Compare the Reuters data to the Compustat data with the same stocks in a Custom List

Research
1

I wanted to test a Sim to compare the Reuters data to the Compustat data with the same stocks in a Custom List.
First I ran the Compustat Screener with this buy rule; AvgDailyTot(20) > 100000 & AvgDailyTot(20) < 1000000, and I got 976 stocks (for today).
Next I downloaded them to Excel and copied them into a Custom List to run on the Reuters data.
From that list there were 27 stocks that still had the 3 prefix on their tickers, so I deleted them from the list.
When I saved the list, there were 21 stocks that were invalid, so I ended up with 932 stocks that are valid in both the Compustat and Reuters data.
Here is the list that was invalid: AKO.B AUNFF CSCTF CWSRF FES GSV JE LSG NNA PACEF PFLT PRIS.B PRQNF PWCDF QADA QRM SENEA SPLP STB TGD XTLS
I then copied the final list to a Custom List for the Compustat test server.
So now I have a list that I can compare one to one (at least for today) with both Reuters and Compustat.

Next I created 2 very simple 20 stock Sims using the buy rule; InList(“CompustatADT100k_1M”), and one sell rule; NoBars > 21.

Here is the Compustat Sim: http://38.102.52.102/port_summary.jsp?portid=911068 ; annual return = 26.26%, max drawdown = 45.23%, Winners = 68.55%, Total Market Value = $1,377,016

Here is the Reuters Sim: http://www.portfolio123.com/port_summary.jsp?portid=911196 ; annual return = 20.84%, max drawdown = 59.94%, Winners = 71.3%, Total Market Value = $842,004

At least in this test the Compustat data set wins, but there still is the question as to how can there be so much difference using the same stocks and rules?

Below is the list of stocks that were used in both Custom Lists:

AAMRQ AAU AAV ABCB ABCD ABTL ACAD ACLS ACPW ACUR ADGE ADK ADY AERL AEZS AGX AGYS AHS AIQ ALCO ALIM ALNC AMBO AMBT AMCN AMID AMRI AMSWA AMWD ANAC ANAD ANAT ANCX ANGO ANX AOSL AP APAGF APFC APP APPY APRI AQ ARC ARDNA AROW ART ARTNA ARWR ARX ASI ASMI ASTI ASTM ASUV ASYS ATEC ATHX ATLO ATRC ATRI ATSG ATX AUDC AV AVHI AVL AVNW AWRE AXTI AZC BAGL BALT BANC BANF BBGI BBNK BBX BCA BCBP BCDS BDGE BDRBF BEAT BELFB BERK BFIN BGFV BGMD BIOD BIOF BIOL BKEP BKMU BKR BKYF BLDP BLIN BLMT BMRC BMTC BNCL BONA BONE BORN BRD BREW BRP BSPM BSRR BSTC BTFG BTN BTX BUSE BWEN BWINB BXC CAAS CAC CACB CAE CAK CARB CASS CAST CBK CBNJ CBRX CBZ CCBG CCF CCH CCIX CCM CCNE CCRN CCRT CCSC CCUR CCXI CDCAQ CDI CDTI CDXS CDZI CEA CEBK CECE CEDU CEMP CERE CFI CFK CFNB CGA CGEN CGR CHDX CHFN CHKE CHLN CHOP CHRM CIA CIS CISG CKSW CLCT CLDT CLIR CLMS CLNT CMRG CMT CNBC CNBKA CNEP CNTF CO COBZ COHU COKE COOL CPAC CPIX CPLP CRD.B CRFN CRME CRNT CRRC CRT CRTX CRWN CRY CSBC CSBK CSFL CSII CSRE CSS CSTE CSUN CSVI CSWC CT CTCH CTEL CTFO CTGX CTHR CTO CTS CUI CUTR CVM CVTI CVV CWCO CWST CWTR CXDC CXPO CYAN CYD CYTK CYTX CZNC DATE DAVE DBLE DEER DEJ DEXO DFR DFZ DGICA DGII DHIL DHT DIIBF DIMI DJCO DLA DM DNN DOM DQ DRCO DRD DRL DROP DRRX DRWI DSCI DSGX DSPG DSS DUSA DVOX DWCH DYN EAC EBF EBSB EBTC EBTX ECTE EDAC EDGR EDS EFSC EGAS EGBN EGT EKDKQ ELON ELOS ELRC EMAN EMCI EMMS ENV ENVI ENZ EPM EQU ERII ESBF ESEA ESLT ESP ESSA ESSX ETAK ETM ETRM EVAC EVC EVOL EXAC EXFO EXM FALC FARM FBN FBNC FBNK FBRC FC FCBC FDEF FENG FF FFBH FFCH FFN FFNW FHCO FIBK FISI FIX FIZZ FLIC FLML FLOW FLWS FLY FMD FNLC FOLD FONR FRF FRM FRNK FRP FRS FSCI FSM FSRV FSTR FUBC FULL FVE FX FXCB GABC GAIN GASS GBL GCAP GENT GGAL GHM GIVN GKNT GLAD GLCH GLP GMK GMXR GNE GNMK GOOD GRIF GRNB GROW GSAT GSBC GSE GSH GSIT GSJK GSOL GST GSX GTN GTXI GUID GURE GV GZT HALL HBIO HBM HBNC HCKT HDSN HFWA HIFS HIL HILL HIMX HLYS HMPR HMST HNH HNRG HOGS HOKU HOME HSFT HSKA HSOL HSON HTBK HTCH HTCO HURC HUSA HVB HVT HWCC IBA IDSY IEP IFT IHC IIIN IMI IMKTA IMMR IMN IMSC INFU INSM INVE IPCI IRIS IRS ISH ISLE ISR ISRL ISS ITIC ITPOF ITRN ITWG IVAC IVAN IX IXYS JAX JCS JHX JKS JMBA JMP JOEZ JOUT JRN JST KBALB KCLI KEQU KEYW KGJI KH KMG KNDI KONA KOPN KRG KRNY KSWS KTEC KUB KVHI KYO LACO LAKE LAS LAWS LBAI LBY LCAV LCI LCUT LDL LEDS LEE LEI LINC LION LIOX LIWA LLEN LMIA LMLP LMNR LMOS LNDC LOCM LODE LONG LOV LPH LTS LUB LVB LWAY LYTS MAG MASC MATR MAXY MBLX MBRG MBVT MCEP MCHX MCRI MCS MDCI MDTH MDW MEA MEDW MEET MEG MEI MELA MEMP MEMS MERC MERU METR MFI MGIC MGPI MGT MIICF MIL MITL MLR MN MNI MNTX MOBI MOSY MOTR MPAA MPR MPX MRLN MRVC MSEX MSFG MSL MSO MSPD MSW MTR MTRX MVG MXL MY NABI NAK NASB NATH NATL NATR NAUH NBS NBY NCIT NCQ NCS NCTY NEWS NFBK NGL NGPC NHC NICK NKBP NKSH NL NLS NLST NM NMFC NMR NMRX NOA NOAH NOVB NPD NPTN NRCI NRT NTE NTWK NUTR NVAX NVDQ NVEC NVMI NWLI NWY NXST OCC OCFC ODC OESX OGXI OIIM OILT OKSB OLP OMAB OME ONCY ONFC ONP OPAY OPXT OPY ORBC ORCC ORN ORRF OSH OSTK OTIV OTT OUTD OVRL OWW OXF PACR PAM PAMT PATH PATR PBNY PBTH PCBC PCBK PCCC PCO PCOM PCYG PDFS PDII PDO PEIX PENX PEOP PERF PERI PFSW PGTI PHX PIKE PIP PKOH PLFE PLG PLM PLPC PLUS PMBC PMTI PNSN PNX POPE PPHM PPP PRGX PRMW PROJ PRSG PRTS PRXI PSDV PSEM PSTB PSTI PTI PTIE PTNR PTX PULS PVD PVSW PWAV PWOD PZE PZN PZZI QTWW RARS RBCAA RCKY RDA RDNT REDF REED REGI REIS RELL REXI RICK RLH RMCF RMTR RMYI RNO RNST RNWK ROMA ROYL RRMS RSYS RTIX RUK RVM SAAS SALM SANW SAPX SASR SATC SB SBCF SBSA SBX SCHS SCMF SCMP SCR SDIX SEH SFE SGA SGK SGU SHI SHLO SHOR SHP SIF SIFY SIGA SILC SIM SJW SKY SLI SLTM SMI SMRT SMS SMSI SMT SMTX SNBC SNMX SNSS SOL SOQ SORL SPA SPAR SPNS SPRT SQNS SRCE SRSL SSN SSRX STEL STEM STFC STKL STL STN STNG STRC STRL STRM STV STXS SUBK SUMR SUNS SUPX SVA SWS SXCL SYBT SYMM SYMX SYN SYNM SYUT SYX TA TACT TAOIF TAOM TARO TAS TAT TBBK TBNK TCBK TCL TCPC TCRD TCX TEAR TELK TEU TGX THER THFF THM THQI THTI TIII TINY TIS TISA TMS TNAV TNP TOWR TPL TREE TRIB TRIT TRK TRNO TROV TRR TSEM TST TTGT TUC TUES TVL TXCC UACL UBSH UCFC UCTT UDRL UEPS UFI UG ULTR UMH UNTK UNXL UQM URG URRE URZ USAK USAT USEG USHS USLM UTSI UVE UVSP VASC VCBI VCO VHI VIAS VICR VIPS VISN VIST VLNC VNDA VOG VPG VRML VRNM VRS VSEC VSTM VTNC VTSS WACLY WASH WAVX WBCO WD WEST WEYS WF WFD WHG WINA WLB WLFC WMAR WNS WRES WSBF WSCI WSR WSTG WSTL WTBA WWAY XIN XRA XRM XRS XUE YONG YORW ZA ZBB ZGNX ZIGO ZIXI ZN ZNH ZOOM ZSTN ZX ZZ

Denny :sunglasses:

2

I find the same as Denny . I use custom lists quite a bit and find the results vary dramatically using the Compustat vs Reuters data . Can this all be the result of point -in time data ?
It raises an alarm bell as to how robust our solutions are and whether our models in 123 have any predictive value as to expected future outcomes ?

I would not have expected such significant differences .

Any thoughts

Hugh

3

Hugh,

Even though I question what the causes are of the differences between my 2 Sims, I have no doubt about the predictive value of the P123 Sims. Having been a member since 2004 I have experienced GREAT returns due to the predictive value if the Sims.

Denny :sunglasses:

4

Hi All

I played around with denny’s list — called it denny .
Then ran a simple 1 year test on 10 stocks called denny1 - made it public . Ranking system 1 week return — should be the same in reuters and compustat . Ranking system also public . I get a 40 % + differential in return over 1 year using the same list and a ranking system that should deliver the same results . I called the simulation on the beta server denny2

compustat : http://38.102.52.102/port_summary.jsp?portid=911278 128% p.a

reuters : http://www.portfolio123.com/port_summary.jsp?portid=911346 77 % pa

ranking : http://www.portfolio123.com/app/ranking-system/97541

I then created two new portfolios as of today and got different portfolios .

reuters : http://www.portfolio123.com/port_summary.jsp?portid=911348

compustat : http://38.102.52.102/port_summary.jsp?portid=911282

Do we have differences in closing prices between the two data banks ? If not the results should be exactly the same or close .

Any thoughts ? Something does not look right here .

5

Hugh,

One difference between your 2 Sims is the Reuters Sim uses the R2000 Benchmark while the Compustat Sim uses the S&P 500. Since you have market timing that includes the Benchmark performance there will be a difference expected.

Your Ranking System in your Compustat Sim is set to private.

Denny :sunglasses:

6

Denny

I made the ranking public now thks

I also removed all timing rules and changed the benchmark to be the same .

Results still differ and remained the same . The timing was switched off .

I am very curious as to the real reason behind the difference .

Hugh

7

Hugh,

In comparing the 2 Sims, after changing the Benchmark to R2000, I found that of the first 10 purchases the Compustat Sim didn’t buy BXC as did Reuters and instead bought ITWG. ITWG was the second highest ranked stock in Compustat but was not in the top 10 in Reuters; BXC was their 10th ranked stock.

Since your ranking system and Sim rules use only price data (no fundamentals), I can only conclude that there is a price error in one of the data sets. Based on what P123 is telling us about the quality of the 2 data sets, I guess I have to believe that the error is in the Reuters data until we can determine which is correct.

Denny:sunglasses:

8

Hi Denny

THx for the effort .

I more or less came to the same conclusion but hope there is not something more to it . It is interesting that all my results using custom lists have improved using the compustat data – I hope this is a true reflection of what we can expect from the " better " data

H

9

All,

I may have discovered a source of some of the problem. I went to the P123 Stock price table for ITWG for both Compustat’s and Reuters’ data. That stock turns out to be a thinly traded stock back in 7/5/11 when it was bought by the Compustat Sim. Around that time there were many days that had no volume, and had zeros in the table for the open price, and volume. The high, low, and close were copied from the previous day’s data. In the table for the Reuters data there were just missing dates for days with no data when there was no trading.

Because of the problems that no data caused in Sims that traded thinly traded stocks, about a year or so ago in the Forum P123 stated that for those stocks, the price data for the missing days would be filled with the data from the last day the stock was traded. The volume data would remain zero.

It looks to me like Compustat used a similar approach with their data, but at least for this stock, the days in which there were zero volume, there were also zeros in the table for the open price. Therefore, if a Sim was setup to buy and sell on the next open then it might find a zero for the price if it happened to rebalance on a no volume day.

Marco,

Please look into this. Although I only looked at 1 stock, you may need to correct the zero open prices to the last traded day’s open for all thinly traded stocks.

Denny :sunglasses:

10

Denny / Marco

Denny Thank you for your effort .

Marco please look into this for us

Hugh

11

Please see http://www.portfolio123.com/mvnforum/viewthread?thread=5377

Thanks

12

All,

Marco is correct. I checked a dozen S&P 500 stocks and even IBM has zeros for open prices prior to 8/26/2004. P123’s workaround of filling in the zeros with the average of Hi & Low allows Sims that use Next Open to be run without “no data” prior to 8/26/2004. This does explain the differences in the above Sims since they are “Price Only” comparisons and do not have fundamental factors in either the ranking system or the buy / sell rules.

However, I (and we all) wish that they would have told us this a week ago so we would not have had the anxiety we have felt with the big differences in out tests. From what we have found to date (and now P123 has told us), I think that we can make reasonable comparisons only for large caps that have no zero volume days, and only for dates after 8/26/2004. After comparing my larger cap Sims for dates after 8/26/2004, I feel that my current real Ports will behave pretty much the same after the switch to Compustat as they did before.

But for me, I will wait until P123 has corrected the open price data before I try to develop any new Sims or ranking systems (the open price data effects ranking systems that use TA factors also).

Denny :sunglasses: