I remember watching FT71 webinar and seeing an indicator on his non Ninja platform in which there is a horizontal line and a distribution of historical rotations (swing size in ticks) - so one can statistically understand visually what are "normal" rotations and what are extremes /deviations.

I know the swingPro indicator shows the last swing size average rotations - but not statistical distributions around some baseline.

So was wondering if anyone coded something similar fr NT?

Cheers,
Steve

The following user says Thank You to ziggy123 for this post:

I installed the indicator using it on range charts - it does indeed seem as if the numbers do not correspond to the tick swing distribution in CL - unless I'm missing something?

I'm not at my pc at the moment so will have to check in a little while, what settings did you use with the indicator? And also what was the range chart bar size you used? I will test it out.

With the settings, RotationSize is in ticks (minimum movement before new swing) so maybe something like 5+ for CL? And the DistributionDays is the number of previous session days to include in the distribution calc for 1SD & 2nd SD. Probably 20+ is good, but can be 1000 if you want all past days.

I was actually looking at a Renko type chart and not range (My bad) - however - I switched to 1 Minuet chart - and the distribution does not make sense.

I'm assuming that I'm seeing the number of rotations (Swings) distributions and the distributions of the LENGTH of swings? It's the LENGTH distribution which Im interested in.

The distribution is counting rotations at each size (length of rotation) before reversing a minimum amount as specified by the rotationsize value, at the moment it looks like this is 1 tick on your indicator, try making it 5.

ok - i think I see now that the "whole numbers" (ie:3 on the chart should be multiplied by x10 to represent the actual ticks)

So the BLUE lines are the standard deviation - and if I change the setting of rotation to 5 - it means fractal looking at two bars forward and backward to find a "swing" point and then in this specific scenario (attached image above) 30 cents (3) is still within the standard deviation - but there is no reference to actual FREQUENCY NUMBER (You have to count the vertical similar bars) - or I can use the SwinGPro indicator to see number of swing count average, length etc.

Sorry for the late response, there are a couple of differences from FT71's study;

I use same calculation as ZigZag indicator for rotations, both should be similar, but main difference is in FT71's 1min fractal study has a minimum number of bars between fractals, where mine will calculate at every bar. This can cause issues if you are trying to calculate rotations that are smaller than a single bar size (like a 5 tick rotation on CL via 1min bars would be every bar). Best way to overcome this is either use a high resolution tick based bar (like a renko) or a higher rotation value on minute bars.

Also in FT's study, up and down rotations are added to the same bin, I calculate them separately and apply 1st & 2nd SD lines to each separately.

I can see in Ft's study it is using 8 tick minimum rotations & calculating on a 20 session distribution, so to get a similar result try;

DistributionDays: 20
RotationSize: 8

Also make sure you are using a session template only for the rotations you want to calculate in the distribution and have a minimum of 20 days historical data loaded.

Rotations that exceed the 2SD are worth attention, usually representing a shift in the market balance or perspective. Especially if one lines up with a key area of interest you might have marked.

There are some stats which I put together which I also use to assist in entry position, depending on what the larger context of the trade is (if I am looking to go with or fade a 2SD rotation)

Here are some stats (based of last 3yrs of ES RTH, minimum 4t rotation calculation);

Using something like Jigsaw or Bookmap might assist in seeing areas where pull-backs may start to form.

Once a rotation pulls back gets to its mid, it has really become 50/50 so you have to decide what your risk is, or have already build a position via scale in/outs.

For Rotations over 1SD (but under 2SD) there is a high probability (70%) that they will hit their mid, but only 5% of trades reach 2SD so there is an edge in being skilled in catching the turning point of a 1SD rotation back to its mid.

I don't just blindly trade the rotations though, I wait for an area of interest then use these rotational probabilities to build a position (small scale in/out) to buy time and see if it plays out in my favour, goal being to move the risk point further away.

Hope that might give some ideas

The following 13 users say Thank You to amoeba for this post:

Thanks you for sharing, this is quite interesting. But, isn't this expressing the same information as plotting volatility/ATR bands on the price panel? Please forgive my ignorance here. Also wouldn't the rotations be different for each periodicity and therefore meaningful only to the periodicity which you are trading?
Wilson

I haven't done any analysis around ATR so I'm not too sure what its outcomes would be. You could possibly do a statistical study on individual ATR shifts and calculate a distribution to see what constitutes large single ATR moves.

I suppose the difference is this would be a derivative function of price where the rotations are a direct function of price movement.

The indicator attached to this thread is optimally used with tick based charts like renko, as time based bars can contain multiple rotations inside them which would get lost.

For time based bars, the same concept but applied through fractal rotations would be better, I have not made an indicator for this but from other studies the statistical outcomes would be the same.

The following 2 users say Thank You to amoeba for this post:

Thanks Amoeba, your rotation study is great!
Do you happen to have it for Ninja 8?
Would you also happen to have any indicators that would help a Ninja Trader chart look more like FT71. i.e. printing the rotation size and price below the candle on the chart.

NT8 version is something on my list, just time poor at the moment & my primary platform has been SierraChart for a couple years now. I do still use NT for analysis though.

As for the printing rotation size, there is a great indicator already around called PriceActionSwing - have a search for that, I believe they have an NT8 version available too.

The following user says Thank You to amoeba for this post:

Here is a first release for NT8, I have only given it a quick test in historical charts + a little bit of market replay, so if you have time please give it a good check.

Cheers!

The following 4 users say Thank You to amoeba for this post:

Thanks. Since seeing Morads first video i have been interested in this style. He had some other tools that let him know how much heat and draw down he could take and how many ticks of profit he had a positive expectancy of reaching.

Hi thanks a lot for that indicator, I am member of ft71 forum and would like to know if its posible to add more features to the indicator so it can show more and better information of the rotations. Thanks in advance

For example could be great to show the number of points or ticks in the top of the rotation, also could be done something similar as the priceactionswing indicator that shows the tremendous information of the swing. but this are just some easy ideas...

I'm wondering why I'm only seeing the histogram at the bottom, but no the distribution on the right or anything else. What am I doing wrong? Should I use minute, range, ticks; and what value?

Hey, thanks a lot the indicator is working awsome... I just wanted to point some improvemente we could do.
As FT71 is using, we could also mark in the rotations indicator the number of points of each rotation. Also could be great to do it with the cumulated delta indicator but this another history...Hope someone could help.
Another improvement could be to chose between points or ticks on the rotation, as now it is only showing ticks
thanks
Here a better screenshot

The following user says Thank You to pep17 for this post:

Reviving the "older" thread a bit.
This is indeed a great indicator.
@pep17 I'm curious what it is you're asking. I'm sorry but your pictures are kinda tiny and won't expand well.

Where are you asking the rotations indicator to be marked for the number of points for each rotation?
Where are you asking for the number of point in each rotation to be displayed?
What precisely are you asking for the Cumulative Delta indicator to do?
I apologise if I don't follow what you're saying I just want to make sure to get it right.
Cheers

Hi ! I came across this concept from FT71 like everyone is talking about here. Bigups to amoeba for sharing the updated version . However , I am having doubts as to how should I calculate what to put on the the parameters . FT71 says he uses it for the last 20days(lets assume no major vix change)

So when I download the 1 min OHLC data on my excel , should I limit my study of fractals to a particular timeframe ?(as he talks about timeframe as well) . Say I trade 3 min timeframe and hence I put 6 mins ie 6 bars and see the maximum upmoves without cutting the current or previous low and similar for downmoves? Then calculate both positive moves and negative moves and feed it on the indicator?

Please suggest how I shall calculate the rotation for the market I trade.

Not sure if this is exactly what you are after but I have another indicator that uses fractals to calculate the rotations then bins the rotation lengths to calculate 1st & 2nd SD's.

It has two inputs, number of bars included in the fractal pivot & minimum ticks between H's & L's.

I believe FT used 5bar fractals to calculate his rotations. (this could have changed)

The following 2 users say Thank You to amoeba for this post:

Thank you for the quick response !
BTW , I like the statistics you provided for the last indicator and if you could please provide a perspective on how I can calculate the rotation on excel so that I can feed proper data into your indicator . I did try my best but different people do it differently . But your indicator(the last one , not this one) looks a lot like FT .

What I had done was..
Basically what I had naively done was fix a time period . I downloaded 1 min data(I have access to 1 minute open high low close for the last 40days) .

My logic was based on 6 minutes ..so I have six 1 minute OHLC . then I used a logic to see if current 1 minute bar's low is greater than last 1 minute bar's low and I continued it for 6 bars and saw on average ,in 6 minutes, how much continuously price is making higher highs and higher lows . The calculation stops if it cuts the last 1 min bar's low and I calculated the retracement. This was for upside moves. I did the same in reverse for downside move. But while using renko or range bar , its devoid of time WHICH MAKES MORE SENSE TO ME .

Should I do the same study without the time filter and see everyday for the last 40 days how much price moved continuously up vs down and feed it to your indicator ? Anything you suggest? Please reply on your own time . And thanks a lot again for the indicator . I am just confused as how to calculate it

I think this indicator does the calculation you are looking for, it is not for showing the current rotation size like the other one, I use this one to calculate what my 1sd & 2sd levels should be to input them into the other indicator.

If you have a look at the screenshot in the top left corner you can see what the 1Sd & 2Sd rotation sizes are based on (x) bar fractals.

So , I set the chart to 1 minute and load 20days chart and load the fractal and set the maximum lookback to ‘infinite’ rather than 256 bars so it looks into the whole 20 days (I assume)

Mode up and down is average (in ticks)? So if its 10 and my market’s minimum tick size is 0.05 , its 0.5 ticks avg?) and 70 percentile is approx. 1 sd and 90 is 2sd(shouldn’t it be 95?) ?
I feed the data in the previous rotation indicator . I put the mode up down value as rotation size , and 1sd and 2sd I put the values of 70 and 90 percentile respectively . While the indicator does not have a mark for 1sd and 2sd , I just remember the sd values and watch the points as they move and see if the points breach my 2sd values .

This is for 1 minute time frame and obviously I will do it at a higher timeframe .
So am I doing everything right here?

I will have to check my RotationsNT8v1 indicator to make sure it is doing what it is supposed to do, I cant remember how it is calculating its 1sd & 2sd, I will have a look and post back.

Okay thank you ! I will wait for your response .Also please let me know if I am using both the indicators properly . Another thing , I find you are using it on range bar . Shall I use it on renko ?

Lastly , I am new to this forum . Please guide me to discussions about rotation studies and strategies if you can . Also let me know how I can make it up to you for this fantastic indicator and serve the purpose of this community in my limited capacity . Thanks a tonne !

Ok, I had actually forgotten how the RotationsNT8v1 was working as I do not actively use it any more, I do use the FractalSwing indicator each week to get the rotation statistics which I use in SierraChart.

So after you run the FractalSwing indicator on a set number of days, use the Mode number to input into the minimum rotation amount which will get the bands to calculate close to the percentile outputs of the FractalSwing calculator.

Just one query . You mention that if mid of a 2sd rotation is hit , there is 48.5%chance of break top and 48.5% chance to hit rotation mid . Which rotation mid are we talking about here? The rotation move's mid is already hit . Are we talking about the mid of the countertrend as shown by the red line ? If so then isn't it a pullback and not necessarily a rotation ?

Might have been a typo in the diagram, I think it was supposed to be 48% to hit the 2sd rotation finish, or 48% to hit the 2sd rotation start (after Mid has been tagged). So essentially once mid is tagged there is equal chance* of the top or bottom of the rotation being hit.

* based on historical analysis, past performance not indicative of the future results, etc, etc..

Please see the attatchments . I got the 3 minute fractal mode and 1sd and 2sd data . I then made a unirenko chart with the brick size same as the fractal mode and put up the rotation indicator and fed the mode and 1sd and 2sd data. The issue I am facing is that I am getting almost all the signals outside 2sd . Not even 30 minutes in (and it stayed that way throughout the day). It has been the same ever since I have been using this and moreover today was a rangebound day with very narrow range . Is there something wrong that I am doing ?

Was the tick size 0.05 for the Nifty? the FractalSwing output is in the measurement as the instrument (not ticks), so Mode of 14 would be 280 ticks.

The RotationsNYv1 indicator is in ticks (maybe where the confusion is), so for the MinimumRotationAmount should be closer to 280.

Also, the 1Sd & 2SD need to be percentiles, so 0.7 & 0.9, which would be similar to the FractalSwing output. This might be what is causing you confusion as this indicator attempts to auto calculate the percentile bands for you, rather than you entering them manually like you have.

When I get a change I will make a v2 indicator that allows you to just put that value you want for each band. (so you can input the values from the FractalSwing indicator)

This version is different to v1 in that it does not attempt to auto calculate the percentile bands and allows you to just enter values. Also I have made the unit values consistent with the FractalSwing indicator so you do not have to convert to ticks.

The following 2 users say Thank You to amoeba for this post:

Thank you for being so patient with me ! So I am using the 1 minute information only . Feeding the fractals info to rotation study as it is . Say the mode is 8 points and tick size is 0.05 . SO ,Using a renko chart of the brick size 160 to get an even brick of 8 points . The chart looks normal now . Thank you !

You can also use a smaller brick size & just leave the MinRotationSize=8, this should give the same result but with better resolution, depends on your preference

I am not sure if anyone has been doing this yet or not, but I do not use this as a counter trend tool. I actually use this as a "continuuation of trend" tool.

I have 4 very hard rules I follow and if you'd like to help me build it into a strategy please send me a message!

Thank you,

The following user says Thank You to Ratdog for this post:

HI ! Just getting the hang of the indicators you provided and they are a blessing . Thank you so much . Just wondering when NInjatrader comes up with an update , I will have to update it as the other indicator packs are bought from a company which will update theirs . So in that case , will it be okay for you to provide the 2 indicators in the updated platform ? If so , how shall I get in touch with you ? Here only ?

HI ! Just getting the hang of the indicators you provided and they are a blessing . Thank you so much . Just wondering when NInjatrader comes up with an update , I will have to update it as the other indicator packs are bought from a company which will update theirs . So in that case , will it be okay for you to provide the 2 indicators in the updated platform ? If so , how shall I get in touch with you ? Here only ?

Hi ! Sorry to bother you again . Was wondering something about the fractal indicator you provided. There are 2 parameters. Number of bars , set to 1 and minimum ticks set to 3.Should I keep it default if I wanna watch number of rotations in 1 min for the last 20 days ?

Also , for past 20 days , I am loading 1 min chart and in ''data series'' I am loading 20 days and setting the your fractal indicator parameter "maximum bars lookback bars" to infinite. Am I doing it right ?

That setting should be fine for 1min bars, it would represent the most sensitive fractal pattern recognition (the NumberOfBars value represents the number of bars to count on either side of a fractal High or Low). So Higher values here create larger rotations as smaller fractals are filtered out.

The "maximum bars lookback" is actually a Ninjatrader setting, not something from my indicator, I would leave it as the default value (maximum bars lookback). You can have a look on the ninjatrader help docs to better understand this setting, but basically it is a limiter to how far back an indicator can look at historical bar data (as the further back an indicator looks the more ninjatrader is required to keep in memory). My indicator does not look further back than the NumberOfBars value.

You say your indicator does not look further back than the numberofbars. What I understand from your points is that it starts calculating from 1 bar only . The larger I make it , the larger the rotations will be(its not of much use making it so large as I will not get a continuous upmove or downmove of 95 percentile).

But what I am asking is I understand that 1 NumberOfbars(sensitive to lower fractals) rotations is what I am looking for BUT I am looking to see the statistic over last 20 days . So I load 1min chart and set days to "20" . Now your indicators need to calculate rotations for 1 bar but needs to know the start point and end point.So , I set the maximum lookback to 20 days.As I understand from your explanation of it as a ninjascript and also as it seems,its the correct way . Is it? That's all I guess. Thanks a lot

As many days as you have loaded into your chart via the data series is what the indicator will use as the sample set, the lookback variable has no effect here.

The indicator calculates from the bars loaded into the chart at load time, not looking back from the current bar, so if you load your chart and select Days to Load : 20, you will get a sample of rotations from the last 20 days.

The following user says Thank You to amoeba for this post:

I was trying to put hands on Rotation 1&2 tools. What input settings would you suggest for EURUSD, RANGE 40. I experimented myself also using FractalSwing with no material success. I read all thread and apllied all answers already.

I dont have EURUSD, so had to look at the 6E, not sure if the tick size is the same.

When you say Range 40, I take that to mean Range Bars of 40t. For these indicators, they only count rotations during a session, if the bar size is too large and no swings are occuring during the session it will not be analysed.

Maybe try a 5min or 30min chart and see how that works.

I took EUR, DELTA 300 and chose following MinimumRotationAmount ' s = 10, 30, 50 to test.

Setting of MinimumRotationAmount = 30 makes a great sense in two ways (1) signal 1 - prediction of reversal at 95 level, (2) in case prediction fails this failure becomes signal 2 - space for continuation. Which means this 95level with that setting in that context (asset, peridocity of bar, state of market) is a real pivot at the edge of consolidation. When i say signal it is not a trigger to place a trade based on it alone. It is more like graet contextual information.

So i call this value of 30 an optimal value on a chart. Now, the question is - is it possible to derive this optimal value with FractalSwing ?

BTW this type of questions resembles searching for dominant cycle with spectrum analizers, or searching for optimal renko brick value with renko analizers.

I don't use it quite the same way you are exploring - not to say there is a right or wrong way.

The FractalSwing indicator is a tool I use to calculate a distribution of all rotations and find the 70th & 90th percentile bands of those rotation sizes.

From that, I primarily look at 90th percentile rotations during the intraday session and try to interpret based on context, location & time of day.

The RotationsNT8v1 automatically calculates this distribution and displays the percentile bands, whereas the RotationsNT8v2 does not automatically calculate values and requires the user to enter the point value of each percentile band, this version works well in tandem with the FractalSwing indicator as you can use the values provided by the FractalSwing results.

IMHO i try to do exactly the same thing. If i understand you correctly - FractalSwing+Rot2 produces an output of what i call optimal value (as reffered to in my previous post).

When i run FractalSwing and copy its output into Rot2 it does not make sense. I copy them exactly the way you described in this thread. I receive from FractalSwing an output which looks typicaly sth like 0,002445 and i am unable to copy that into Rot2.

Hi ! Forgive me for bothering again but I was thinking of some things. Your fractal swing says 90 percentile but when I open the script it seems to say 2 standard deviations as 90 percentile . But 90% is 1.64 standard deviations. I am a bit confused here as I can't code and its kind of very important that I know how many standard deviations is it calculating. Can you bring clarity to this ? How do I get 2 and 3 standard deviations?Maybe I just multiply 70 percentile with 2 and 3? OR ..(70 percentile value - mode)*2 and then *3? I hope I could explain what I wanted to know. IN other words 70 percentile value shown is (Mode value+ 1*sd) right?

Even though it had been working with 90 percentile , I would like to tell you that I made a very silly observation. Its 90 percentile and its not sigma as we know it. Its the cumulative probability that 90% of all observations fall below this value. 95 won't mean 2 sigma...it implies 95percentile. Rotation such high is rare. Better to stick with 90 percentile

I just stumbled upon this indicator and it is great. Thank you.

Is there a way to filter this by hour of day. For example, the last 30 days I want to know rotation lengths from 6:30-8:30pst and the lengths from 8:30-10:30 pst..... I know the rotations are a larger in the morning, then they're smaller midday, and large again in the evening. It would be good to be able to make adjustments for different times of the day.

This should work, the code only uses Bars.IsFirstBarOfSession, so if you create sessions of varying times it will only compute the SD's of fractals inside those times.

You can. You just need to set up a bespoke Trading Hours template in NinjaTrader.

--------------------------------------------------------
- Trade what you see. Invest in what you believe -
--------------------------------------------------------

I wanted to know if you had a Sierra Chart version of the indicator available?
I have seen you mention the Fractal Swing indicator in this thread. However, I can't find anything for this indicator in Sierra Charts.

Thanks mate!

Cheers,
Dylan

The following user says Thank You to DylanSmoo for this post: