|
PureBytes Links
Trading Reference Links
|
Be careful with this idea, as it is a very slippery slope. Capturing the
magnitude and accuracy of the spread depends a lot on what time-frame you're
referencing in your data series. As an example, let's use EOD daily data.
The only thing you can be sure of is at what time and price the market
opened, and at what time and price it closed. To calculate an accurate
spread, you must be sampling the two (or more) data series at the SAME time.
If one market closes at 2:00pm and the second market closes at 3:00pm, then
"close of
data1 - close of data2" will not give you an accurate spread, as they are
not being sampled at the same time -- you would need to compare the close of
data1 (2:00pm) with the price data2 was trading at 2:00pm. If the markets
open and close at the same time, you can get an accurate picture of what the
spread was at the open, and what the spread was at the close -- the
difference of the opening and closing spread will be the KNOWN range of the
spread for that day, given the time series (daily data) you're working with.
Since with daily EOD bars, you don't know what "time" the highs and lows
were made (you don't know if both data series lows/highs were made at the
same time) you could be creating a false high and low for the spread using
the "high/low of data1 - high/low of data2". BTW, the above is also true for
intra-day bars ( 5min, 15 min, hourly, etc.), if you don't have the tick
history to determine when the high and low of the bar was made.
Now...Since I don't have a real-time data feed, I'm drawing a logical
assumption on the following... if you are comparing intra-day/tick data,
and comparing the shortest time period possible, then "close of data1 -
close of data2" should capture the highs and lows of the spread fairly
accurately, as you are calculating the spread "real-time", and the daily
high and low of the spread will be captured from the real-time data
comparisons/calculations.
----- Original Message -----
From: "Tanoto Sau Ian" <sitanoto@xxxxxxxxxxxxxx>
To: "List, Omega" <omega-list@xxxxxxxxxx>
Sent: Saturday, February 02, 2002 4:21 PM
Subject: Regarding Spreads
> HI,
>
> The current spread function in ST is the difference between "close of
> data1 - close of data2". This will return only 1 value and naturally only
a
> line chart is plotted out from this information.
>
> I am wondering if it is possible to kinda create a bar chart for spreads,
> with "high of data1 - high of data2" equating to the high of the spread
bar,
> "low of data1 - low of data2" being the low of the spread bar, "open of
> data1 - open of data2" the open of the spread bar and lastly "close of
> data1 - close of data2" will be the close of the day for the spread bar.
>
> As spreads do fluctuate through the course of a day trading sesssion, it
> would be most helpful if we could capture the actual magnitude of the
> spreads' movement. In short, I think it will be a better way to represent
> them than just taking the difference in the closing. For example, a spread
> that closes at around 50 for the last few days may probably be having
> fluctuations to as high as 60 and as low as 40. Thus, there is actually a
> lot more action than a relatively straight line of around a value of 50
> would show.
>
> tks
>
|