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

Re: DATA Discrepency BE AFRAID!!!



PureBytes Links

Trading Reference Links

Ron:

If have ever visited the futures exchanges--or better yet--pit traded for a
period of time, you'd see that the prices we get from data vendors [and the data
the exchanges send out] are only fairly crude representations of where the
market is. Remember, except for the totally electronic exchanges, we are seeing
prices that are derived from a group of clerks standing outside the pit, typing
away as fast or slow as they wish. Often, I'll do an order in a pit like the
Nasdaq futures [where I have direct phones to the pit] and after I make the
trade and confirm the trade and price is moving away from where I made the
trade, my broker will be shouting at a pit 'recorder' to get the price on the
*blanking* screen.

I had a seat on the CME for a stretch of time. Trading in the currency and S&P
pits for any period of time there will give you a better feel for the 'noise'
contained in the data most of us use. You clean it the best you can and trust
techniques you use that are successful over time. Data streams from different
sources may not match--especially if they are using different data plants.

What do you do in this case? You'll have to watch both data streams with your
indicator and then choose which you trust and which is successful. This may be
the lesson that shows you just how subjective most indicators are.

Be careful out there! This is real money!

Best,

Tim Morge

Ron wrote:

> Please consider the following example:
>
> DTN Feed_IBM 60min_Stochastic Crossover
> A crossover buy is generated using DTN data.
>
> Quote.com Feed_IBM 60min_Stochastic No Crossover
> No crossover in Quote.com data using same stochastic parameters.
>
> What do you do? Both feeds have been working fine all day and tick counts
> are within 95% of each other but obviously there is a discrepancy in the
> data between the two feds. Its very hard to get to independent feed to match
> data exactly, especially on the lower time frames.
>
> Since no two feed are the same does it make sence to test systems on one
> data source and trade it using another. No matter how reliable your data
> source is for historical data there is no guarantee that your real time data
> matches the intrinsic characteristics of your historical data source. Infact
> most often the performances will NOT be the same for the same system. This
> highly limits the type of systems that can be traded in real-time and tested
> historically. SO BE AFRAID!!!
>
> This data problem has been bugging me for a long time and unless you have
> direct feeds from exchanges and full control over the data distribution and
> filtering process you may be trading based on artificially adjusted data.
>
> So where do you draw the line.
>
> All comments appreciated. Thanks