Halls of Shrewd'm / US Policy
No. of Recommendations: 12
A similar screen mentioned by mungofitch last year.
"Speaking of KISS, try this:
Member of Nasdaq 100
price / (52 week high + 52 week low) top N"Google (AI): describe price/((high+low))
The formula {Price}/High+Low} calculates the relative position of a current price within a period's trading range.
It normalizes the asset's current price against its absolute price boundaries, making it highly useful for measuring volatility and trend strength in financial markets.
The expression resolves into distinct ranges depending on the price action:
Near the High: If the current price is close to the session high, the sum of the top and bottom values is approached, causing the ratio to approach ~0.5.
Near the Low: If the price is hovering near the low, the sum of the limits grows relative to the price, dropping the ratio closer to ~0.25 or lower.Similar backtest results. There is a lot of overlap between this and the absolute 52 week momentum screen.
Today's picks in rank order
-this- -52wk momentum-
SNDK SNDK
MU LITE
WDC WDC
STX MU
AMD STX
ARM AMD
LRCX LRCX
AMAT MRVL
INTC KLAC
WBD MPWR
No. of Recommendations: 10
The formula and the summary don't match?
This is comparing price to its 52 week range boundaries:
price / (52 week high + 52 week low) top N"
But this mentions "session" high, which would normally mean a single day's trading session?
If the current price is close to the session high, the sum of the top and bottom values is approached, causing the ratio to approach ~0.5.
By the way, I think this worked quite well in both 2024 and 2025. CAGR staring with a 3 I think? Depending on depth and hold, of course.
Similar backtest results. There is a lot of overlap between this and the absolute 52 week momentum screen.
Makes sense.
The quirk of using the midpoint of high and low is that "off low" works best when the market is rebounding from a sharp selloff, and "off high" works most of the rest of the time, so it can sometimes give you something that works a larger fraction of the time rather than just a high CAGR momentum that relies on fewer good periods.
Jim
No. of Recommendations: 4
But this mentions "session" high, which would normally mean a single day's trading session?
It's coming from an AI, which is likely responding to what most of its inquiries are -- short term. Other hits on my googling steered me to long treatises on "Average True Range", which focuses on "typically 14 days".
I take "session" to mean the period of interest, which for this strategy is 52 weeks.
FWIW, I thought that "current/(hi + lo)" (or "current/((hi + lo)/2)") would have the same rankings as "(current - lo)/ (hi - lo)"*, because they both are measuring where the current price sits in the 52 week range.
But...the rankings are not the same.
using the midpoint of high and low is that "off low" works best when the market is rebounding from a sharp selloff, and "off high" works most of the rest of the time,
Seeing as the average CAGR is around 22% (assuming the GTR1 backtests are right) you don't need to care about "gains coming off low". You don't need to capture them. All you need is some simple timing scheme to help you avoid the worst of the big drawdowns. Control your risk and the gains will take care of themselves.
-----------------
* Which is what popped up when googling "calculate current price in hi low range".
No. of Recommendations: 1
please post gtr forms for testing and alteration
No. of Recommendations: 2
No. of Recommendations: 4
No. of Recommendations: 3
No. of Recommendations: 3
No. of Recommendations: 0
Examining it in chatgpt.
180 days in the time period I have available there reduces Ulcer Index to 14 and only slightly reduces CAGR with mdd to -28.89%
No. of Recommendations: 6
mdd -62.96
That BCC step isn't doing much of anything.
The purpose of timing is to avoid the full brunt of a large decline, not to improve the return.
What's the purpose of the 3 day lag?
Although I could see that might help you avoid some stock like LITE which had a 1 day gain of 11% when it suddenly popped on the screen at #2 position at 1026% in 52 week momentum.*
Lag 1 or lag 3 doesn't have much difference in the backtest. Harder to do in real life, too. But I guess you could run the screen on Monday and do the trades on Thursday.
I have fantasies of trading this 5 stock portfolio when the portfolio has grown to $500,000. ;-)
Merrill already gives me grief for many of the stocks & ETFs I try to buy. Some they just flat-out refuse to let you buy. I wonder what they'd think when a private investor places a $100,000 buy order. I expect to have the phone ring about 10 seconds after I clicked the SUBMIT button, them asking if I did it or if some scammer got into my account. (That happened once at BrownCo, when I accidently clicked SELL instead of BUY.)
----
* I went to buy it on the 21'st, it opened at 870, put in a bid inside the spread at 903 and watched it rise steadily all day long, running away every time I increased my bid. Never got it, it topped out at 977.
I got pissed at myself, missing out due to being greedy and trying to buy well below the ask. Almost just went with the next stock, then I remembered "Discretion over buy-and-sell decisions in aggregate can turn a model that generates a market beating return into a sub-par return. IOW, follow the signals religiously!"
So I waited a few days, saw the drop back toward 900, bought at 921, then kicked myself when it dropped down to 810.
Yeah, and today it closed at 1029.
No. of Recommendations: 4
Examining it in chatgpt.
180 days in the time period I have available there reduces Ulcer Index to 14 and only slightly reduces CAGR with mdd to -28.89%
I think we need to be careful in the numbers we look at. The first link you posted had CAGR's from 26.2% to 31.6%. UI from 14.7 to 21.7
It is too optimistic to assume that the real-life portfolio would get anything but the worst figures. It is virtually impossible that it would get the averages.
Especially since the averages are for independent cycles. The min/max CAGR probably won't be the same cycle as the min/max UI or min/max MDD.
What I do is download the full "daily portfolio values" and pick out the cycle with the worst ending value and use that one for all my investigation and statistic computations. It's a bit more work, but I'd rather expect 16% and get 22% than expect 22% and get 16%. And for detailed work I trust my Excel skills better than I trust a LLM chatbot.
No. of Recommendations: 5
I look across regimes and many drawdown, crashes, corrections, recessions periods to see extent and recovery times compared to NDA/QQQ and SPY (which are the main alternatives for cheap exposure to many US stocks at once). I am testing to verify when gtr numbers are not repeated in Norgate data and AI analysis and when they are (generally they are though today I found an anomaly). I test and test and test.
But results can be better than expected, too, as in in the last 8 months. #6 pick in the strategy was up 30% today, e.g., and it is not alone; the period since October has been gangbusters, far better than the strategy would have suggested likely.
I lived through the 87 30% crash, the tech 70% crash, the 2000 and 2002 disasters, the covid crash so I realize bad things happen to good people.
But I believe in data; that is all there is. I am not Long Term Capital Management; I realize the past does not prepare you for all eventualities. But it partially prepares you, and using it enabled me to retire at 40 without much money so I am still crunching numbers.
No. of Recommendations: 3
by buying the overlaps (averaging 2.4 stocks) only between the 9 month momentum top 5, and the top 5 of price/ (high price in last 9 months + low price in last 9 months), both using my suggested SPY>325 day SMA rule, you would have recovered from the worst crash since 1999 in 130 market days (-56%) and had a CAGR of
Post-1999 summary:
Series CAGR MDD Sharpe Beta vs SPY UI mDD recovery delay
Overlap-only 28.4% -57.0% 0.87 0.82 15.6 130days
Timing rule would have kept you out of the market in 2000 and 2001, and drawdown in 2008 was -20% and it would have taken 736 market days to recover compared to 888 for SPY.
No. of Recommendations: 2
by buying the overlaps (averaging 2.4 stocks) only between the 9 month momentum top 5, and the top 5 of price/ (high price in last 9 months + low price in last 9 months),
Does this mean only the ones that are in both screens, or the ones that are on either screen?
If the former, there could be periods (probably rare) when you owned NO stocks. But anyway you could own anywhere between 0 and 5 stocks.
If the latter you could be owning anywhere between 5 and 10 stocks.
Do you equal weight every month or only EW the new stocks?
I could see operational PITA when increasing or decreasing the number of stocks.
I could see another interesting variation to test. First the top 25 or 33 stocks from one screen and then the top 5 or 10 from those by the other screen. (Now we are starting to get silly.)
Just a casual comment ... we are deep into nosebleed territory with any of these variations. Hard to believe this is realistic, long-term. But it appears to have lasted 40 years, so for me it only need to last another 10-15 years.
No. of Recommendations: 2
I could see another interesting variation to test. First the top 25 or 33 stocks from one screen and then the top 5 or 10 from those by the other screen.
No joy. There must be so much overlap that it does not matter. Both variations pick almost the exact same stocks.
This was no HTD and no timing.
step2: [[Actual closing Price; share_lag=1 days; quote_lag=1 days]/[Actual closing Price; share_lag=1 days; quote_lag=252 days]] Top 5; Cash When Fewer
CAGR 26.227407
GSD 43.084263
UI 15.380290
MDD -74.043823
AT 3.675156
step2: [[Closing g-price; quote_lag=1 days]/[[Highest closing g-price over 252 days; lag=1 days] + [Lowest closing g-price over 252 days; lag=1 days]]] Top 5; Cash When Fewer
CAGR 26.522161
GSD 42.430981
UI 14.554314
MDD -73.333794
AT 4.969617
step2: [[Actual closing Price; share_lag=1 days; quote_lag=1 days]/[Actual closing Price; share_lag=1 days; quote_lag=252 days]] Top 30
step3: [[Closing g-price; quote_lag=1 days]/[[Highest closing g-price over 252 days; lag=1 days] + [Lowest closing g-price over 252 days; lag=1 days]]] Top 5; Cash When Fewer
CAGR 26.966204
GSD 42.214531
UI 15.020716
MDD -74.518349
AT 4.947662
step2: [[Closing g-price; quote_lag=1 days]/[[Highest closing g-price over 252 days; lag=1 days] + [Lowest closing g-price over 252 days; lag=1 days]]] Top 30
step3: [[Actual closing Price; share_lag=1 days; quote_lag=1 days]/[Actual closing Price; share_lag=1 days; quote_lag=252 days]] Top 5; Cash When Fewer
CAGR 26.461454
GSD 41.920460
UI 15.040699
MDD -73.663803
AT 3.895647
No. of Recommendations: 6
Given that those two momentum-ish screens are almost identical, there is no reason to use both. It's just two slightly different way to measure the same thing.
I mentioned this screen on a different universe of stocks in post #5524. That universe is roughly the Russell 1000, filtered by some quality metrics that mungofitch has mentioned. Sales growth, Return on Assets, etc.
Mkt Cap above 2 Billion (mid cap+) is the top 978 of the Russell 1000.
Mkt Cap above 10 Billion (large cap+) is the top 676 of the Russell 1000.
With this set of metrics, the #20 position by 52 week momentum is:
Mid cap+: 271%
Large cap+: 211%
Nasdaq 100: 91%
Russell 2000: 510%
Link to screener:
https://finviz.com/screener?v=141&f=an_recom_buybe...
No. of Recommendations: 4
"by buying the overlaps (averaging 2.4 stocks) only between the 9 month momentum top 5, and the top 5 of price/ (high price in last 9 months + low price in last 9 months),
Does this mean only the ones that are in both screens, or the ones that are on either screen?"
Only the overlaps was best.
"If the former, there could be periods (probably rare) when you owned NO stocks. But anyway you could own anywhere between 0 and 5 stocks.
If the latter you could be owning anywhere between 5 and 10 stocks."
Yes the average was 2.5 stocks in an "overlaps only" screen.
"Do you equal weight every month or only EW the new stocks?
I could see operational PITA when increasing or decreasing the number of stocks."
The test pretended equalization of dollars every month.
"I could see another interesting variation to test. First the top 25 or 33 stocks from one screen and then the top 5 or 10 from those by the other screen. (Now we are starting to get silly.)"
Results degraded adding more stocks.
"Just a casual comment ... we are deep into nosebleed territory with any of these variations. Hard to believe this is realistic, long-term. But it appears to have lasted 40 years, so for me it only need to last another 10-15 years."
My latest version tests to 12/31/99 at 31% CAGR. Worst drawdown -56 same as maret with almost 4X return and faster recovery.
No. of Recommendations: 1
What's the risk off asset, 0% yielding cash?
No. of Recommendations: 2
Risk off asset is a very short-duration income thing like SGOV or BIL or NEAR or ICSH.
SPAXX would also be okay.
Basically, cash equivalent which pays a little income with essentially no risk.
In my timing backtests I assume cash-cash which has no yield. The above would pay you 2%-4% while you wait.
