No. of Recommendations: 2
To DrBob2 , Bill2m , or Jamie Gritton, or anyone else with insight,
I am thinking of using the “Up_5%” screen and have been reviewing the backtest data on the Jamie Gritton’s Backtest URL https://backtest.org/ .
When I run the backtest, the stocks his backtest site picked on the first week of each month do not seem to match the same stocks you report in your weekly published stocks for the same screen.
For example, in recent months, on Bill2m's weekly published picks for this screen,
On Nov 4, SI Rankings published Stocks IEG, PZN, FCBC, TREE, and OPBK.
Where Jamie's back test reports using on Nov 5 stocks DAVA, SNBR, FIX, FRGI and PMM for its first five picks.
For December, SI Rankings published GTN, FN, MTLS, STRL, FBR on Dec 1 , whereas Jamies backtest used GTN, FOXF, FN, CASY, and AMEH for its first five stocks in the backtest.

I was wondering what the explanation might be for the different stocks selected using the same Screening criteria. Are Jamies backtest done on different database?

Thanks for any insight that can be provided.

Lester
Print the post Back To Top
No. of Recommendations: 0
Are Jamies backtest done on different database?

Yep. Jamie's backtester is based on data from one particular company that shall remain nameless, and the SI rankings are based on AAII's stock investor pro data.


Mark
Print the post Back To Top
No. of Recommendations: 0
On Nov 4, SI Rankings published Stocks IEG, PZN, FCBC, TREE, and OPBK.
Where Jamie's back test reports using on Nov 5 stocks DAVA, SNBR, FIX, FRGI and PMM for its first five picks.
For December, SI Rankings published GTN, FN, MTLS, STRL, FBR on Dec 1 , whereas Jamies backtest used GTN, FOXF, FN, CASY, and AMEH for its first five stocks in the backtest.

I recall/believe (and yet could be very wrong) - that to avoid potential copyright issues - the backtest's reported stocks are 'annonymixed'. That is the underlying data points to the correct stock that the screening would show, however the backtest stock tickers uses a mapping to a scrambled database.

Any corroboration or refutation would be appreciated - as much for my own learning.
Print the post Back To Top
No. of Recommendations: 9
Are Jamies backtest done on different database?
Yep. Jamie's backtester is based on data from one particular company that shall remain nameless, and the SI rankings are based on AAII's stock investor pro data.


If by nameless you mean Lord Voldemort, no my backtester's done with that. I ceased and desisted that data, as I encourage anyone still hanging on as a customer of their to likewise cease and desist.

My current data source is quite compatible with the SI picks, so I would expect the picks in question to be nearly the same. There will always be a certain amount of discrepancy, since my price data doesn't include OTC stocks or units (and SI seems to have a number of the latter). But given a relatively simple screen like Up_5% that only contains one sorting rule at the end, the only difference you should see is if those OTCs or units show up in the rankings list and not in my own list. You should never see something there backtest.org picks something that the rankings didn't.

Unfortunately, that doesn't really count as an insight - I'm confused by this myself. Bill2m - we ought to get together and see why we're coming up with different results on this screen. Perhaps we'll discover something that leads to a more reliable backtest.

- Jamie
Print the post Back To Top
No. of Recommendations: 1
From DrBob2 MI Post Number: 159304 on 18 Mar 2004:
Here's a GARP screen that just keeps on ticking. Inspired by a SIPro screen, this screen looks for companies where the earnings estimate for the next quarter has been increased by at least 5% during the last four weeks. It adds a value requirement (P/CF < 20) and sorts by 26-week return.

The screen (Q1 refers to the next quarter):
1) Price > 10
2) Average daily volume > 20K
3) EPS Est. Q1 > 0
4) EPS Est Q1 last month > 0
5) EPS Est Q1 > EPS Est Q1 last month x 1.05
6) # Estimate revisions up > 0
7) # Estimate revisions down = 0
8) P/CF > 0
9) P/CF < 20
10) Sort by high 26-wk return
Enjoy.
DB2

++++
Also , From DrBob2 MI Post Number: 207943 on 08 Mar2008:
The Up 5% screen now has four years of post-discovery data.
For a 5-stock monthly the CAGR is 37% with a GSD of 33 (S&P 500 is 9 and 8).
+++
Thanks for replies from Jamie and the rest regarding this.
Just wondering if the above are still the screening parameters being used in the current weekly stocks being published by Bill2m and per Jamies backtest site, or have the parameters been altered any over the years??

Again, thanks for clarifications.

Lester
Print the post Back To Top
No. of Recommendations: 4
Just wondering if the above are still the screening parameters being used in the current weekly stocks being published by Bill2m and per Jamies backtest site, or have the parameters been altered any over the years?

Don't know about changes, but I've sometimes thought that I should have used a 7% increase as the cutoff. Then I could have called the screen 7Up.

DB2
Print the post Back To Top
No. of Recommendations: 7
The screen definition has not changed. SIP changed the definition of their Cash Flow Per Share field on May 14, 2010. The SIP version of Cash Flow is used in the Rankings.
https://boards.fool.com/recovering-si-pros-cash-flowshare-30...

Today's picks for each screen are:

Up_5pct  Up_5pct_badCF_SIP  SI_Rankings
BSBR BSBR BSBR
VIRT FBP FBP
FBP AJRD AJRD
AJRD WMGI DCOM
DCOM DCOM EVRG
ORIT EVRG MLI
GLOG MSL
EGOV CEIX
LAMR
CCR
SMMF
AVX
SONA
OZK


{Up_5pct} picks are quite different as expected. There are only 4 picks common with {Up_5pct_badCF_SIP}. Jaccard index = 4/18 = 0.22.

{Up_5pct_badCF_SIP} and SI_Rankings picks are close. There are 5 picks common. Jaccard index = 5/9 = 0.56. GTR1 data values today are:

            Ticker                WMGI       MSL     CEIX       MLI        BSBR      FBP       AJRD     DCOM     EVRG
[SI EPS Est Q1] 0.031 0.082 1.02 0.48 0.221 0.181 0.285 0.322 0.424
[SI EPS Est Q1-Last Month] 0.027 0.074 0.835 0.48 0.205 0.168 0.265 0.304 0.403
[SI EPS Est Q0-Revisions Up] 2 1 1 0 2 5 1 3 4
[SI EPS Est Q0-Revisions Down] 0 0 0 0 0 0 0 0 0
[Security Type] 12 11 11 11 31 11 11 11 11
[Actual closing Price] 30.36 11.4 33.23 31.52 13.1 10.67 39.92 19.84 58.1
volume 1,564,194 39,342 185,705 224,895 867,440 1,748,053 947,655 150,056 1,457,070
[SI Price] 30.36 11.4 33.23 31.52 13.1 10.67 39.92 19.98 58.1
[Net Cash Change/Share 12m] 3.7 8.94 9.67 -0.06 1.02 2.12 2.87 1.09 4.82
[PCFsip] 8.21 1.28 3.44 -525.33 12.84 5.03 13.91 18.33 12.05
[Total Return % over 126 days] 15.44 -18.15 -25.07 -1.46 38.86 28.21 15.81 13.85 4.74


https://boards.fool.com/si-rankings-2019-02-09-34130738.aspx...
http://gtr1.net/2013/?~Up_5pct:s20101231h21i1f0.4::styp.a:ne...
http://gtr1.net/2013/?~Up_5pct_badCF_SIP:s20101231h21i1f0.4:...

https://boards.fool.com/comparing-gtr1-picks-and-si-rankings...
Print the post Back To Top
No. of Recommendations: 2
I have to admin I haven't been keeping track of Robbie's work in correcting SI's errors. But the sentence "the SIP version of Cash Flow is used in the Rankings" suggests backtest.org should still match the rankings. Yet it doesn't. Looking at Dec 1, I seem to be including a lot of stocks that the rankings don't - I had 23 pass the cutoffs while there were 7 on the rankings list.

Looking at the first stock I passed that the rankings didn't, FOXF:
price     63.71
avd_10d 385
eps_eq1 0.465
epspm_eq1 0.44
epsum_eq0 9
epsdm_eq0 0
pcfps 19.8
prchg_26w 55.39

Those numbers all point to passing the rules of the screen. So are we using different rules, or different values?

- Jamie
Print the post Back To Top
No. of Recommendations: 1
I have monthly snapshots of some SIP data. Using that I generated {Up_5%} picks using the SIP PCF field. The 9 picks I got are also in the list on backtest.org, but backtest.org has an additional 17 picks on this date. These 17 extra picks have some blank data fields, mostly [SI PRICE/CFPS]. Does backtest.org deblank [SI PRICE/CFPS]?

Up_5% using SIP PCF
borisnand using March 4, 2016 SIP data MXL GGAL BKH LITE ALRM GTS SATS PJT MANU
backtest.org Purchased March 7, 2016 MXL NPTN POWR GGAL BKH LITE ALRM GBL TTWO AMAT GTS JKS RYAAY STO TSEM WSBF EQC UGP SATS SGC TSC BLKB PJT CXP MANU SMLP
SI Rankings 2016-05-08 RYAM POST BMA PGC INBK


picks on backtest.org but not in borisnand list, March 4, 2016 SIP data:
TICKER  PRICE  VOLUME--AVERAGE DAILY 10D  EPS EST Q1  EPS EST Q1-LAST MONTH  EPS EST Q0-REVISIONS DOWN  EPS EST Q0-REVISIONS UP  PRICE/CFPS  PRICE CHANGE 26 WEEK
AMAT 19.39 10699 0.36 0.335 0 19 25.99
BLKB 57.28 171 0.506 0.48 0 6 0.54
CXP 21.09 765 0.19 0.04 0 1 -0.47
EQC 27.15 869 2.5 7.1
GBL 37.31 52 0.88 0.62 0 1 27.55
JKS 23.35 803 1.284 1.169 0 2 18.89
NPTN 11.57 1180 0.163 0.142 0 6 61.59
POWR 18.63 1845 59.78
RYAAY 85.09 537 11.01
SGC 17.57 26 3.66
SMLP 14.94 1291 0.233 0.171 0 2 -35.1
STO 15.87 4086 0.113 0.083 0 2 8.7
TSC 13.07 36 1
TSEM 13.57 415 0.733 0.688 0 3 8.3
TTWO 36 1340 0.196 0.18 0 6 26.54
UGP 17.7 575 5.73
WSBF 13.95 50 8.14
Print the post Back To Top