Kreslik.com - traders community since 2006

Techguy wrote:I'm currently conducting an optimization test on whether or not the ATR can filter out bad breakout trades due to low volatility.

My initial premise was that if the ATR is under 2 pips on 1 minute bars, that would be a good way to filter out trading during times of low volatility.

My suspicion is that low volatility makes for poor trading conditions, especially during break outs.

My thinking is that Instead of making trading hour filters, just let the ATR filter out the low volatility time periods.

Your thoughts???

I will post my results when I have them.

Techguy wrote:Test Data Set: EURUSD Jan 01, 2017 to June 01, 2017, using 5 min bar data. Based on a lot size of 1.00.

The results were a mixed bag:
For my breakout code, the data distribution was only generally conclusive, as the scatter plot showed that parameters did not always correlate to like results.

However in general the test results were:

Very Low ATR's of 1 to 2 pips yielded lots of trades (880), low profit factor (1.02-1.04), and typically high drawdowns up to 31% (29 to 31%)

Lower ATR's, for example in the 3 to 5 pip range yielded typically about 200 fewer trades on average (514 - 902) the highest profit factor (1.08 to 1.14), medium drawdown (17 - 30%)

High ATR's, for example 7 to 9 pips, yielded low trades, low profit, and the lowest drawdown.

NOTE: RETESTING IN THE STRATEGY TESTER HAD WORSENING RESULTS USING ATR OPTIMIZED PARAMETERS.

My conclusion:
Don't use it.
I used an ATR filter of 3 pips, with an averaging of 1 bar. Results were bad for me. Multiple bar averaging in the ATR calculation did not appear to make trading worse or better.

See attached report to draw your own conclusions.
OptimizationReport-ATR-Sort by Profit.gif

Techguy wrote:Test Data Set: EURUSD Jan 01, 2017 to June 01, 2017, using 5 min bar data. Based on a lot size of 1.00.

The results were a mixed bag:
For my breakout code, the data distribution was only generally conclusive, as the scatter plot showed that parameters did not always correlate to like results.

However in general the test results were:

Very Low ATR's of 1 to 2 pips yielded lots of trades (880), low profit factor (1.02-1.04), and typically high drawdowns up to 31% (29 to 31%)

Lower ATR's, for example in the 3 to 5 pip range yielded typically about 200 fewer trades on average (514 - 902) the highest profit factor (1.08 to 1.14), medium drawdown (17 - 30%)

High ATR's, for example 7 to 9 pips, yielded low trades, low profit, and the lowest drawdown.

NOTE: RETESTING IN THE STRATEGY TESTER HAD WORSENING RESULTS USING ATR OPTIMIZED PARAMETERS.

My conclusion:
Don't use it.
I used an ATR filter of 3 pips, with an averaging of 1 bar. Results were bad for me. Multiple bar averaging in the ATR calculation did not appear to make trading worse or better.

See attached report to draw your own conclusions.
OptimizationReport-ATR-Sort by Profit.gif