Post New

For those few of you doing weekly trading this is a weekly-traded strategy to buy 5 stocks from the Nasdaq100 with
a CAGR over twice the Nasdaq100's 11.5% and S&P's 12.5%,
with a better beta than both (.8 v. 1.2 for the Nasdaq100 and 1 for the S&P)
and a sharpe of 1 (.975 v. .57 for the Nasdaq100 and .44 for the S&P).
-51% max drawdown is also better than both Nasdaq100's -80.7% and S&P 59.7%.

You take the Nasdaq100's stocks
Narrow to 1 year top 15 momentum.
Narrow by requiring two day rsi under 50.
Narrow by requiring one week return under 1%.
Then keep 5 by lowest ratio of 1 week/1 year return.

Cycle: 0 1 2 3 4
Bottom Positions Held Variant Avg 1st Trades: 19850201 19850204 19850205 19850206 19850207

5 0 27.254822 27.962633 28.401411 27.590551 26.103434 26.216082
28.186230 26.731035 28.413044 27.865431 28.399183 29.522448
0.975264 1.038504 1.005030 0.991735 0.934993 0.906058
15.842601 15.148137 15.700482 15.547487 16.144173 16.672726
11.254422 9.788255 10.057002 13.682628 10.784611 11.959613
0.802961 0.821991 0.780715 0.826275 0.776602 0.809222
30.174393 29.798952 32.243382 29.559303 30.034771 29.235546
23.835537 23.560780 23.507339 24.242435 23.966789 23.9003
https://gtr1.net/2013/?h5f0.10000::nas100.a:nenull...

I took the old ones I already reported on and was curious if we could get the beta down below the S&P, and did, and still have a good sharpe, drawdown (relatively speaking of course), and excellent CAGR.

For those few of you doing weekly trading this is a weekly-traded strategy to buy 5 stocks from the Nasdaq100 with
a CAGR over twice the Nasdaq100's 11.5% and S&P's 12.5%,
with a better beta than both (.8 v. 1.2 for the Nasdaq100 and 1 for the S&P)
and a sharpe of 1 (.975 v. .57 for the Nasdaq100 and .44 for the S&P).
-51% max drawdown is also better than both Nasdaq100's -80.7% and S&P 59.7%.

You take the Nasdaq100's stocks
Narrow to 1 year top 15 momentum.
Narrow by requiring two day rsi under 50.
Narrow by requiring one week return under 1%.
Then keep 5 by lowest ratio of 1 week/1 year return.

 			Cycle:	0	1	2	3	4
Bottom Positions Held	Variant	Avg	1st Trades:	19850201	19850204	19850205	19850206	19850207
 
5	0	27.254822		27.962633	28.401411	27.590551	26.103434	26.216082       cagr
 		28.186230		26.731035	28.413044	27.865431	28.399183	29.522448       gsd
 		0.975264		1.038504	1.005030	0.991735	0.934993	0.906058        sharpe
 		15.842601		15.148137	15.700482	15.547487	16.144173	16.672726       ldd
 		11.254422		9.788255	10.057002	13.682628	10.784611	11.959613       ui
 		0.802961		0.821991	0.780715	0.826275	0.776602	0.809222        beta
 		30.174393		29.798952	32.243382	29.559303	30.034771	29.235546       TR
 		23.835537		23.560780	23.507339	24.242435	23.966789	23.900339       AT
 

https://gtr1.net/2013/?h5f0.10000::nas100.a:nenull...

I took the old ones I already reported on and was curious if we could get the beta down below the S&P, and did, and still have a good sharpe, drawdown (relatively speaking of course), and excellent CAGR.

https://gtr1.net/2013/?h5f0.10000::nas100.a:nenull...

https://gtr1.net/2013/?h5f0.10000::nas100.a:nenull...


But 15.7% from 20030103 to 20251128
And 15.9% from 20100104 to 20251128
And 13.9% from 20150102 to 20251128

Having looked over the years at several other screens that used very short periods for holding and lookback, they all seemed to work by luck.

In real life, you can't get the prices that the backtests assume. This makes a big difference; a few cents in a trade that lasts for a week vs. a few cents in a trade that lasts for a year. Or even a month.

BTW, if you eliminate the 2nd and 3rd rules, 2003-2025 is 18.9%. 19.8% if you never rebalance existing positions.
And 18.3% if the period is a month instead of a week (21 days vs. 5 days).
Then 18.3% if the only rule is "tr(1,253) top 15" (21 days), or 19.8% (5 days).

Year ago I did an experiment; I used the strategies we talk about here weekly in one account and monthly in another.
My weekly strategy did better. Every year.
It wasn't luck.
Rsi2, rsi3, 1 week bad return, connorsrsi2 all work in multiple universes over long time periods.
It is what enabled me to quit work at age 40.
If it is "luck" it is luck virtually every year.
After discovery and before.

Only updated through the end of November, but the screen currently selects no stocks.

Not a bad thing; it averages 2.9 stocks. Maybe excluding stocks is how it beat the market while taking less risk than the market.

You can get the average up to 4.1 stocks by increasing the 2nd step to tr(1,253) top 35.

This gives a CAGR of 31% with a sharpe of 1 and a beta of 1 and mdd of -54% and still beats the Nasdaq's and the S&P's sharpe, beta, dradown, and doubles their CAGR.

You can bring the average # of stocks to 4.5 by bumping tr(1,253) to top 50. CAGR 32%.

You can get 4.78 average stocks bumping it to tr(1,253) top 75 stocks. Still 32% cagr, 1 sharpe; beta goes to 1.2.

If you read the academic and practitioner literature for decades now they confirm that short term momentum reversal is consistent, universal across asset classes, capturable, persistent. You can take screens people use here at at final steps based on rrs, rsi, tr(1,5), connorsrsi or other screening criteria on the universe of your choice and see it yourself: many screens, even with the spreads paid weekly, more than justify weekly trading. I am now up 120% in MU in 4 months as a result.
It isn't "chance" when you can test back to 1985 to verify the result. Zacks universe, WER, Nasdaq100. Take your pick as your universe.
I was serious that this board and this technique let me trade lawyering 28 years ago for working fewer hours per week with this method. It won't eliminate crashes mostly but good years more than make for it, like last year. Thanks to Robbie you can run tests of your choice, and if you use online universes/screeners you can find picks on your own since I assume he will get tired some day.
The sources of info are not always in agreement, e.g. I ran one on the Nasdaq100 1 week ago and some online sources had not yet added the latest addition to the index.

and you can get to 4.9 average stocks picked by skipping the tr(1,253) step completely and still get 32% CAGR, 1.2 beta, 1 sharpe.

The point wasn't to say this is the best screen; it is to stimulate discussion so someone tries to generate good ideas to test, or comes up with a new screen.

That is how I have made money, and what the board is for. I realize not many people have ever been into weekly screens here, and market-timing is something that generates more posts than screens and has for years here.

Thanks for posting another interesting screen with suggestions (weekly trading, short-term momentum reversal). The OP screen looks for stocks:

Large Cap Growth (Nasdaq 100)
high 1-year momentum
low 1-week momentum (short-term momentum reversal)
weekly trading

The short-term momentum reversal steps (steps 2 and 3) reduce CAGR, Beta, drawdown, and depth, while more than doubling the number of trades (AT). (Depth is the average count of stocks passing the final step.)

Results from 19850201 to 20251128 (longest backtest):

    Screen      CAGR  SAWR  GSD  DIGSD  LDD  LDDD3  MDD  UI  Sharpe  Beta  TI  AT  Depth
  OP_screen      27    12   28     32   16     8    -51  11   0.98   0.8   30  24   2.9
   no_step2      28    12   33     37   19     11   -55  16   0.91   1.0   26  27   3.8
   no_step3      31    13   34     37   18     11   -60  17   0.97   1.0   27  27   3.9
no_step2_step3   30    12   42     46   23     15   -74  22   0.84   1.4   21  9    5.0

Results from 20151127 to 20251128 (most recent 10 years):

    Screen      CAGR  SAWR  GSD  DIGSD  LDD  LDDD3  MDD  UI  Sharpe  Beta  TI  AT  Depth
  OP_screen      14     9   25     29   16     10   -39  12   0.60   0.8   17  22   2.8
   no_step2      12     7   30     35   19     14   -51  21   0.49   1.0   13  26   3.7
   no_step3      17    10   31     35   19     13   -47  17   0.66   1.1   17  26   3.8
no_step2_step3   23    11   40     45   23     15   -50  22   0.73   1.5   17  9    5.0

I prefer the version with higher CAGR and less trading. Low Beta would help when blending with other screens, but this screen's CAGR is so high that I prefer to just evaluate it on its own.

In the longest backtest, the OP_screen has 90% of the CAGR while holding 40% cash on average (compared to no_step2_step3). This is a surprising result, but does not hold in the most recent 10 years, when the OP_screen CAGR is much lower than no_step2_step3.

The idea of using technical indicators as short-term market timing signals reminds me of some of Zeelotes' posts:

Author: zeelotes
Subject: Managing Risk with Saul Picks
Date: 12/30/22
"Back on 11/23/2021 I posted on the Saul Investing board, and here on the Mechanical Investing board a relatively simple way to manage risk by using overbought / oversold indicators built off of the Software as a Service sector. The method finds times when these four groups of stocks become way overbought to produce a sell and way oversold to produce a buy. ... What you need is to find technical analysis indicators that identify points of extreme on the overbought and oversold sides. You do this for every stock in the universe and then average their readings to arrive at a consensus. Once you've done that you simply determine thresholds for buy and sell. ... "
https://www.shrewdm.com/MB?pid=95445777

Author: Zeelotes
Subject: Saul Investing - Managing Risk!
Date: 1/14/2022
"... The point of that post was to say there is a way to use overbought/oversold indicators applied to the individual stocks identified to pinpoint times of risk and times of opportunity. ... "
https://yorickm.com/Message.php?pid=35025423

Author: Zeelotes
Subject: Managing Risk in Volatile Picks
Date: 11/23/2021
"... Years ago I discovered that it was relatively easy to take a basket of stocks and determine points when they were excessively overbought and when they were excessively oversold. ... Using cash during the dangerous periods results in being invested in the more volatile investments just 69.71% of the time. In other words, you sit in cash about 30% of the time. ..."
https://yorickm.com/Message.php?pid=34986288

Are there specific overbought / oversold indicators, or sets of them, that anyone has used profitably on volatile stocks that they’d be willing to share, either with the board or by private reply? :-)

You can get the average up to....

That's seems like a fairly broad mound of toast.

Eric Hines

You can triple the Nasdaq100 by using Zacks Ranked #1 stocks and trade weekly and get a beta of 1 and double the sharpe of the Nasdaq100. 199912 forward, 5 stocks weekly with .2 spread/trade:
z rank=1
ratio(adv(1,5),adv(1,20)) top 10
ratio(tr(1,5),tr(1,253)) bottom 5

https://gtr1.net/2013/?h5f0.20000::rank.z:et1:rati...

 nasdaq100 v. strategy  v. S&P500
cagr  7.8       v.  23.5        10.4
sharpe .35           .78          .52
beta  1.3            .98         1.09
mdd -80.4          -59.0       -59.7
2025                38% thru 11/28, robbie's test max

So you can double the S&P and triple the Nasdaq100 with less risk assuming the backtest repeats of course going forward.

Reminds me of this strategy from ancient MI times:
https://yorickm.com/Message.php?pid=13845697

You can triple the Nasdaq100 by using Zacks Ranked #1 stocks and trade weekly... cagr 23.5

Zacks claims "Zacks Rank #1 average gain of +23.9% per year from January 1, 1988 to December 1, 2025"

Oh, "usually numbering around 200 to 250 out of over 4,000 stocks."
So a pretty vacuous, vague claim.

But you have to subscribe to see the list. Subscription: $249/year.

Zacks' internal backtesting is with 20-20 hindsight since it only releases new Zacks rank and other data changes over the weekend so you can use it until Mondays, but its backtesting pretends you had it all at market close Friday and their backtests pretend you traded with the new Zacks rank at Friday's close.
When I hand-tested strategies with real-world stock prices it cut Zacks the company's strategies' returns in half.
But with Robbie's tester you are getting what you code for not 20-20 hindsight.
And my example here assumed a 1 day delay before you trade so it should be avoiding Zacks' backtester's issues.

trading the zacks strategy monthly also works but ups the maximum drawdown:
<pre
average the worst start
cagr 22.87 17.4
gsd 36.2
sharpe .75 .59
ldd 20
ui 18.9
beta 1.0 1
ti 23.25
at 12.38
mdd -66.0 -75.4

Warning: I can't get adding daily dollar volume and price criteria to work correctly; when I add them to screens I am getting weird results that can't be correct.

19991229-end of Robbie's tester date 20251128 , assumes .2% spread each trade, pick five stocks

 nasdaq100 v. strategy  v. S&P500
cagr  7.8       v.  27.6        10.4
sharpe .35           .87          .52
beta  1.3           1.04         1.09
mdd -80.4          -59.0       -59.7
2025                32.7% thru 11/28, robbie's test max


I modified the former screen I submitted during this thread, gaining 4% CAGR with same sharpe and beta
and max dd as previous one.

zack rank 1
ratio 1 wk daily dollar average volume/20 day top 15
ratio 1 wk return/ 1 year bottom 10
ratio Market Cap/float bottom 5

https://gtr1.net/2013/?h5f0.20000::rank.z:et1:rati...

Holding monthly instead is still excellent:
cagr 25.45  twice the S&P and 3 times the Nasdaq100
sharpe .79  beats both
beta  1.03  beats Nasdaq100, same as S&P's
mdd -61.82  2% worse than S&P's, 18% better than Nasdaq100's
2024  7.18

https://gtr1.net/2013/?f0.20000::rank.z:et1:ratio%...

Post New

Mechanical Investing FAQ

---

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