|
PureBytes Links
Trading Reference Links
|
Joe Slater wrote:
>
> I have noticed that for the daily data I get for things like CBOT bonds or
> the DJ, the open is the open of the afternoon session, and the close is that
> of the previous day session. This means that gaps are small.
>
> However, large gaps can show up between the "suspend" of the evening and the
> "resume" of the day session. Since these show up in the middle of the daily
> bars, doesn't this have the effect of drastically increasing possible
> slippage making it quite difficult to test systems based on daily data?
>
> Does this mean you really need tick-by-tick data to test daily systems?
>
> Joe Slater
Some data vendors provide "day session only" data as an option. The one
I currrently use does not but you might check CSI or others. If you
have Midam data, these might serve as a pretty good approximation for
day session-only data since they don't trade at night and match the big
contracts' prices quite closely.
I think the question about whether you need tick by tick (or say 5 min.)
data to test an end of day system is a different question. If you have
a day-of-entry stoploss and it is sometimes within the likely range for
the day you get in, then you probably need intraday data and also to
make sure you take particular steps that your day-of-entry stops are
actually executed by TS (see a thread on this list on that topic - I
think you put your day-of-entry stop in a separate system and use
"include system" command).
I am still in the beginning stages of learning about systems, so I
haven't worried about the day/Globex session problem but I'll have to
realize my results are very approximate until I do.
|