Range trading on EUR/USD 1h timeframe
Forums › ProRealTime English forum › ProOrder support › Range trading on EUR/USD 1h timeframe
- This topic has 14 replies, 4 voices, and was last updated 6 years ago by
Vonasi.
-
-
09/09/2019 at 8:10 AM #106791
The system uses the ADX to filter out trending time.
A combination of 3 indicators used (Stochastic, RSI, CCI) to generate trade signals.
– From those 3 oscillators any 2 crosses above or under they trigger levels at the same time a LONg or SHORT signal is generated.
Stop Loss is defined as multiplication of ATR.
The code has a risk management built in to dynamically calculate position size according to risk taken.
09/09/2019 at 8:14 AM #106901Hi, thanks a lot for this original idea. I’d like to push it into the library but for an unknown reason the backtest has odd behavior this morning (see attached picture), do you have screenshot of the equity curve? Did you backtest prior to January 2018?*
09/09/2019 at 10:44 AM #10692209/09/2019 at 1:27 PM #10695209/10/2019 at 7:52 PM #10710909/10/2019 at 8:16 PM #107116GraHal – Yours is a perfect first candidate for my Robustness Tester – Random!
Here are the results of 45 tests of opening positions in random like formations. We are looking for as flat a line as possible with values as close as possible to each other.
I’m afraid it is a fail!
On a brighter note at least all the tests were in profit. Unfortunately some of the equity curves were horrible!
09/10/2019 at 9:11 PM #107123I have not been backtesting it for more than 10K candles because it is not a holy grail to work all the time in every market conditions.
I backtest mostly 3 months on specially selected markets. As I don’t try to force my strategies on markets but use the right tool for the current condition.
I specifically said that it is a range trading strategy (it won’t work in trend simple as that).
ADX[100] < 15.0 is to pre-filter market conditions for range only.
If a market is historically unlikely to range this strategy will not work on it. But say CHF/JPY hardly ever trend so as EUR/USD since a year.
1st you select a market which is mostly ranging and never trending then apply this strategy and optimise values to last 3-4 months data.
Values to optimise (important in this order):
1. Switch off (comment off) money management all completely and set a fix positionLong = 1, positionShort = 1
2. Comment off all stop loss (no stop loss at all) and ATR
3. Set all values to default. (Stochastic[20,5] trigger bottom = 20.0, top = 80.0); (RSI[14] trigger bottom = 30.0, top = 70.0); (CCI[100] trigger bottom = -100, top = 100)
4. Optimise CCI[length] (min=50, max=200, step=50) / CCI trigger bottom line (min=-300, max=-50, step=50) / CCI trigger top line (min=50, max=300, step=50)
5. Optimise RSI[length] (min=5, max = 20, step= 1 or 5) / RSI trigger bottom line (min=10, max=40, step=5) / RSI trigger top line (min=60, max=90, step=5)
6. Optimise Stochastic[length,smooth] (length min=1, max=20, step=1 / smooth min=1, max=10, step=1) / Stochastic trigger bottom line (min=10, max=30, step=5) / Stochastic trigger top line (min=70, max=90, step=5)
7. Uncomment ATR and stop loss
8. Optimise ATR[length] (min=1, max=20, step=1) / and stop loss myATR multiplayer separately for long and short stop loss (min=1, max=10, step=1)
9. Uncomment money management.
Results are as best as they can be.
These steps are to be taken in strict order.
Never apply to market which is likely to trend.
09/10/2019 at 9:32 PM #10712809/10/2019 at 10:51 PM #107136does it have similarities to monte carlo testing?
Yes – except it doesn’t just randomise the trades you have in your strategy orders list like some of the popular monte carlo testers do – it applies your strategy to a market and only allows it to trade in random sequences. This means that each test can have completely different trades in it to those in another test. You really should get to grips with it as it is a very powerful tool and can save a lot of time (and hopefully money)!
09/10/2019 at 11:01 PM #107137How does aglit996 version (in the 1st post) look on your tester Vonasi?
I’m afraid not very well but then again I’m not surprised as I got the same equity curve that you did in your first post and that is not a great starting point for any test!
Image attached of robustness test results anyway – even if they are very ugly.
09/10/2019 at 11:04 PM #10713909/11/2019 at 11:32 AM #107182Results are as best as they can be. These steps are to be taken in strict order. Never apply to market which is likely to trend.
Thank you for taking the time to specify optimising order etc.
range trading strategy (it won’t work in trend simple as that).
The equity curve for your settings and even my settings ( 🙂 ) confirms above.
What I liked about my version is that even in the very obvious uptrend, the equity curve stayed more or less around zero … no big losses!
Of course the truth will out (as they say) in Live Forward Testing. I will report back when I have about 50 trades (Demo Live) on my version.
If Vonasi is right re his robustness tester results then I’ll be crying in my beer! 🙂
Now my coffee has worked through … I’ll go and get my head around the VRT (Vonasi Robustness Tester).
Thank you for sharing your strategy with us all @aglit996
09/11/2019 at 12:03 PM #107185If Vonasi is right re his robustness tester results then I’ll be crying in my beer!
You could get lucky and hit a run of good trades and get a great equity curve in just 50 trades. The robustness tester says that in the long run however the strategy might not perform exactly as it did in the exact sample of trades that resulted when the strategy was created.
09/11/2019 at 12:34 PM #107189The robustness tester says that in the long run however the strategy might not perform exactly as it did in the exact sample of trades that resulted when the strategy was created.
Surely above is what we all expect and accept anyway … history never repeats itself exactly etc.
I do understand where you are coming from … your VRT is an indication of the expected closeness of results between Demo Backtest compared to Live running.
09/11/2019 at 1:19 PM #107191No! It is making sure that we back test our strategy with it opening random trades.
Imagine if you set an auto strategy running but it kept on getting stopped and so you had to set it running again at random times. If your strategy is robust then it should not matter at what times you set it running or what times it trades. If however your strategy is not very robust then you might suddenly find a very different equity curve.
It is probably easier to consider one of the earlier robustness testers that I wrote – the one that does just two tests – one test where it only allows trades to be opened on even numbered bars and the other where it only allows them to be opened on odd numbered bars. If a strategy is robust then it should make no difference whether you decide to trade on odd bar candles or even bar candles. Then just multiply that concept up to 45 tests or more all with different allowed trading sequences. If the strategy is robust then all the results should be pretty even in average gain and % of winners.
Perhaps this discussion would be better taking place on the robustness tester thread as it has drifted a little from the subject title – my fault entirely for introducing the test to the analysis of the original strategy.
-
AuthorPosts
Find exclusive trading pro-tools on