[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: atr



PureBytes Links

Trading Reference Links

Dr. Bhanja,

I don't think there is any one "best" method.  The most straight-forward
way is to use the MetaStock built-in indicators.  For custom programming
you should specify what you are using and be consistent.

Wilder didn't use EMAs because that was not practical for manual
calculations in pre-computer times.  I've given Wilder's smoothing
algorithm.  There must be a mathematical connection between that and
EMAs because it is a fact that the Wilder algorithm can be replaced by
an EMA of twice Wilder's period minus one (e.g. Wilder's 10-period
smoothing corresponds to a 19 period EMA).

Please note that you have replied to a message containing an error.  I
followed it up with a correction.  The last line of the ATR (EMA) should
read:
Mov(TR, Pds, E).

HHP
===================

"Dr. R.BHANJA" wrote:
> 
> In most of the averaging of indicators ( mainly developed by Wielder like
> DI+, DI-, ADX etc.), Equis uses Wilder's averaging method. Yes, the values
> from simple moving average method and Wielder's method differs a lot, mainly
> if the period of averaging is less. So what should be the best method to
> approach.
> Furher what about the exponential moving average. Wielder has never used
> exponential moving average in his book.
> BHANJA
> ----- Original Message -----
> From: HHP <hhp@xxxxxxxx>
> To: <metastock@xxxxxxxxxxxxx>
> Sent: Sunday, July 16, 2000 12:04 AM
> Subject: Re: atr
> 
> > Here's the complete formula for ATR.  Plot it and the MStk ATR (for the
> > same number of bars) in separate inner windows, then drag one onto the
> > other.  They merge exactly except at the beginning of data due to
> > different initialising.
> >
> > =====================================
> > {ATR (Wilder)}
> >
> > Pds:= Input("ATR Periods?", 1,100,10);
> >
> > TR:= Max(H, Ref(C, -1)) - Min(L, Ref(C, -1));
> >
> > (Cum(1) <= Pds) * (Sum(TR, Pds) / Pds) +
> > (Cum(1) > Pds) * ((PREV * (Pds - 1) + TR) / Pds;
> > =====================================
> >
> > You can also use an EMA for double the number of periods minus one,
> > which will be almost an exact overlay (e.g. ATR (MStk) 10 periods, ATR
> > (EMA) 19 periods).
> >
> > =====================================
> > {ATR (EMA)}
> >
> > Pds:= Input("ATR Periods?", 1,100,19);
> > TR:= Max(H, Ref(C, -1)) - Min(L, Ref(C, -1));
> > Mov(TR,20,E);
> > =====================================
> >
> > HHP
> > ===========================
> > HHP wrote:
> > >
> > > Adam,
> > >
> > > Welles Wilder wrote the book on ATR as he did on almost all the other
> > > tried and true indicators, but that was in pre-computer days, so he used
> > > a smoothing method that was (relatively) simple to calculate by hand.
> > > See 'New Concepte in Technical Trading Systems' by J. Welles Wilder,
> > > 1978, Section 3 'Volatility', p.23.
> > >
> > > For an ATR(7) he would start by summing the True Range for the first 7
> > > days and divide by seven.  For day 8 and all subsequent days he would
> > > take the previous day's value times six, add the current day's value,
> > > and divide by seven.  (Prev * 6 + ATR(1)) / 7.
> > >
> > > I see no advantage in using ATR Classic.  As long as we are consistent,
> > > and define what we are doing, we can use whatever we wish.  I'm sure
> > > Welles Wilder would use a different method today.
> > >
> > > HHP
> > > ===================
> > >
> > > Adam Hefner wrote:
> > > >
> > > > Ian,
> > > >     You are very correct!! After reading your mail, I decided to find
> > > > out what is going on.... below is the code I used:
> > > >
> > > > r1:=H-L;
> > > > r2:=H-Ref(C,-1);
> > > > r3:=Ref(C,-1)-L;
> > > > At:=If((r1>r2) AND (r1>r3),
> > > >     {then}r1,
> > > >     {else}If(r2>r3,
> > > >            {then}r2,
> > > >            {else}r3));
> > > > At;
> > > >
> > > > The value  of "At" comes up with the exact value of ATR(1), indicating
> that
> > > > the "True Range" calculation that Equis uses is correct... so I have
> no Idea
> > > > what kind of averageing method Equis uses to plot there ATR!!!!   I
> may
> > > > start using  the  MOV( ATR(1), 14, S ) after this discovery.
> > > >      Thanks for sharing your findings!
> > > >
> > > >     Adam Hefner
> > > >
> > > > ----- Original Message -----
> > > > From: "Ian Burgoyne" <iburgy@xxxxxxxxxxx>
> > > > To: <metastock@xxxxxxxxxxxxx>
> > > > Sent: Saturday, July 15, 2000 11:15 AM
> > > > Subject: Re: atr
> > > >
> > > > > Adam,
> > > > > I've come across this before with comparing the ATR to a moving
> average of
> > > > > the ATR and found them to give different values. See attached chart
> which
> > > > > gives an example.
> > > > > In the data window shown (in red type)the top value is
> mov(atr(1),10,e)
> > > > and
> > > > > the value underneath is mov(atr(1),10,s) and the bottom value is
> > > > > atr(10). I think the "average" in the ATR calculation is a different
> > > > method
> > > > > at least in comparison to an EMA and SMA.
> > > > >
> > > > > regards...Ian
> > > > >
> > > > >
> > > > > >From: "Adam Hefner" <vonhef@xxxxxxxxxxxx>
> > > > > >Reply-To: metastock@xxxxxxxxxxxxx
> > > > > >To: <metastock@xxxxxxxxxxxxx>
> > > > > >Subject: Re: atr
> > > > > >Date: Fri, 14 Jul 2000 08:00:57 -0500
> > > > > >
> > > > > >Al,
> > > > > >     If you used a ATR(1) and then plotted a 10 day simple moving
> average
> > > > > >of   this ATR(1)..... it should calculate the same value as an
> ATR(10).
> > > > > >Now  if you needed an "Exponential ATR" you could plot a 10
> exponential
> > > > > >moving average of ATR(1).
> > > > > >    I believe this is what the previous ( Mike) e-mail was trying
> to
> > > > show.
> > > > > >
> > > > > >     Adam
> > > > > >
> > > > > >
> > > > > >----- Original Message -----
> > > > > >From: "Al Taglavore" <altag@xxxxxxxxxx>
> > > > > >To: <metastock@xxxxxxxxxxxxx>
> > > > > >Sent: Friday, July 14, 2000 3:10 AM
> > > > > >Subject: Re: atr
> > > > > >
> > > > > >
> > > > > > > ATR(1) would simply be the true range for one day.  I fail to
> see the
> > > > > >value
> > > > > > > of taking an "n" day moving average of one day.  I am looking
> for the
> > > > > > > average true range of price over "x" period of days....what is
> the
> > > > > >average
> > > > > > > price movement for the past 10, 50 day period.  If, as is the
> case for
> > > > > >WMT,
> > > > > > > the 10 day ATR is   2 2/16, and the 50 day ATR is 2 7/16, after
> price
> > > > > >has
> > > > > > > moved, during the trading day, 2 points, I would anticipate
> little
> > > > > >reward
> > > > > > > to buy the stock as I could only presume a futhur movement of
> 2-7
> > > > > > > sixteenths.  If however, the stock fell 2 1/2 points, I have a
> low
> > > > risk
> > > > > > > entry point for a countertrend trade.  If I owned the stock from
> a
> > > > lower
> > > > > > > price point, after the 10/50 day ATR is reached I have a good
> exit
> > > > point
> > > > > > > for my day trade.
> > > > > > >
> > > > > > > Al Taglavore
> > > > > > >
> > > > > > > ----------
> > > > > > > > From: Bob Jagow <bjagow@xxxxxxx>
> > > > > > > > To: metastock@xxxxxxxxxxxxx
> > > > > > > > Subject: RE: atr
> > > > > > > > Date: Thursday, July 13, 2000 8:28 PM
> > > > > > > >
> > > > > > > > Right. The Equis ATR(period) matches Wilder's original version
> and
> > > > the
> > > > > >TR
> > > > > > > > isn't a builtin.
> > > > > > > >   ATR(1) is actually the TR so taking its ma will give SMA or
> EMA
> > > > > > > versions
> > > > > > > > of ATR -- Chande uses the SMA for stops.
> > > > > > > >
> > > > > > > > Bob
> > > > > > > >
> > > > > > > > -----Original Message-----
> > > > > > > > From: owner-metastock@xxxxxxxxxxxxx
> > > > > > > > [mailto:owner-metastock@xxxxxxxxxxxxx]On Behalf Of Mike
> Campbell
> > > > > > > > Sent: Thursday, July 13, 2000 2:18 PM
> > > > > > > > To: metastock@xxxxxxxxxxxxx
> > > > > > > > Subject: Re: atr
> > > > > > > >
> > > > > > > >
> > > > > > > > Al Taglavore writes:
> > > > > > > >
> > > > > > > > > Neither.  As Welles Wilder developed it, a moving average
> was not
> > > > > >used.
> > > > > > > > > MetaStock has it programmed.  Simply pull up the indicator
> and
> > > > type
> > > > > >in
> > > > > > > the
> > > > > > > > > number of days.  Today's ATR is the distance from today's
> low to
> > > > > > > today's
> > > > > > > > > high OR from yesterdays close to today's high.....whichever
> is
> > > > > >greater.
> > > > > > > > > This accounts for any gaps from the previous close to the
> low of
> > > > the
> > > > > > > > > current day.
> > > > > > > >
> > > > > > > > I believe you are mistaken there.  What you described is the
> "true
> > > > > > > > range" calcuation.  ATR is some moving average of THOSE
> values.
> > > > > > > >
> > > > > > > > Otherwise, what would the "number of days" have to do with it?
> > > > > > >
> > > > > >
> > > > >
> > > > >
> ________________________________________________________________________
> > > > > Get Your Private, Free E-mail from MSN Hotmail at
> http://www.hotmail.com
> > > > >





  • References: