|
PureBytes Links
Trading Reference Links
|
Hello david,
Just a bit of follow up--
as a quick check I looked at tick entries in Future Source and
Gloal Server (Q.Feed) for the past two days. QFeed is showing aaprox
35000 tick entries for the emini sp while FS shows approx 13000.
I talked to FS and they didn't understand why this would happen. They
did state ticks were sometimes bunched in busy periods and then sent
thru but there is no evidence in my entire mini tick file of out of
sequence timestamps. Likewise, there are many consiectuvie ticks at
the same price so if they are blocking unchanged ticks it seems to not
be all of them.
All that doesn't really get to my problem in that the indicator only
uses up and down ticks, not total tick volume or unchanged ticks.
--
Best regards,
Jim Johnson mailto:jejohn@xxxxxxxxxxxxxxxx
Thursday, April 04, 2002, 2:57:33 PM, you wrote:
dbs> The fact that you get 2 different results already eludes to the fact that
dbs> you are
dbs> getting different tick counts. The upticks and downticks functions are
dbs> already
dbs> counting ticks for you.
dbs> If I remember correctly, FS only transmits quotes when there is a change
dbs> in price.
dbs> Probably to save bandwidth because they are a satellite feed or possibly
dbs> because that's
dbs> the way they get it from Bridge. This may not be true of quotecom.
dbs> FS uses a rather unique timestamp that can wreak havoc on tick functions.
dbs> Where
dbs> the Signal server ends a bar, say on a 5min chart where the time is 1045,
dbs> with the
dbs> last tick with a 1045 timestamp, FS begins the 1050 bar with data having
dbs> the 1045
dbs> timestamp. It's been a few years since I messed with this but I think the
dbs> above is
dbs> correct. I was working with code that wrote to a realtime ascii price file
dbs> and the various
dbs> systems read from that file to make calculations. My intentions were to
dbs> force TS to
dbs> provide the same results in realtime as historical. TS had a difficult
dbs> time extracting
dbs> the ascii price that corresponded to the realtime bar because of the way
dbs> TS stores.
dbs> Also, with high volume markets, such as the minis, data gets jammed
dbs> in the FS pipeline. Somehow, duplicate packets of the jammed data are
dbs> transmitted to
dbs> you after other data gets thru and timestamps are out of order and false
dbs> duplicate
dbs> ticks end up in your data. The jams can occasionally cause out of sequence
dbs> errors in
dbs> other contracts that aren't high volume.
dbs> I would trust the quotecom results in your indicator over the FS results.
dbs> Good luck.
dbs> Jim Johnson wrote:
>> Hello omega-list,
>>
>> I'm working with Bill Blau's Tick volume Indicator (TVI) from his
>> book Momentum, Direction and Divergence, chapter 4. And I'm getting
>> very differenct looking plots under TS4 w FutureSource and TS2k w
>> q.feed.
>>
>> Perhaps this is because the two feeds are providing different tick
>> counts. How would I check that? Any other ideas on why this would
>> occur? Thanks.
>>
>> I'd be happy to provide the code to anyone who wants it. The basic
>> formula is
>>
>> TVI = (triple.ema(Upticks) - triple.ema(Downticks) )/
>> (triple.ema(Upticks) + triple.ema(Downticks))
>>
>> --
>> Best regards,
>> Jim Johnson mailto:jejohn@xxxxxxxxxxxxxxxx
|