Why Keltners in Stockcharts so different from ToS's?

[floydddd]

Anyone know why the Keltner results in ToS are so different from those at stockcharts.com? See pix for one example, plus the ToS code I'm using that allows for 3 different inputs to match what stockcharts allows. i've tried changing up all those average types but that doesn't seem to matter. Any other possibilities for why and/or what to do about it? generally speaking, i believe the stockcharts numbers are the accepted standard, so it'd be nice to get ToS to align with them.

for the example, the stockcharts low band reading at the close is $60.79. the tos reading is $62.52.

thanks!

declare upper;
declare weak_volume_dependency;

input displace = 0;
input factor = 3;
input length = 20;
input lengthATR = 10;
input price = close;
input averageType = AverageType.EXPONENTIAL;
input trueRangeAverageType = AverageType.WILDERS;

def shift = factor * MovingAverage(trueRangeAverageType, TrueRange(high, close, low), length);

def average = MovingAverage(averageType, price, length);

plot Avg = average[-displace];
Avg.SetDefaultColor(GetColor(1));

plot Upper_Band = average[-displace] + shift[-displace];
Upper_Band.SetDefaultColor(GetColor(8));

plot Lower_Band = average[-displace] - shift[-displace];
Lower_Band.SetDefaultColor(GetColor(5));

 

[armybender]

Well, my first thought is "is your trading system sensitive enough that the differences matter?" I have an extremely difficult time reasoning through a scenario where that difference will matter. Channels are extremely "imperfect" in the first place because they just show a range of possibilities based on history, which can't account for a zillion other things that actually drive market prices... like open interest at specific levels, for example. That's real vs a channel which is "fake" because it's just an arbitrary calculation based on arbitrary parameters.

However, regarding your question specifically, the documentation on StockCharts.com shows exactly how they're calculating the indicator. It doesn't talk about using a Wilders average though, so perhaps that's the difference. It could be a different average type. You're also using a "factor" of 3, but their formula shows an addition or subtraction of 2x the ATR, not 3x. Those are the most likely culprits. I think you're comparing apples and oranges, to some extent.

Why don't you just duplicate their exact indicator in ThinkScript code and see if they truly are different if you use the same parameters?

Here's the documentation from their page. This was simple to find.

https://chartschool.stockcharts.com...-overlays/technical-overlays/keltner-channels

Keltner Channels Calculation​

There are three steps to calculating Keltner Channels. First, select the length for the exponential moving average. Second, choose the time periods for the Average True Range (ATR). Third, choose the multiplier for the Average True Range.


Middle Line: 20-day exponential moving average
Upper Channel Line: 20-day EMA + (2 x ATR(10))
Lower Channel Line: 20-day EMA - (2 x ATR(10))

 

[rad14733]

@floydddd Have you verified that ALL settings parameters are identical between the two versions... Some platforms use a Simple Average for both length and factor, as well as different lengthATR value... For my Keltner Channels I use the same parameters as your code above with the exception of using a factor of 1.5 or 2 and a lengthATR of 14 (ATR Default)...