|
PureBytes Links
Trading Reference Links
|
Hello Yarroll;
Pl. go through the book "Trade your Way to Financial Freedom" by Dr. Van K.
Tharp. The book will solve your questions. The AWL plays a very important
role in system testing. But I fear the way which you are testing your
systems can show you any thing. Go through that book before testing the
sytems.
Bhanja
----- Original Message -----
From: Yarroll <komin@xxxxxxxxxxxxxx>
To: <metastock@xxxxxxxxxxxxxxxxxx>
Sent: Friday, December 07, 2001 10:42 PM
Subject: What gives? System performance summary
> Hello List,
>
> I was just wondering. Does anyone use the Average Win/Average Loss ratio
> from the system's Summary Report? If so, then how it can be used to any
> useful purpose?
>
> This might be a weird question to ask, I know. But I was just testing for
> various trading setups. Lets see, Ive got 2 setups that show me the
> following:
>
> Net Profit Total trades Winning Trades Losing trades Avg Win/Loss
> -91 2 1 1
> 0.91
> 587 7 5 2
> 0.55
>
>
> I know this is hardly usable since the 1st setup has only 2 trades. But
just
> as an example...
> Setup # 1 has a much better Avg Win/Loss than #2 (0.91 vs 0.55) and yet it
> can well be in the negative...
> Setup # 2 has less than breathtaking Avg Win/Loss but, hey, isn't 587
nicer
> than -91?
> Does Avg Win/Loss tell you anything. Or, should it be used only in
> conjunction with something else (for example, Net Profit in the positive
AND
> a nice Avg Win/Loss, etc.)
>
>
> I also have a 2nd question. This one can be a bit tricky...
> Lets suppose I'm testing yet another trading setup. It's more or less
ready,
> has all those results nicely lined up. Next step, I want to add a FILTER
to
> my setup. Lets say this filter is that the price action takes place above
> Mov(C,100,S), whatever. Lets say I somehow suspect that if I use this
> filter, my setup could be overall less profitable, (some entries after all
> have been filtered out), yet the remaining entries would bring more profit
> per good entry than loss per bad entry.
>
> Is there any way at all I could figure it out ?
> a) Avg Win/Loss? That would be the natural way to go... But those figures
> bring in those weird results as shown above.
> b) a simple Hit Rate? (winners divided by losers) But this is also hardly
> usable... for obvious reasons - you could end up having a lot of small
> winners, and a few large losers.
>
> Please advise.
>
> Thanks,
> Yarroll
>
|