Try visiting boards you don't normally visit, starting with the 'All Boards' link above. Or request a new board by posting to 'Any Missing Board Requested Here'.
- Manlobbi
Halls of Shrewd'm / US Policy❤
No. of Recommendations: 2
I am using three screens. The tiny titans, trending value, and reasonable runways. I wonder if it is better to use 3, 6 and 12 months as a look back period, or just 6 and 12 months. I heard that 3 months can be just noise.
No. of Recommendations: 14
Zeelotes beat this subject to death some time ago.
I think he concluded that a "what's working lately" (WWL) approach works best looking at more data and switching less often. A few years, IIRC. But at the time almost all screens were considered only 4 or 5 deep, so you need a lot of data to get any statistical support at all.
Perhaps that lookback would be shorter if you're using extremely deep, and markedly different, screens so that there is some statistical support to your measurement.
e.g., if you're running a 40 stock dividend portfolio or a 40 stock no-dividend growth/momentum portfolio, maybe a short lookback makes sense?
A random example of that effect comparing broad and different portfolios.
Compare the relative performance in the last 3 months of a portfolio of all dividend payers, and of the S&P 500.
If the dividend payers did best, then in the next month, a portfolio of the highest half of dividend payers does 15%/year better than it does when the S&P did best in the prior 3 months.
This is sort of a slow-motion minimal version of Zee's speculative/defensive signal.
If you used that as a signal, checking once per month and going with hidivs or SPY, you'd switch horses 2.7 times a year (1.35 round trips to SPYland) in theory beating the S&P by several percent a year before friction. over 7% in my quick test which was monthly starts only over 38 years.
As for statistical noise, I once created a rule of thumb that 4000/([number of trade dates in a test]*[number of picks in the screen]) will get you the error bar size on the CAGR you measure.
e.g., a 10 stock screen monthly for 10 years would be 4000/(10*120) would be plus or minus 3.3% measurement error as a crude estimate, assuming reasonable turnover. Worse for a screen with low turnover.
Jim
No. of Recommendations: 1
I did my own research, and it looks like 3 month momentum is subject to mean reversion, while 6 and 12 month momentum is more stable. So therefore, 6 an 12 month look back periods are best, with a possible final sort of 2* 26 + 52.
No. of Recommendations: 15
The most effective look back period for relative strength is proportional to the forward holding period.
If you are holding for one year, use RS52 or RS26. If you are holding for 3 months, use RS13.
This applies to very short periods also. If holding for one minute, use the momentum over the last minute.
I don’t have figures in front of me but this follows my research on the subject 25 years back.
If holding for x days into the future, use the relative strength close x days (say 0.5x to 2x days) in the past.
- Manlobbi
No. of Recommendations: 4
A random example of that effect comparing broad and different portfolios.Agree this is a case where "it depends".
I measure the "AVGDEV()" of each screening time period of ROC.
It doesn't answer which time period is better but does indicate how different screens and lookback differ.
example where AVGDEV is very small
6- QEW: Equal-Wght'd ETF's Momentum Equity LT-12m IT-6m IT-3m ST-1m
Tuesday, August 26, 2025 Symbol Style Box ROC252 ROC126 ROC63 ROC21
FT:Dow 30 Equal Weight EDOW Large Value 7.0% 11.3% 6.9% 2.7%
Invesco S&P 100 Eq Wght EQWL Large Value 7.9% 12.4% 5.9% 2.5%
Direxion:NASDAQ-100 EWI QQQE Large Growth 8.1% 17.7% 4.5% -1.5%
iShares:MSCI USA EW EUSA Mid Blend 3.0% 11.8% 6.2% 1.2%
Invesco Rus 1000 EW EQAL Mid Blend 0.6% 10.5% 6.6% 2.3%
Invesco S&P500 EWght RSP Mid Blend 1.4% 11.0% 6.0% 1.8%
AveDev = 3.0% 1.7% 0.5% 1.1%
GD_
No. of Recommendations: 2
This is my research.
A 3 month look back period is not recommended because it is subject to reversion to the mean. It seems that 6 months and 12 months are better. This is from Deep Seek.
Practical Example
Imagine two stocks:
Stock A (The "Hot Stock"): Up 25% in 3mo, Up 30% in 6mo, Up 40% in 12mo.
Stock B (The "Steady Winner"): Up 0% in 3mo (flat), Up 25% in 6mo, Up 50% in 12mo.
Under your old screen:
Stock A passes all three filters. Stock B is eliminated immediately by the 3-month filter, even though it has fantastic 6 and 12-month performance. You are forced to buy Stock A, which carries higher risk of a pullback.
Under the new screen:
Both stocks pass the 6 & 12-month filters. Now, your composite score decides:
Stock A: (2 * 30) + 40 = 100
Stock B: (2 * 25) + 50 = 100
They have the same composite score! This screen would allow you to buy the steady winner (Stock B) that has demonstrated excellent long-term momentum but may have simply paused or consolidated in the most recent quarter. This is often a much healthier entry point.
Conclusion: Your composite momentum formula is very well-designed. The flaw in your process is the mandatory 3-month filter. By removing it, you fix the structural weakness of your strategy while strengthening its focus on high-quality, sustainable momentum. This change is highly likely to improve your risk-adjusted returns over time.
I guess the crux of it, is that you don't want stocks that are rocketing up, but ones that are increasing slowly and steadily.
No. of Recommendations: 3
Conclusion: Your composite momentum formula is very well-designed. The flaw in your process is the mandatory 3-month filter. By removing it, you fix the structural weakness of your strategy while strengthening its focus on high-quality, sustainable momentum. This change is highly likely to improve your risk-adjusted returns over time.
I guess the crux of it, is that you don't want stocks that are rocketing up, but ones that are increasing slowly and steadily.
Why not use a weighted average? Any advantage?
sum(12m*4,6m*3,3m*2,1m*1)/10
GD_
No. of Recommendations: 7
A couple of observations:
The optimum lookback for WWL (picking among screens) is decidedly NOT the same as the optimum lookback for ranking stocks within a screen.
The optimum lookback for stocks within a screen varies with the screen, since different styles of investing have different user populations. e.g., sales growth types versus high yield types.
I think one of the main factors for the variance among styles is that according to some research, and a few pokes at it myself, the optimum lookback varies as (among other things) a function of the turnover ratio of the stock(s) in question. The research noted that a higher turnover ratio means that the weighted average entry price among holders willing to trade is different, leading to a different strength of disposition effect, which seems to be the main reason that momentum exists as a factor. People being too hesitant to sell losers, and too quick to sell winners.
Jim
No. of Recommendations: 1
The optimum lookback for WWL (picking among screens) is decidedly NOT the same as the optimum lookback for ranking stocks within a screen.
The optimum lookback for stocks within a screen varies with the screen, since different styles of investing have different user populations. e.g., sales growth types versus high yield types.
Thanks for that, it helps to keep my focused.
GD_
No. of Recommendations: 1
That would also work.
No. of Recommendations: 7
I am using three screens. The tiny titans, trending value, and reasonable runways.Do you mean the O’Shaughnessy Reasonable Runaways stock screen? Are the steps similar to Cornerstone Growth (mutual fund HFCGX)?
Define {zgHFCGX2023}
step0: ordinary U.S. stock
step1: no IPOs
step2: [MktCap] > 175
step3: recent 10Q filed
step4: [PS] < 1.5
step5: year over year earnings increase
step6: [Total Return % over 63 days] > 0
step7: [Total Return % over 126 days] > 0
step8: [Total Return % over 252 days] Top 50
https://gtr1.net/2013/?~zgHFCGX2023:h252f0.4::styp...
No. of Recommendations: 3
Close.
1. Market Cap > 250M
2. P/S <1
3. price change 26 week > 0
4. Price change 52 week > 0
6. Tost = 2 * 26 week relative strength + 52 week Relative Strength
7. Tost > 0
8. final sort by descending by tost
No. of Recommendations: 5
P123 Version: (slightly modified)
Universe($ADR)=FALSE AND Country("USA") AND Universe(NOOTC)
Universe(MasterLP)=FALSE
#APeriods > 2 //No IPOs
AvgDailyTot(20)>500000 // Enough liquidity to trade
LatestActualDays < 60 //Recent 10Q filed
Pr2SalesTTM < 1.5 & Pr2SalesTTM != NA
EBITA > EBITPY
Ret3M%Chg > 0
Ret6M%Chg > 0
Ret1Y%Chg > 0
FOrderOLD("Ret1Y%Chg",#All,#Desc,#Previous,TRUE) <= 8
Five Year Backtest (Yes, I know that's not long enough, but it's the subscription I currently have.)
Screen 25% Annualized, -50% Max DD, 1.41 Beta, 0.97 Sortino
S&P500 14% Annualized, -33% Max DD, 1.0 Beta, 0.94 Sortino
No. of Recommendations: 6
The most effective look back period for relative strength is proportional to the forward holding period. If you are holding for one year, use RS52 or RS26. If you are holding for 3 months, use RS13.
This implies that for the canonical monthly hold the best look back would be four weeks. That certainly hasn't been my experience over the years.
DB2
No. of Recommendations: 12
This implies that for the canonical monthly hold the best look back would be four weeks. That certainly hasn't been my experience over the years.
The effective look-back period for momentum is proportion to the forward return in which it works, rather than equal to, though I gave examples with them being equal, I apologize.
You would intuitively expect the look-back period (for momentum) to be at least twice as long than the holding period (and more meaningful at 3-5 times longer or so) such that we are actually capturing 'part' of the momentum. (It sort of doesn't make sense if the 'part' is longer the whole).
In any case, the holding period rises in proportion to the relative strength loopback period.
Holding for 1 minute - use momentum the last 4 minutes
Holding for 1 week - use momentum the last month
Holding for 1 month - use momentum the last 4 months
Holding for 6 months - use momentum the last 24 months
Holding for 12 months - use momentum the last 4 years
- Manlobbi
No. of Recommendations: 6
The most effective look back period for relative strength is proportional to the forward holding period.
This implies that for the canonical monthly hold the best look back would be four weeks. That certainly hasn't been my experience over the years.
In time series statistical analysis, the rule of thumb is that you can predict with some reliability one-fourth to one-third of the time frame for which you achieve stationarity on the past (stationarity: jargon-speak for getting a stable pattern from the data). Thus, if you have a statistically reliable pattern over 12-month intervals, you can reliably predict 3-4 months into the future.
Eric Hines
No. of Recommendations: 3
If you are into 1-week holds you want to exclude last week's momentum if you are using 1 month backward momentum "up" as a good thing; this gives you a 10 times higher return on the Nasdaq100, the S&P500, and 2 times on WER stocks, than if you include last week. For short term trading you want a bad last week, not a good last week.
