[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

A confidence shaking observation concerning TS/eSignal data.



PureBytes Links

Trading Reference Links

List,

Today I made a chance observation with respect to a data 'discrepancy' in my 
system.  What I 'saw' was a difference between data for which I had  hard 
copy values recorded close to the time at which they had been received in 
real time from eSignal and what I now see stored in my data base for these 
values.  I would not have noticed this if I had NOT had the hard copy numbers 
to look at.  To be specific, the first observation concerned data received 
via eSignal into TradeStation 2000 on 8/3/2001 (last Friday) for $ADV, the 
number of advancing issues on the NYSE.  The data are for a 60 minute chart 
using natural hour bars.  Here is the 'comparison'':

            'Close' DATA for 8/3/2001 for $ADV
              Hard Copy      Currently in data base
0700           1091         885
0800        1149            904
0900        1232            1064
1000        1242            1201
1100        1265            1214
1200        1228            1250
1300        1206            1397
1315        1197            1426

Once I saw this I went back to other data for this ($ADV) 'price' series.  I 
did NOT do a systematic check so I do not yet know the extent to which this 
shows up in my data base.  BUT I see enough incidences where this is the case 
to  have my confidence in the data I use with TradeStation really shaken.  I 
HAVE seen enough to know that I am not 'seeing things' and that I am not 
having a 'psychotic episode'.  Thinking about this, I have concluded that the 
only way I would have observed this is to have hard copy data recorded close 
in time to when the data were first received AND to have the data currently 
still residing in my data base.  In my case, this is not true except for a 
few isolated cases.  I DO have some hard copy S&P futures H/L/C data 
coincident with the $ADV data.  These S&P data in the data base DO agree with 
the hard copy data for the dates I have looked at .. BUT I have not done a 
systematic look at this side of the issue yet .... nor have I gone through 
all the $ADV data for which I do have both hard copy data and currently 
stored data to compare.  I AM very interested in trying to 'understand' just 
what the hell is going on though.  I need my confidence back!!!

I have scratched my head over just how such a situation might arise.  One 
thought was that modifying the original data through the importing of 
'refresh data' from either the TradingTechnologies site, HistoryBank or Mark 
Brown's site might possibly contribute.  In my case I do NOT generally 
'refresh' intraday data from any of these 3 sources. I DO routinely download 
certain symbols, indices and futures/stock data, from one or the other of 
these sources for symbols which I do not track on a daily (real time) basis.  
I DO sometimes 'refresh' intraday data when I have had an 'outage'.  That has 
been, fortunately, a rare occurence in my case.  In the case of Friday's 
data, I did not do ANY download ..... because I could not connect with the 
HistoryBank server and did not go to the other sites.  So I am pretty 
confident that I am seeing a real difference between data recorded in real 
time and those which I now see in a chart and data window.  At this stage I 
certainly cannot rule out (or in) the possibility that these observations are 
unique to me and my 'system'.  For the sake of you others out there, let it 
be just me <g>.

Now what to do?  Any suggestions from you good people on this list would be 
much appreciated.  One thing that I have thought might be a useful next step 
would be to compare data from one or more of you folks with the data in my 
data base and those for which I have hard copy.  How to do that?  I think the 
best way to do that would be to look at data from (a) 'volunteer(s)' in the 
form of your data 'exported' to an Excel file with H/L/C S&P futures data and 
$ADV data from a 60 min. natural hour bar chart.  I could then do the same on 
my side and then combine the data and very efficiently look for any 
differences between the data you send me, data from my database and hard copy 
data which I have. To keep it within bounds at this time, data for the Sept 
S&P futures and $ADV for July and August to date would to me to be sufficient 
as a starting point.  Again, to be specific I am using EXCEL 2000 (9.0.2720) 
to try to front run any potential EXCEL spreadsheet compatibility problems. I 
am using eSignal for data with TradeStation 2000 (sp5).  To keep it apples to 
apples, it would be best I think to initially at least look at data 
accumulated on as close to identical data accumulatioin set ups as possible. 

ANY thoughts anybody has on this issue would be greatly appreciated ... and I 
hope none of you have this problem.  Like I said, it sure shook up my 
conficence in my data accumulated with my current set up.

Regards,

Lee Scharpen