|
PureBytes Links
Trading Reference Links
|
PO ecrit:
> I cant say for sure but the problem is probably due to the databaseS:
>
> TS4 has three different databases:
>
> The signal.dat ( tick file) and signal.idx ( index file)
> The signal5.dat ( tick file) and signal5.idx ( index file), optional and
> built (updated) during the morning update The daily files ( on file per
> symbol).
>
> Daily data are updated during the day from the tick file
>
> When an error exist in the tick or 5 min datafile it may affect the daily
> bars after the morning update This should explain why the exit signal
> disappeared during the night (a bad tick affecte the 5 min datafile eve
> before the morning update)
>
> More the exit that disappeared was on the low of the daily bar.
NO - it was NEAR, but not the low ! the EXIT was 126770 (as said) and the
low 126650 (as said) - there was NO EFFECT by this !
DH wrote:
> good stuff from pierre on rebuilding the databases
>
> > I cant say for sure but the problem is probably due to the databaseS:
>
> i think, in THIS case, the problem was much simpler. the day in question
> opened in the lower half of the range, then went up, then went below the
> open and closed near the open. in realtime, the system correctly bought on
> a stop near the high and exited on a money management stop when the price
> went down.
>
> but at the end of the day ts downloaded the daily bar and recalculated the
> system based on that. since the day opened nearer the low, ts ASSUMES that
> the price went down before it went up (wrong in this case). so, the
> stoploss was never triggered on the updated daily data. only by setting
> the bouncing ticks to a high value can you trigger the stop.
>
> of course, doing that will give you other historical trades that are
> wrong the other way.... showing an exit when there was none.
HEY GREAT :-))....so its not the "SYSTEM REPORT" (® OR) any more but
the "SYSTEM GUESSING REPORT" (® ZeBrA)
> if TS2K
> really does use the tick database for deciding the order of trades on
> daily data, i see that as a major improvement.
isnt that REAL easy ?
PO ecrit:
> Probably true if the entry was on the same bar than the exit, but I do not
> remember if it was the case here.
come on Pierre, you have all details - somehow noone seems to be able to
read mails when a REAL problem is there ! Sure it was on the same bar !
> TS4 makes no difference with daily bars when they are generated by tick
> files or true daily ( OHLC) database and when a recalculation occurs. The
> only solution to have historical result in this case meeting realtime is
> to use intraday bars and evaluate the daily bars from the intraday bars.
> Cumbersome, but solved by TS2000i
>
> Sincerely,
>
> PO
thanks
rgds hans
|