|
PureBytes Links
Trading Reference Links
|
At 04:57 AM 9/24/98 -0700, you wrote:
>>Any system that uses Buy on Close is flawed anyway. The results are not
reproducible in real time. PERIOD !!!
>
>Not exactly. The problem is when the signal generating the Buy on Close
order is generated by the same close.
>
>A system that 'Buys on Close' were the signal was generated the day before
is perfectly acceptable.
>
>I make this point not to be nitpicky but rather to point out that you just
can't scan EL code and if you see 'Buy on Close' then you reject things out
of hand. The important point is were the signal was given vs were you exectute.
>
>Scott Hoffman
This is NOT TRUE as Scott says if, for example, your S&P500 stock index
trading model is used to generate signals when based on the cash index and
then when the trade is executed based on the futures. This is a valid
method of trade entry and system testing in these circumstances, because you
have 15 minutes after the Close of the cash to enter the trade, or you can
also enter in the globex. Entering at the open the next day generally
degrades performance in real time and in back testing ( at least on longs),
as compared to executing the trade 'On the Close', as noted. We buy on close
and sell on close as standard procedure and the real time results generally
equate with the system results, provided the tradeable prices are used to
calculate the gains or losses on the trades, the same as they would in the
actual trading.
Michael Paauwe
mpaauwe@xxxxxxxxxx
|