheet...

https://docs.google.com/spreadsheets/d/1lq5UYfa8fc...

Using the daily data for the NASDAQ index from YAHOO, starting on 1/1/1985, I find the following
(How are tables created on this site?):

B&H:
CAGR: 11%
Max DD: 78%

Using EMA(13):
CAGR: 12%
Max DD: 45%

Using EMA(13) and 2x leverage:
CAGR: 24%
Max DD: 77%

Using EMA(13) and 3x leverage:
CAGR: 33%
Max DD: 93%


With some data mining, I found EMA(36) works better:

1x leverage:
CAGR: 13%
Max DD: 29%

2x:
CAGR: 25%
Max DD: 51%

3x:
CAGR: 35%
Max DD: 67%


SMA worked, too, but the returns were lower, and the Max DD's were much worse. Of course, who wants to do the analysis and trading every single day? Looking at monthly data, on the other hand, SMA gave better results than EMA, particularly the data-mined SMA(9):

1x leverage: (return is 1%-point less than B&H, but the Max DD is a lot less)
CAGR: 10%
Max DD: 32%

2x:
CAGR: 17%
Max DD: 61%

-Jim

<<Of course, who wants to do the analysis and trading every single day? Looking at monthly data...>>

Jim, would a weekly analysis of EMA(36) be possible?

The data you provided is intriguing...!

I was able to essentially duplicate the weighted NH-NL signal that came out of the GTR1 spreadsheet. That's the good news.

Computing 3X QQQ from 1/3/2020 to 8/11/2023
The bad news is that with timing, the return isn't anywhere near the 2010 to 2023 return. Better than SPY, though.
Recall that the 2010-2023 of 3X QQQ was 38.27%.

B&H:
SPY: 6.78% CAGR
QQQ: 5.89%
3X: -5.40%

A few timing schemes:
150 day lookback Relative Strength---
QQQ: 8.02%
3X: 13.98%
But changing the number of days by 5's had wildly differing returns.

225 day SMA ---
QQQ 6.98%
3X: 12.12%
Very short SMA periods were worse.


QTR1 weighted Nasdaq NH-NL ---
QQQ: 6.02%
3X: 10.14%

I was able to essentially duplicate the weighted NH-NL signal that came out of the GTR1 spreadsheet.

I would be interested to know how to do this, since, as was discussed on the old board, there are many possibilities. Stockcharts and Barcharts provide readily available versions of NHNL, but they calculate it differently from GTR1, and from each other. In practice I'm not sure the difference matters much, but knowing how to calculate the only back-testable version would be helpful.

Baltassar

QTR1 weighted Nasdaq NH-NL ---
QQQ: 6.02%
3X: 10.14%

Ray, thank you for doing this research. What values of the NH-NL indicator were you using to determine to be in or out?

Ray, I think I know now -- buy if indicator positive, sell if negative?

Jim

Using the daily data for the NASDAQ index from YAHOO, starting on 1/1/1985, I find the following

Are the results based on determining ema at the close and trading at the close, or trading next day at open(?) or close(?).

Craig

I would be interested to know how to do this, since, as was discussed on the old board, there are many possibilities. Stockcharts and Barcharts provide readily available versions of NHNL, but they calculate it differently from GTR1, and from each other. In practice I'm not sure the difference matters much, but knowing how to calculate the only back-testable version would be helpful.

It's probably easiest to just show the spreadsheet. Here's the link, anybody can view it, and copy or download to their own account to muck about with it.

Positive is the buy signal, negative is sell signal.

https://docs.google.com/spreadsheets/d/1KiMrNPrwpT...

My timing backtest assumed you bought at the close based on the signal of that day, presuming you could get the NH-NL figure just before the close. That may give a more optimistic picture than you could get in real life. An extra day lag has a lower return.

lsmr409, the google docs link you posted seems to be protected. I am having to send a "request access permissions" request. Can it be open to all?

Thanks for posting it.

Tails

Tails and others,

I think I have the settings correct now to view the Google sheet... :-)

Here you go:

https://docs.google.com/spreadsheets/d/1RyyevCoR2h...

I think I have the settings correct now....

Works like a champ, now, at least for me.

Thanks

Eric Hines

Craig,

Unfortunately, those results are based on computing the ema at close and trading at close on the same day, so probably not very realistic (the results for monthly trading might be a little more realistic).

I also used the NHNL data, using the same dates and approach:

1x leverage (the CAGR is no better than B&H (11%), though the DD is better (78%)):
CAGR: 10%
Max DD: 54%

2x (given that B&H has 78% DD, this appears to be a reasonable alternative to that):
CAGR: 20%
Max DD: 82%

Maybe I'll examine the results using opening prices, but I'm not sure these results are enticing enough to warrant the time.

-Jim

I do not have any idea of how you can tell when you should own TQQQ and when you shouldn't.

What about something simple. Using a weekly chart compare the close of QQQ to its 52-wk moving average. If the close is above the MA use TQQQ; if it closes below then exit.

You would have exited in late September 2000 and been out until mid-March 2003. There were ins and outs over the next five years, but the final exit for the next bear was in mid-August 2008. The return to TQQQ would have been in early June 2009. More ins and outs (including exiting for four weeks in the spring of 2020) but mostly long. Exit again in early February 2022 with a return a year later.

There are obviously whipsaws and such, but it might be worth looking into.

DB2

"What about something simple. Using a weekly chart compare the close of QQQ to its 52-wk moving average. If the close is above the MA use TQQQ; if it closes below then exit."

Interesting premise, but surprisingly not as good as a buy and hold of TQQQ, according to Portfolio Visualizer (since 2011):

https://www.portfoliovisualizer.com/tactical-asset...

Just not sure I'd be able to handle that buy and hold drawdown of TQQQ.

Interesting premise, but surprisingly not as good as a buy and hold of TQQQ, according to Portfolio Visualizer (since 2011)

Using QQQ, one can go back to the turn of the century and include the two bear markets before 2011. Using a lookback period of 12 months and weekly trading checks, the system grew $10K into $88K while buy-and-hold only made it to $77K (a 14% improvement).

DB2

Using a weekly chart compare the close of QQQ to its 52-wk moving average. If the close is above the MA use TQQQ; if it closes below then exit.

I did look into that. Along with a bunch of other schemes & lookback intervals.
Here's the problem:
1/1/2000 to 8/11/2023 (CAGR's):
Untimed TQQQ: -5.4%
52 wk SMA TQQQ: 12.6%

But....
1/1/2000 - 1/1/2010:
Untimed TQQQ: -43.5%
52 wk SMA TQQQ: -4.9%

1/1/2010 - 8/11/2023:
Untimed TQQQ: 38.3%%
52 wk SMA TQQQ: 27.5%

Note: TQQQ inception date was 2/9/2010

The first 10 years was horrible, you had to use timing to not quite keep your head above water.
The next 13 years was great, although timing cut the CAGR drastically.

I do not think that anybody would stick to TQQQ in a period like the 2000-2010 years.

Using QQQ, one can go back to the turn of the century and include the two bear markets before 2011. Using a lookback period of 12 months and weekly trading checks, the system grew $10K into $88K while buy-and-hold only made it to $77K (a 14% improvement).

I get QQQ grows to $52,166 and TQQQ grows to $165,869. 52 week SMA.

But looking closely at my spreadsheet I made a mistake. It evaluates weekly but trades daily, which is wrong.
::sigh:: back to the drawing board.

But looking closely at my spreadsheet I made a mistake. It evaluates weekly but trades daily, which is wrong.

Well, that was an easy fix.

Not much change in the results, though.
52 week SMA, trade weekly
............... QQQ....3X
2000-2023 8.00% 14.73%

2000-2010 4.12% 0.93%
2010-2023 10.95% 26.07%

$10,000 on 1/1/2000 grows to $61,560 for QQQ and $256,632 for 3X QQQ. But all that TQQQ growth came in the last 13 years.
To 1/1/2010 it only grew to $10,967.

But all that TQQQ growth came in the last 13 years. To 1/1/2010 it only grew to $10,967.

But during those 10 years wasn't the benchmark index (QQQ) down some 50%? Significant outperformance in that respect.

DB2

But during those 10 years wasn't the benchmark index (QQQ) down some 50%? Significant outperformance in that respect.


Yes.
$1000 in QQQ dropped to $480. Slightly more than 50% down.

However, timed by 52 week SMA
$1000 in QQQ went to $14,900


B&H CAGR
...............QQQ.......3x
2000-2023 5.89% -5.40%
2000-2010 -6.88% -43.52%
2010-2023 16.39% 38.27%

Timed by 52 week SMA
...............QQQ.......3x
2000-2023 8.00% 14.73%
2000-2010 4.12% 0.93%
2010-2023 10.95% 26.07%

Timing helped during the bad period, but hurt in the good period. C'est la vie.
OTOH, you don't know ahead of time if the next 10 years are going to be good or bad.



By way of comparison, SPY (using the 52 week SMA of SPY)
B&H CAGR
.............SPY
2000-2023 4.86%
2000-2010 -2.46%
2010-2023 10.59%

Timed
.............SPY
2000-2023 4.35%
2000-2010 2.34%
2010-2023 5.85%

Wth SPY, same. timing helped in the bad period, but hurt in the good period.

----------------
FWIW, the QQQ 52-wk SMA has been a buy since 3/20/23.
Before that it was a sell since 4/11/22 except for a 1 week whipsaw the week of 2/6/23.

I took a look at NHNL trading, checking weekly on Friday close, and trading on next-trading-day close.

From 3/15/99 to 3/20/23, I broke the GTR1 NHNL values into deciles, and found total QQQ return within each decile:

1st decile (NHNL values from 5.75 to 15.07) => QQQ return -16.32%
2nd decile (3.99 to 5.47) => QQQ +13.32%
3rd decile (2.73 to 3.98) => QQQ -0.89%
4th decile (1.89 to 2.73) => QQQ +70.37%
5th decile (1.14 to 1.89) => QQQ +10.49%
6th decile (0.32 to 1.14) => QQQ +162.57%
7th decile (-0.76 to 0.32) => QQQ +72.45%
8th decile (-2.21 to -0.76) => QQQ -21.04%
9th decile (-4.98 to -2.23) => QQQ +14.01%
10th decile (-29.53 to -5.00) => QQQ -4.56%

Holding TQQQ when NHNL values are in the 4th to 7th deciles (or possibly even 2nd to 7th) might be something to consider?

2nd decile (3.99 to 5.47) => QQQ +13.32%

Range here should read 3.99 to 5.74. Typo.

Just not sure I'd be able to handle that buy and hold drawdown of TQQQ.

A lot of that gets mitigated if you've put only a fraction of your portfolio in TQQQ. Or any other security or collection of securities.

Eric Hines

A timed TQQQ with a min max NHNLDiff
20100211 till 20230310 with 0.4% friction CAGR 33.5%, GSD 46%, MDD -43.8%, UI 17.6,AT 6.4
Untimed TQQQ same period CAGR 34.6%, GSD 86.5%, MDD -81.7%, UI 25.2, AT 0
Almost same gain but lower GSD, MDD, UI

https://gtr1.net/2013/?~Timed_TQQQ_NHNL_MidValues:...

But caution this is during a strong up market where QQQ had a 16.6% gain over the same period.

RAMc

for those who just want to check current status
https://www.marketinout.com/chart/market.php?bread...
https://www.wsj.com/market-data/stocks/newfiftytwo...
https://www.barchart.com/stocks/highs-lows/summary
https://school.stockcharts.com/doku.php?id=index_s...
https://www.nasdaq.com/market-activity/nyse-52-wee...

Tails and others,

I think I have the settings correct now to view the Google sheet... :-)

Here you go:

https://docs.google.com/spreadsheets/d/1RyyevCoR2h...

Mt code for NHNL in BCC does not match the code in GTR for Dummies. I am guessing there were several variants for the NHNL code. Would you mind posting your GTR1 url for NHNL so that I can see if I can replicate the numbers.

When I use your numbers for one on my screens, I get better results than BCC>0 by not being invested when NHNL is in the top or bottom 10% of its range.

Thanks

Craig

Sure Craig, here is the BCC GTR1 url I've been using. Does this give you what you need?

http://gtr1.net/2013/?!!QlpoMTFBWSZTWVb2ik4AAiLfgB...

lsmr409

Thank you, exactly what I need.

Craih

lmsr409

As usual, I am not certain of what I am actually doing with GTR1.

I just want to make sure I understand the code. Does the "-1" in the field lag and price lag for the url you provide mean that you are trading at the close of the day you are checking?

Also, on 20230310, NHNL has a value of -6.51. Given that there is a -1 in field values, is the NHNL =-6.51 calculated for values from the close of 20230310 to close of 20230228?


Thanks

Craig

Craig, I'm afraid I'm not going to be too much help with the details of what GTR1 is doing with the lags. I may have once known some time ago, but memory has faded...!

In a previous post (#859), I showed some manual research I did on weekly returns of QQQ given NHNL values produced by GTR1 as of Friday close, since in general I knew that these values would be available sometime over the weekend (back when Robbie was able to update data regularly). I then used next-trade-day close prices in calculating QQQ returns. In this way, I was pretty sure I was avoiding hindsight bias. I was interested to see that with QQQ at least, there seemed to be a 'Goldilocks' range of NHNL values, neither too high nor too low, when QQQ performed the best over the following week.

Todd

Post New

Mechanical Investing FAQ

---

Contact Shrewd'm
Contact the developer of these message boards.