- ago
Hello fellows! Maybe it's a stupid question but it's Monday and I am too lazy to think:

Is it the same an upscaled indicator than the same indicator in the same timeframe with a longer period? Example: Aroon Uptrend (25) upscaled from a 15 minutes analysis to 60 minutes versus Aroon Uptrend (100).
1
313
Solved
2 Replies

Reply

Bookmark

Sort
Cone8
 ( 28.25% )
- ago
#1
The answer is:
1. could be exactly the same,
2. could be close to the same,
3. or may be quite different

.. depends on the indicator.

For AroonUp, I think you'll find #2 is the answer -

CODE:
using WealthLab.Backtest; using System; using WealthLab.Core; using WealthLab.Indicators; using System.Collections.Generic; namespace WealthScript123 { public class CompressionSample : UserStrategyBase {        //create indicators and other objects here, this is executed prior to the main trading loop public override void Initialize(BarHistory bars) {          if (!bars.IsIntraday || bars.Scale.Interval > 15)          {             throw new InvalidOperationException("Example is intended for an intraday BarHistory, 30-min or less");          }              _arUp = AroonUp.Series(bars.Close, 100);          PlotIndicatorLine(_arUp);                              TimeSeries compClose = TimeSeriesCompressor.ToMinute(bars.Close, 60, bars.Market);          _arUp2 = AroonUp.Series(compClose, 25);          // expand to the base scale and plot          _arUp2 = TimeSeriesSynchronizer.Synchronize(_arUp2, bars);              PlotTimeSeriesLine(_arUp2, _arUp2.Description, _arUp.PaneTag, WLColor.Fuchsia);       } //execute the strategy rules here, this is executed once for each bar in the backtest history public override void Execute(BarHistory bars, int idx) {           }       //declare private variables below       AroonUp _arUp;       TimeSeries _arUp2; } }


2
Best Answer
- ago
#2
I think if it's a "linear" indicator, then whether it's upscaled or compressed, the results would be about the same. Even for the ATR indicator, where the next value is a function of the previous value, you would expect some differences, but the results would be "somewhat" close. Most indicators are based on linear methods.

Now for a nonlinear indicator, the results would be totally unpredictable. I have one I created myself. Even though it's based on a robust statistical analysis, there would be certain events that would appear at a higher resolution scale that would be absent on a compressed scale. So the robust analysis won't remove all the weird behavior--unfortunately.

The Median indicator and the TimeSeries.PercentRank property are technically nonlinear transforms. Yes, their results might be unpredictable when compressed, but I would like to think they wouldn't get too weird because we use them frequently in robust statistics. My point is, just because it's nonlinear doesn't automatically mean it gets weird when compressed.
1

Reply

Bookmark

Sort