[Fwd: Std deviation Exit]

["alan myers" <a.myers@xxxxxxxxxxxx>]

This didn't get through so I'll try again.  My apologies if it appears
twice.Message-ID: <3649A6A3.BA5E038D@xxxxxxxxxxxxxxxxxx>
Date: Wed, 11 Nov 1998 09:00:51 -0600
From: A Myers <a.myers@xxxxxxxxxxxxxxxxxx>
Organization: @Home Network
X-Mailer: Mozilla 4.05 [en]C-AtHome0404  (Win95; U)
MIME-Version: 1.0
To: Michael Stewart <MPST@xxxxxxxxxxxxx>
CC: Omega <omega-list@xxxxxxxxxx>
Subject: Re: Std deviation Exit
References: <0d4f80042120bb8UPIMSSMTPUSR03@xxxxxxxxxxxxx>
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Michael and Group,

This is more of a question than an answer.  Since the average loss is
improved, the average win is what needs the work.  Wouldn't average wins
improve if you were to use a variable factor instead of a fixed factor? 
You could change the factor based on time or on price movement so that
it loosens as the trade progresses.  

What I've been trying to do is alter the factor based on the trade
statistics.  Tight stops at the start of the trade, when the risk of
being wrong is the greatest, looser in the middle, when the market has
proven you right, and tightening again when the profit (or duration of
trade) is greater than average.

I got the idea from M. Edward Borasky in a Chuck LeBeau's Traders Club
Bulletin:

"My approach is to make stops depend on "percentiles" of the recent true
range. That is, you take the past 59 days' true range values and sort
them from lowest to highest. The reason I use 59 is that the percentiles
are easy to calculate. The nth value is the n/60th percentile. For
example, the lowest true range is the 1/60th percentile and the highest
true range is the 59/60th percentile. The median is the middle value,
the 30th. The 90th percentile is the 54th value. So let's assume you
want a stop that will only be hit 10 percent of the time and be missed
90 percent of the time. That's the 90th percentile, which is the 54th
value in the list! If you're a little more gun-shy, take the 75th
percentile, which is the 45th  Value. If you're are real risk-lover, go
for the 95th percentile, the 57th value. I think you want to use the
true range rather than the high minus low because that accounts for
overnight risk."

I'd submit code, but I'm still working on it.  I'm not adept at TS
programming yet, so what takes people on this list a few minutes to do
is something that I have to puzzle over for days or weeks!

~Alan


Michael Stewart wrote:
> 
> This std deviation trailing and profit taking exits appear to decrease Avg
> trade win or loss and ROA. Exits losing trades quickly and cuts winning
> trades short, which would imply that in order for a system which uses them
> to be profitable and cover slippage and commissions a high % of winning
> trades are required.
> 
> But how does that sit alongside letting your profitable trades run and
> cutting the losers short?  Apart from widening the std band and possibly
> using discretion on the exit would anyone who uses (or has used) a std d
> exit have any thoughts to add?
> 
> { Profit Taking Stop}
> 
> Inputs: Factor(3),SDLen(30);
> VAR: VolIndex(0), LongLimit(0), LongStop(0), ShortLimit(0), ShortStop(0);
> 
> VolIndex = Factor*StdDev(Close - Close[1],SDLen);
> 
> IF MarketPosition(0)=1 THEN
>    BEGIN
>       LongLimit = EntryPrice(0) + VolIndex;
>       LongStop = EntryPrice(0) - VolIndex;
>       ExitLong ("Exit #2 Ln +") Next Bar at LongLimit Limit;
>       ExitLong ("Exit #2 Ln -") Next Bar at LongStop  Stop;
>    END;
> 
> IF MarketPosition(0)=-1 THEN
>    BEGIN
>       ShortLimit = EntryPrice(0) - VolIndex;
>       ShortStop = EntryPrice(0) + VolIndex;
>       ExitShort ("Exit #2 Sh +") Next Bar at ShortLimit Limit;
>       ExitShort ("Exit #2 Sh -") Next Bar at ShortStop  Stop;
>    END;
> 
>    { Trailing Stop}
> 
> VAR: BestLong(0), LongExit(0), BestShort(0), ShortExit(0);
> 
> IF MarketPosition(0)=1 THEN
>    BEGIN
>       BestLong = MaxPositionProfit / (CurrentContracts * BigPointValue) +
> EntryPrice(0);
> 
>       LongExit = BestLong - 3*StdDev(Close - Close[1], 75); {30}
> 
>       ExitLong ("Exit #1-Lng") Next Bar at LongExit Stop;
>    END;
> 
> IF MarketPosition(0)=-1 THEN
>    BEGIN
>       BestShort = EntryPrice(0) - MaxPositionProfit / (CurrentContracts *
> BigPointValue);
> 
>       ShortExit = BestShort + 3*StdDev(Close - Close[1], 75); {30}
> 
>       ExitShort ("Exit #1-Sht") Next Bar at ShortExit Stop;
>    END;