|
PureBytes Links
Trading Reference Links
|
CwWaring@xxxxxxx wrote:
> How many bars would you consider to be a satifactory test of a 5min system?
Not that easy to say - I trade 5 min bars myself, but compile over hundreds of
issues. With a single issue, this is going to be a problem, since the further you
go back the more you risk alienating yourself from current conditions. I'm sure
the spoo traders can answer this better than I, but I'd say that in this case, a
year would be reasonable. Just make sure to keep this in perspective, though.
This sort of thing is probably more helpful in comparing different systems and
approaches - which can at least help you decide between strategies. Make sure
that you also factor in how each particular system suits your trading style and
comfort level - in other words, only compare systems which you are comfortable
enough to trade properly when real money is on the table.
> How do I change increase the length of the data to be analyzed? I have
> incresed the number of days to 100&500 with no change. maxbars back are 50 in
> almost all cases. The lookback is not more than 25.
Max bars back only have to do with indicators and systems referencing data within
their formulas - not with the period of the test. First, make sure your portfolio
is set up to receive the amount you want, then make sure you actually have the
data for the period, and finally make sure the amount of days back is set up on
your chart. If it ain't on the chart, it won't be in the results.
> I have also seen some comments from some suggesting that their systems don't
> need more than a few weeks test as they seem to adjust/reoptimize frequently.
> How can that be a successful strategy?
I suppose you could just test for a few weeks, and the info wouldn't be entirely
useless, but one would be foolish to rely on such an insignificant sample,
especially when longer ones can be used without much extra trouble.
Regards,
A.J.
|