jmont1 

As mentioned before Renko bars are notoriously erratic for strategy Analysis. What makes things worse is they are not able to use "Tick Replay" that forces the ticks in the original millisecond they occurred. Just ran another Strategy Analyzer using multiple offsets to the exits. They are all coming back with the same result. This might be accurate but of course concerns me.

I am very concerned about this.

Enough so that I'm going to set things aside for the moment and go back to the drawing board. My plan is to test this same basic approach using Range bars instead of UniRenko.

Unlike UniRenko bars or even standard Renko bars, I do know NinjaTrader "supports" Range bars. The thinking here is that results seen in the Strategy Analyzer should more closely mimic live trading, at least in theory.

Range bars are created within Ninja using Tick Data. Therefore any back testing will gobble up a ton of computer memory. So I'll just test one week at a time and see where that leads.

I'd hate to keep spending tons of hours chasing a UniRenko strategy that cannot be reproduced live. Most likely I'll just whip up some sort of mini-scalp strategy that takes a lot of trades. Doing so will allow me to better watch it live in action. Then I'll compare this mini-scalp approach side-by-side using two charts. One chart with UniRenko bars and the other with Range bars.

We'll flush out this critter in a hurry.

OK, if it crashed let's take the Range out since it is not critical to the core. I will see if anything else can be made simpler.

Range bars are more likely to provide correct back testing so with the "knobs" we can try and find good settings.

Please look in your "Log" tab and find the time that it stopped. See if there is an error message there.

Update: @Fugitive69 This version has your PSAR's as lines on the chart.

Range is removed and so is the 30 second timeframe that supported it. So we are back to your core to see how to proceed next.

I've never looked at the Log Tab before - DUH. Lots of good stuff in there.

Around the time of the crash it posted this in the Log:
Strategy 'kKrunchJMont2': Error on calling 'OnBarUpdate' method on bar 3068: You are accessing an index with a value that is invalid since it is out-of-range. I.E. accessing a series [barsAgo] with a value of 5 when there are only 4 bars on the chart.

This was early evening, about 2 hours after the night session opened. Maybe it just needed more bars from market open to evaluate a portion of the strategy, don't really know. But I also find this error in the log as well somewhat close to that same time frame:
Strategy 'kKrunchJMont2/-1': An order placed at '10/29/2018 5:02:36 PM' has been ignored since the order was submitted before the strategy property BarsRequiredToTrade had been met.

If you rework and rename the PSAR indicator to show lines instead of dots, I usually have a lighter green for the "slow" line and a goldish color for the "faster" line. Whatever might work on both a dark and a light background is fine with me actually.

Thanks for your help.

fugtive69, OK, I think this version should work well. I lowered the bar required count to avoid that error and the range may have caused the other one so it is removed.

Back to back testing now.

Tried to get an indicator to create a correlated Index of the currencies "6's" such as 6A, B, C, E and DX. Unfortunately I failed in converting the NT7 version of "Index.cs" which seems close if not exact to what I am looking to accomplish.

But I am not going back to NT7.

So I stumbled into this possible correlation of 6E, DX and ES that at least back tests well. But I have not had a lot of live experimenting with it while I was distracted with another opportunity. I am posting it here with the hope someone else can figure a way to improve on it.

The basic of it I that if 3 bars (typical white soldier, black crow) is across these instruments it is time to enter a trade. The setting is DX opposite 6E and ES. So if DX is down the others are going up and vice versa.

I have run it on a 2 1/2 minute chart set as 150 seconds.


Please let me know if you have any good ideas to extend this theory OR a way to get that composite index indicator working.

I am placing this here rather than in the strategy thread since it is too new to confirm that I would call it a working strategy.

Good Trading ALL !!!

Update: GOing to tweak this a bit further before reposting.

Trying using the PSAR - Kkrunch is a bit too sophisticated for me yet. But I will probably try it later.

jmont1 

Tried to get an indicator to create a correlated Index of the currencies "6's" such as 6A, B, C, E and DX. Unfortunately I failed in converting the NT7 version of "Index.cs" which seems close if not exact to what I am looking to accomplish.

I feel I've seen something like this within the past few months. I'll dig around to see if I can find it again.

Spent a considerable amount of time trying to correlate these two bar types without success. They are just two completely different animals. The short version of the story is that I'm going to concentrate on Range Bars for awhile and see what I can uncover.

Toward that end here's a revised kKrunch indicator. Seems to work best for me when Range Bars are set to 10.

Although I'm forced to go back to the drawing board the good news is that trades that show up on the Strategy Analyzer also show up on a live chart (unlike with UniRenko bars). Still a long way to go and tons of tweaking to do but the glass is half full - there is hope.

Fugitive69 

Spent a considerable amount of time trying to correlate these two bar types without success. They are just two completely different animals. The short version of the story is that I'm going to concentrate on Range Bars for awhile and see what I can uncover. Toward that end here's a revised kKrunch indicator. Seems to work best for me when Range Bars are set to 10. Although I'm forced to go back to the drawing board the good news is that trades that show up on the Strategy Analyzer also show up on a live chart (unlike with UniRenko bars). Still a long way to go and tons of tweaking to do but the glass is half full - there is hope.

@Fugitive69, looked at the modifications and see you are using higher values but cannot be sure if everything has adjusted.

Assuming you are still using a high value ( ~ 1000) and a lower value like the old 250 I recommend you add the lower value lines on the indicator panel. I.e...

AddLine(Brushes.YellowGreen, 250, "250Line");
AddLine(Brushes.YellowGreen, -250, "Minus250Line");

Or something similar. Then it is easier to visualize the entry confirmation.

Although I would not bother putting a third line in, I do recommend you add a third parameter to your entries which I probably already mentioned. So the previous > 500, < 250 should aslo have a minimum on the second bar like <100 so drastic reversals do not put you in just because it is below the 250,

A MAJOR benefit would be derived by making your line values variables. So you can easily adjust them for different instruments. This assumes that Kkrunch for ES would use different values than NQ or GC for instance. Should not hard to do. And likely save you time after the initial pain putting in six (6) variables.

=====

Aside from the more accurate coloring based on the bar dominant value do you plan to act on that? No need for an immediate answer. Best if it is answered when you post an updated strategy suing this indicator.

My draft strategy is trying to add in the PSAR and currently used it for a basic reduction of number of trades. I was about to start using the Kkrunch which I have not been able to determine how well it adjusts the win ratio. It is a bit surprising how radically different the bar values are for V2 to Ver3. Using the data box with both on the chart shows a marked difference in values.

I had noticed that there was a latency for bar value and direction change in V2 and ver3 does seem more responsive.

Your post is just in time for me to start using the new version rather than post with the old one if I decide to keep that element in the draft.

I have not started looking at range bars yet but I do think it is a better setup than Uni's since it probably backtests better.

Good Trading All !!!

U P D A T E: By making the variables for Kkrunch you will be able to do an optimization with Strategy Analyzer to possibly tweak the lines to a btter area, say 750 instead of 1000 or even 650.

Of course there is a likelihood of "curve fitting" to your test data but still would be interesting to be able to tweak it like that.

@jmont1 - Good points here, thank you.

I think the scale got away from me when I started using 10 Range Bars and only realized after the fact the inherent problems it created. So I made a few changes:

[1] Reduced the scale by a factor of 10 to make everything more manageable.

[2] Allows the user to select both the Upper and Lower lines of the oscillator (defaults are 150, -150)

[3] Colored the region between the Upper and Lower lines

It's very early yet and I'm testing 3 different approaches. Maybe by Monday I'll have something more concrete to share.

Oddly enough I can do quite well on the longs but can't get to first base on the shorts. Very intriguing.

But that's OK since my primary goal above all others is to ensure I am able to reproduce on a live chart what I see on the Strategy Analyzer. Otherwise I'm just spinning my wheels. In the mean time I'm receiving on heck of an education so it definitely feels like a win win.