Clemmo's Trading Odyssey

I mean that price tends (averages) to revert nearly 75% of the way from a prior trend's measured distance, tip to tail, before a trend carries on. This figure has increased over time, as it gets trickier to shake out cynical retail and algo traders. The other percentages are common 'Fibonacci expansion' percentages that are used by technical analysts.

I agree that most moving averages are close approximations of each other, if that's what you mean? What part of the JoGET code reminds you of the HMA calculation?

3 Amigos

Another idea based on something from an ICT video.

Timeframe: D1 only
Generally speaking I don’t put much stock in the information that can be obtained from bars or candlesticks. However banks (or big/smart money/or just everyone?) use this pattern very reliably to determine their daily bias. (Says who??? Is this a direct quote of something Huddleston said? Why am I repeating something I have no idea about? Terrible.) This is strong enough to knock cycles entirely out of whack. (in your opinion) It is therefore a primary consideration above even cyclical considerations. (is it reallly???)

Look for three bar patterns examining high and low prices only (wicks).

This is a bearish pattern. The middle bar is flanked by two lower highs. I suppose another way to put this is that price was making higher highs up until it didn’t. There is a strong (how strong - measure - I did and there's no sign of any such propensity, just another Huddleston hustle) probability that the following daily bar will be bearish and that its high will not exceed the high of that third bar, and that its low will exceed the low of that third bar. Trade only that ensuing daily bar, and set the stop at the high of that third bar. Once price declines by 20(?) pips below the low of the third bar, move the stop to secure the profit and stalk. After the day ends, remain in the trade only if the stop doesn’t get hit and price continues to move in your favour.

Vice versa for bullish.


Position size - use 80% Kelly (no way José! You don't even know your win rate!) divided among your basket (if possible) of trades. The main disadvantage of this system is that it’s slow-ish and the number of opportunities is limited (? -check). (Not limited at all. These 3 bar patterns are as common and as valuable as dirt.) Should not be a difficult EA. (oh, you'd be surprised)

Entry - enter on the open or use a cycle to help you refine it? Enter between 2-4am before New York? Needs some study. (If you are going to be so precious about precise entry then there is a more sensible place to try to enter than after a 3 bar swing -namely top or bottom of a range)

Test out the various 3 bar indicators. (spoiler alert - they are all terrible; not just because this has no predictive value but because the coding standard for free indicators is fairly low) Then do a step-by-step.

Future - now can we add cycle knowledge to this to make it better or are we just reducing the robustness of something very simple? (spoiler alert - you can't either make this better with cycles or reduce the robustness of something that has no predictive value)

Huddleston gives us a more stringent signal condition. Although my examples above (coincidentally) also meet his standard. He gives this example.

Middle bar is flanked by two lower highs, therefore expect bearish future direction. However in addition, the third bar extends below the low made by the second bar, AND closes below the low of the second bar.


Testing notes:
Use the more stringent conditions to improve win rate - see notes on Huddleston where we require the third bar to close below/above the low/high of the second bar AND the third bar must break through the lowest low or the highest high of the second bar’s body.

Stringent condition sample test - success over 80% but only 21 trades year to date.
https://drive.google.com/open?id=0B7...zAyczd4cTRJTk0 (where is this file?!? Is this my fault or Google's?)

Less stringent conditions provide success of 68% and more than double the number of trades. Blue and yellow = win, red = loss.
More testing required. Also test lower timeframes - eg. H4

It seems that up trades have higher probability of success in up phases and vice versa, not surprisingly. (and this already tells us what we need to know - the trend direction; one thing to try is using this idea but only for re-entry points) Can we improve the win rate of the less restrictive swing condition using cycles on D1 or lower? Initial check isn’t that encouraging.It is more significant that price is in a trend or a range. This system wins in ranges, even though we’re only trading one bar of the trend. So, probably simpleton will work better if you don’t use the restrictive definition of a swing high/low. Nah.

Testing
Testing on USDJPY in 2017.

As stated a valid signal is generated when the middle bar is flanked by two lower/higher bars and the third bar breaks below/above the level reached by the body of the bar and closes below/above that level.

There were 21 signals of this type generated on D1 by USDJPY in 2017, of which 17 ‘continued’ for a success rate over 80%. Can we expect sufficient trading opportunities using this system if we trade multiple instruments?

If we make the conditions slightly less stringent by allowing signals whose third bar doesn’t close above or below the second bar’s high/low, and allowing third bars that don’t break through the high/low of the second bar, we get 56 signals, of which 38 are winners, or a success rate of 68%. Still pretty good. Nearly good enough for a trading strategy on its own.

For something that is so simple and effective I am surprised that this isn’t more widely known/repeated. I became a subscriber to the channel based on this tip.(It was just good luck; keep on reading!)

Further checking - XAUUSD 2017 - 30 ‘strict’ signals generated. 24 wins. 6 losses. 80%
Less strict (don’t enforce third bar breakthrough or close below second bar extreme) generates 50 trades with 32 winners, 64%.

Testing H4 on XAUUSD using stringent condition
From August to now
20/27 trades = 74%

Testing EURUSD W1 using stringent
9/16 wins from Sep. 2015 to now = 56%
Often much of the range is done by the time the week is over and so you’re getting into the retracement during the following week.

Testing EURUSD D1 from Jan 2017 to now
Stringent
17/29 = 58% ! Why so low?
Remove all signals with third candles crossing the (wrong) band?
15/23 = 65%, still not stellar but better.

Need reliable dashboard indicator.

I just don’t know if we can get reliable indicators from a single candlestick or candlestick pattern. Look at EURUSD - 65% win rate isn’t sufficient for daily trading, or is it? (It is sufficient, but the real rate is not that high. Let's do a new test using data from this year.)

New screenshot of 3 day swing (stringent) test - you can see adding ATR breakout signals doesn’t help accuracy either.


I’m ready to abandon this one, despite the frequency that the universe keeps bringing up ‘3 amigos’. :/
It's funny how things like that happen. Clustering is a property of randomness.

3 Amigos updated test
One of the things I remember is how difficult it is to find a 3-bar-swing indicator that actually works properly. Whatever it is about it, most of the ones you can find either flag incorrect patterns or miss clearly valid ones. I adapted by doing it manually (slower) or adding several indicators together. That's why you see groups of different arrows on this chart.

This doesn't solve the problem entirely though, as you can see in this shot the purple square highlights an arrow that is on an invalid pattern.

The rule I'm using is:
"A middle bar that is flanked by 2 lower or higher bars, where the third bar breaks the high or low of the central bar." The highlighted pattern has a third bar whose high is lower than the middle bar high, so it's invalid.
This is a little more stringent than just looking for V or ^ patterns, but I don't require the third bar to close above or below the high or low of the middle bar.

The rules are simply:

1. Look for 3 bar swings on D1 that match the above pattern
2. Enter at open of bar 4.
3. Exit at the end of that bar or imagine we set a SL at BE when that bar closes.
4. SL at the low or high of the middle bar (bar2). If the SL isn't hit but we don't have a profitable exit on bar 4, then exit at the first profitable daily close (or imagine we set SL at BE).

Now let's count.

I said this back in 2019.
"I made this system when I was relatively new to trading. This system is susceptible to TREND BIAS and the market in general is a pattern-destroying machine, so that I don’t favour trading with patterns of any kind. I don’t trade this and I don’t necessarily recommend it, although there might be some ideas you will find useful in your own systems."

Except after checking EURUSD 2021-2022 I find there were 49 losses vs. 89 wins on daily charts. It seems like the losses are slightly worse, probably because there's no trailing for wins, no minimum target; exits are more important than entries, etc.

What's changed? Well, for one thing Fibre is now strongly trending, both up and down. I need to do a test on a ranging bit of history. Almost by definition in that case the signals should be reversed.

Is it because I'm not using Huddleston's particular rule? Flipping Huddleston. He's cost me nearly as much time chasing rabbits down rabbitholes as Hurst.
This probably breaks even or just slightly worse. I'm considering,

1. Limiting losses to the first unprofitable day
2. Letting wins run for an additional day
3. Using fixed TP and SL at a favorable ratio
4. Sticking to Weekly charts?
5. Using the signals only as entry points in a trend-biased method
6. Using bands. Once again.

This probably breaks even, at best, no matter what you do. The pattern is the least important factor. Right? Right??!?

Ah crap, add it to the testing pile.

just for you to know Clemmo17

I willl suppress this message under 24 hours to keep your thread clean

Simpleton

Ok, now this. This I KNOW doesn't work. I have, however, sadly lost the testing documentation.

Yeah. Works in a trend. As you'd expect. Trades on the trend side, trend instead of forming peaks. Troughs are short-lived.

Timeframe: D1 only

Simply enter long after a 3 candle swing low, and enter short after a 3 candle swing high. Ignore future signals that are the same or use them to pyramid.

Stop placed just under the previous candle extreme?

See 3 Amigos for a visual description of a 3 bar swing.

Use 80% Kelly to size positions, and divide this amount around a basket of uncorrelated instruments.

Smart trailing could make this system great. Say, allow no retrace greater than 50% of the gain to date? Or use a band target? Above all don’t allow retrace once price has extended above the previous bar’s high/low. Allow no price to close lower/higher than prior low/high? Test profitability of different approaches during same time periods.

This strategy is good for a first EA programming exercise.


Future:

1. Do you get better results on H4? Lower? (of course not)
2. Take a cue from something like ZigZag and only consider a new signal within a certain number of bars? To avoid choppiness.
3. Ignore dojis? (small range, open/close prices nearly the same) (yes, since by definition it can only work in a trend)
4. Ignore inside bars?
5. Ignore gaps?
6. Early testing shows there are too many signals, too many bad results. What’s odd is how no indicator I can find finds all the signals.
7. Performs poorly in ranges. Can we avoid ranges? (avoid summers?) (ranges happen in the winter too, dude)
8. What if we use candle bodies only? (turn off wicks) (or how about Heiken Aishi. No I'm not serious. OR AM I??)

I think you are saying, using bands is the superior method? I am not very clever, but I think I agree. Of course, I knew this a long time ago, but I resist the easy path, sometimes, foolishly. I have to know the dark side!

FLD 64 (Future Line of Demarcation) Aas a Trend Following Tool


FLD 64 = Price Action projected forward 1/2 Period (here 64/2 = 32)

Revolvr

Attached Image (click to enlarge)

This is one of the oldest systems whose main conclusions still forms some part of my current methodology. It is the beginning of the 'critical path'. Of course this system itself went nowhere.

Fig. 1 : A typical piece of price history that could have come from any instrument and timeframe

Consider just about any slice of price history on any TF. What does every candle have in common?
OHLC, yes. What else?
Shortly after open, and shortly before close there are wicks that extend on either side. Like so (highlighted blue):

Attached Image (click to enlarge)

Fig. 2 Price history with bar wicks highlighted

Rare is the bar that moves without retrace from close to the following close (although it does happen).
Now consider how in any sequence of 2 bars there are two probable situations

 

1. The bars will progress with the trend with a slight offset
   like so
   Attached Image

   
   (downtrend) or like so
   Attached Image

   
   (uptrend)
2. They will not make any trend progress but instead one bar will encompass the other to produce an inside and outside bar combination, like so
   Attached Image

   
   or like so
   Attached Image

   

What all of these combinations have in common is some degree of overlap between bars. It doesn’t matter the trend direction, strength or timeframe, we find this continually owing to the multifractal brownian (or Ornstein-Uhlenbeckian?) motion of price action.

We can (possibly) take advantage of it by considering in any given pair of bars how the overlap might occur. Even if we ignore the current trend there are only a few possible continuations.

Fig. 3 A nondescript price bar

Suppose a bar has just closed and it looks like this. The following bar will most often start very near the close price (assuming trading isn’t very thin like at the start or end of a week).

Attached Image

Fig. 4 Closed price bar with a new price bar right at open.

Right after open it would look something like this. Now two things can happen.

Attached Image (click to enlarge)

Fig. 5 The new price bar starts down but then reverses direction.

1. Price can move down and then reverse (or slow) the trend by returning to the prior open
   
   Attached Image (click to enlarge)

   
   
   Fig. 6 The new price bar starts by going up but then continues decisively down
   
2. Price can move up and then continue down with the trend

In either scenario there is a portion of the price bar that overlaps both sides of the opening. In other words, the price would reach that area/level no matter what, as long as it doesn’t simply move directly in one direction from the opening. As a result we can make a theoretical profit no matter what price does on just about every candle, provided it moves enough that we can beat the spread and commission costs.

There are multiple ways to play this. Some ways might not be practical because of the broker tendency to widen spreads at the start of a new bar (?) Not all TF and times of day would be practical either.

 

1. Right from the open place a trade with a TP the width of a typical wick in the current trend direction. Optional: SL the distance of a standard wick +safety margin away from the trend.
2. Right from the open place a trade with a TP the width of a typical wick against current trend (possibly near an expected band/reversal point). Optional: stop loss the distance of a standard wick+safety margin towards the trend
3. Place 2 trades right from the open with TPs the size of a standard wick one favoring the trend side with a more generous width. Optional SLs wick+safety buffer in either direction.
4. Wait for 20% of the bar time to elapse and then trade in the other direction assuming the retrace will have completed. Optional SL wick size+safety buffer
5. Wait for 20%-30% of the bar time to elapse and then trade in the direction that price has moved assuming that’s the trend direction, set TP for 60%-80% of a recent average bar height. Optional SL wick size + safety
   

In every case we use a TP rather than a trailing stop.
Then there are ways to play the close of each bar, now with added confidence that the bar’s main trend direction has been established.

 

1. Wait for 80% of bar time to elapse then trade against the bar’s trend assuming close price will retrace somewhat. Close trade after next bar open +20% time? Optional SL set at wick size + safety.
2. Wait for price to exceed the previous bar’s high/low and then trade in that direction, assuming that’s the trend; set TP to the height of the current bar x2. SL set just above/below current high/low. Additional certainty after a 3 bar swing high/low?
3. Use an indicator to give a bias towards likely trend direction and upcoming reversals and set the closing trade to trail or set TP to aim for a higher TF bar close.

Just be aware that any mean-reversion system can be torn asunder by a sudden violent trend, if it isn't ready to give up something.

Lion

A simplified (ha) law of the jungle?

The 'Law of the Jungle' systems are a whole category of systems that try to trade correctly on as many timeframes as possible. Needless to say, the horrible ambition, the breathtaking ignorance, are all earmarks of the novice.

Liger, liger, aiming right,
In the markets of the night,
What immortal hand or eye
Dare frame thy fearful symmetry?

The lion/tiger anticipates the actions of its prey and then pounces. This strategy uses indicators but reads between the lines rather than entering on the precise signals to find the best entry/exit.

On one hand this almost sounds like 'man in the window'; but I don't think I had the foggiest notion of it yet.

Indicators:

1. Hurst for MT4 or Sentient Trader (only to find min. 40 week peaks)

1. RSX of Hull Moving Average
   1. RSI: 14
   2. Price: close
   3. RSI type: RSX
   4. Avg period: 80
   5. Avg Method: Hull MA
   6. Other settings default, and set alerts as desired.

2. Averages Confidence Bands 2
   1. Avg length: 80
   2. Price: 0
   3. Avg type: 10 (hull)
   4. Confidence: 90
   5. Shift: 3
   6. Alerts as desired

Add bands if desired (I do)

1. TMADCG mladen-arrowsMod (band1)
   1. Half-length: 80
   2. price : 6
   3. Deviations: 2.5
   4. Alerts as desired

2. TMADCG mladen-arrowsMod (band2
   1. Half-length: 640
   2. price : 6
   3. Deviations: 3
   4. Alerts as desired

We are looking for fluctuations away from the trend, to either avoid them or trade them lighter. The rest of the time we want to be following the daily trend. We survey charts on the D1 timeframe, mostly.

Combining trend and counter-trend trading in one system is asking for trouble. If you want to do both; do them separately.

The 40 week peak trends are our main guides. Price rarely doesn’t jump to the whims of the 40 week peaks and troughs. These also have the added bonus of being fairly evident and even the novice Hurst trader will be able to identify them. You can even find them phasing manually and so you don’t need Sentient Trader or Hurst Cycles for MT4.

A bold statement; but there is some evidence to back it up, at least in some instruments.

RSX is our secondary guide. If it’s in tune with the 40 week trends we’re simply following it. When its trend gets long in the tooth, but the upcoming 40 week peaks or troughs are still far off, then we want to pay mind to what price is doing with the moving average line of the confidence bands indicator.

When price crosses that line on the D1 or threatens to do so, we want to drop down to H4 and find a finer entry/exit.

On H4 we can also drop down to H1 for even finer entries.

Place 3 trades at the right time. A D1 trade that we intend, as much as possible, to hold between 40 week peaks.
An H4 trade that we’ll hold as long as price stays on the correct side of the RSX or the correct side of the MA line on H4 if the RSX is long in the tooth.
An H1 trade that we’ll hold as long as price stays on the correct side of the RSX or the correct side of the MA line on H1 if the RSX is long in the tooth.

The thing to keep in mind is the RSX will change sides once it’s clear that the trend has changed. You should be aware that it’s due to change well before that. Once it has changed sides, price is bound to retrace, sometimes quite a bit, back to the edge of the confidence band, or the MA line or a bit more. That’s where you actually want to enter. Guided by the higher TF (D1 usually) we expect price to stop retracing and go back the way RSX is telling us its bound to go.

 

1. Stay on the right sides of the 40 week trends. Don’t fight them. A correct manual phasing is probably needed. The RSX can actually help show what is a 40 week peak/trough and what is lesser perturbation. Lesser points will not reach the full length of the indicator amplitude but retrace midway. However there will also be some full length cycles shown that aren’t valid 40 week markers. Measure the time manually and verify with a Hurst indicator and then do it the other way round to confirm. Getting the 40 week reversals right is vital to this strategy.
2. Which is why it can't work. Cycle phasing is voodoo.
3. Trade with the trend on D1
4. Trade countertrends (as revealed to you by RSX or MA line crosses on D1) on H4 and H1. We give countertrend moves shorter shrift and manage them on lower timeframes. Nooo!
5. Exit on RSX changes, MA line crosses and possibly, target band touches.
6. It’s important to anticipate rather than rely on RSX. When the trend is flatlining, drill down and look for price crosses over the MA line. These will often coincide with RSX changes on the lower timeframe or anticipate them just by a couple of bars. These are the best entry points for a new/countertrend trade. If you take the trade right at the RSX signal it’s often late and you will need to wait for the retrace to enter. Color changes on the confidence bands indicator are good confirmers, but they make poor (tardy) signals. Or are they in fact better because they avoid the initial retrace, making them more efficient (compounding) than MA cross signals? Compare and evaluate. If you can't tell, just abandon ship.
7. They are better than RSX signals on higher TF though, and are valid last resorts. Give more weight to the ones that occur right at the curve of the RSX, not when its flatlining.
8. If the trade is safely winning don’t exit on close of a MA line cross, but look for a second candle to move strongly away from the line. Often price will bounce off the MA shortly after closing over it and continue the trend. It’s best to give it a chance to do so even at the cost of some profit if it doesn’t.

When scanning:

1. Look for old trends on RSX D1. The line should flat, long-ish and ambling on the top or bottom of the indicator window. Price should be pushing out the inner band.
2. Look for MA line crosses.
3. Look for averages indicator signals that coincide with round corners/signal changes on RSX.
4. Look for band touches that coincide with round corners/signal changes on RSX
5. Look for overlapping occurrences of these things and for the same signals to line up (or threaten to change) on H1, H4, D1.

Entries:
Anticipating is better than waiting for confirmations, but then false breaks need to be closed mercilessly. This will result in some up-front costs but the eventual big trades more than make up for it.
Entry signal is usually the MA cross on a lower TF
A band touch might also be a good signal but it’s riskier or slower to win. ‘Edge band’ trade in the parlance of Hurst.
Exits:
MA cross that has follow-through (one more candle after priror candle close) or an RSX change change, but MA cross is usually earlier and better.

Preferred entry signals in order of preference:

1. MA line cross
2. RSX signal change on a lower TF
3. Target band bounce/touch

Make 3 trades so that you have one long-term trade, one medium-term, and one trade will satisfy your need for action, and pay for the possible harsh retracements that will occur on D1 timeframes.


Process:

 

1. Switch to D1 timeframe. Find the most recent 40 week peak and trough using ST or hand-phasing. Price is likely moving away from that point unless it was very recent, in which case you need to treat it cautiously especially if it’s early but even if it’s overdue.
2. Mark the D1 RSX signals that take place around the 40 week periods. Don’t mark false ones (mid-amplitude blips).
3. Draw lines to show the win/loss of expected trades between the peaks/troughs of the RSX signals. They should make sense. If they don’t - understand why. Did you make an error? Is the market very choppy?
4. If this trade is interesting we should be just after or before an RSX signal, price should have crossed MA or will soon, and have retraced back to the line or the confidence bands edges.
5. Drill down to H4. We should be able to say some of the same things -MA crosses, RSX changes, retraces to the bands, or nearly about to happen.
6. Drill down to H1. Find the finest entry and mark it with a pending order. If it has passed drill down deeper to find an upcoming one.
7. Add 2 more orders depending on how things unfold?

Your faith in your indicators will be your downfall.

Triger
I think this might be how DrDave trades?

Indicators: 3Hull moving averages

Averages confidence bands 9.0 x3
Periods: 30 (fast), 60 (slow), 120 (superslow or guide)
Price: close
Hull moving average
Double-smoothed: true
Confidence level: 99.9, 95, 90
Shift: 5,5, 16

Attached Image (click to enlarge)

Check performance on M1, H1, D1
Then, add this to trading Hurstonian 80 day cycles and you have Around the World in 80 days?
Trade all cycles and you have Law of the Jungle?

The trouble is tigers are fast and pouncy. Moving averages are slow and laggy.

Around the World

Triger + trading 80 day Hurst cycles in agreement with the 40 wk trends.

Use double-agent entries just prior to expected (big/20wk+) cycle changes. Otherwise trade with the previous 40 wk trend.

Use Triger to set exits and to gauge when something isn’t going according to plan.

This could easily work; I'm just too impatient. Also the 80 day cycle is ...as iffy if not more than the rest of them.

Around the World 2

A (batshit crazy) Hurst cycle trading strategy.

Phasing
We use my simplified phasing system and dispensing with FLDs, concerning ourselves only with VTLs.

 

1. The phasing system, briefly, consists of identifying the highest and lowest price points, assuming these represent probably, 18y peaks or troughs, 9y minimum if we’re looking at around 20 years of data. It doesn’t really matter since most people don’t have trading horizons this long, even monthly cycles are longer than we will likely trade.
2. Then we mark the rest of the yearly peaks/troughs using the nominal model expected locations and matching them as closely as possible with actual peaks/troughs that we can find in the price. Generally, we’ll find 18yr/9y/ and 54 month (4.5y) cycles.
3. Then we split the 54M periods into 18M sections of 3. Hurst thought the 54year cycle was 3 groups of 18y and the 54M is three groups of 18M; while this doesn’t always seem to be true it often works out.
4. Then we switch to W1 and do the 40w and 20w peak/troughs. These will generally be mid-way between each 18M section.
5. Then we switch to D1 and find the 80d and 40d peaks and troughs. Important note about this timeframe. Whereas in W1/MN we can just measure the exact distance using MT4’s measuring tool, when measuring in days because weekends aren’t represented on the timeline, but they matter to Hurst’s nominal model, we need to do some mental math to make sure that we’re counting the right number of days - don’t just go by what the measuring tool says.
6. Staying in D1 find the 20d peaks and troughs. Place them on D1 but change their visibility to H4.
7. Now switch to H1. Why did we skip H4? Because it’s hard to figure the distance in days on H4. H1 period separators are individual days and are easy to measure/count. Place the 10d and 5d peaks and troughs.
   

General notes: I usually only do as much recent history as is needed to make trading decisions in the present. This works out to about one visible screen’s worth of price data.

Colours: Use whatever makes sense. I use the colours I learned from the MT4 version of Hickson’s Hurst Cycles for MT4/Sentient Trader.
18 year: maroon
9 year: purple
54 month: orange
18 month: yellow
40 week: yellow-green
20 week: green
80 day: aqua/turquoise
40 day: blue
20 day: magenta
10 day: deep pink (yes, this is very close to magenta but it’s a habit now - be cautious)
5 day: plum
2 day: orchid
1 day: gold
5 hours: white or red (rarely used)

I don’t generally phase lower than 5 day periods but sometimes I’ll be interested in a 1 or 2 day VTL.

I place dashed lines for peaks; and broken lines for troughs.

Once the chart’s been phased, I find the valid trend lines for each view.

1. The main concern is that a pair of connected peaks or troughs must not have a higher degree cycle cutting through it. Otherwise it’s invalidated.
2. The trendline takes the degree of the lowest cycle degree that composes it. So for example an 18M and 40w pair of connected troughs will be a 40w bullish trendline.

A slightly modified nominal model - all numbers are averages only

18y ~15y
9y ~8yr
54M ~63M
18M ~ 24M (the most variable cycle of all?)
40w ~36w
20w ~18w
80d ~63d
40d ~34d
20d ~22d
10d ~10d
It’s important to remember that phasing is a rough guess, there is no regularity that can be depended upon for consistently long periods, and this is all about attempting to divide the probability space.

Not all cycles are equal

1. Of particular importance is the 40w cycle peaks and troughs. They tend to be market movers and it’s rare for a 40w marker not to show a long-term reversal except when it’s working against a >18M cycle and is upended by a 20w cycle.
2. 80d peaks and troughs have an unusual tendency to occur at the midpoint of larger cycles so they’re often much less significant (have less influence on price) than 40d peaks or troughs. Not always but often.
3. Cycles less than 40d can get pushed around by larger cycles so never consider them in isolation. Look at their parents. If they are working with the parent cycle expect them to have more impact. The bigger the cycle, the more impact should be expected. 5d cycles can often have no more impact than daily or bi-daily blips.

Variance and target dates and amplitudes
Whereas Hurst relied entirely on his nominal model and Hickson uses the average period of past occurrences, I am proposing a slightly different tactic. Make use of the amplitude and timing of past sections that match the type of cycles that bracket the period in which you’re trading.

So for example if you were trying to trade a 10d trough up to a 10d peak you’d consider that the higher degree marker after the 20d peak is an 18M trough, and the preceding peak behind the 10d trough was an 80d peak so you’d

1. Keep expectations moderate
2. Find a similar situation that occurred in the past, look for the previous 18M trough, look for an 80d peak that preceded it and see if you can find a 10d peak to 10d trough inside that segment.
3. Measure its slope and place a goldenrod line from lowest to highest point
4. Move it to current time and there you have an amplitude and time estimate
5. Now do the same thing with the variance and measure the deepest countermove in crimson and use that to gauge how patient you’ll need to be approx.
   LOL - this is like assuming that weather on July 01 will usually be the same every year. I mean, actually that's more reasonable than what I'm assuming here. If cycles were periodic and worked as Hurst expected, this would be quite reasonable. In reality, it's just wacko.
6. Since we’re trading a bullish move, look for swings high and then put the countermove gauge at the top of that swing and place a pending order at the bottom of the gauge line. Basically we’re saying that’s a likely place where price will retrace but the trend will still remain valid. Repeat this until we get too close to the end of the target line - be conservative and expect trends to end sooner than they did historically.
7. If the historical section you find was underwater most of the time or had a very shallow amplitude consider not trading this opportunity at the current time.

An alternative

1. Look at the triangle that’s forming from upward and downward sloping VTLs, and use the angle from past data to extrapolate a path to the other VTL
2. That’s your time and price target.
3. If there’s not enough data (because price is at an historical extrema), trade a smaller cycle.

Dealing with the unexpected
Trading expected peaks and troughs using the nominal model, mine or Hurst’s or using any other tool of Hurst’s has been disappointing so far.

I believe that Hurst gives us a rough outline, for multifractal price and time, but reality always has something slightly different in store. It only takes a small difference between what we expect and what actually happens to throw a small account into the gutter. Especially without some expectations for variance.

The work we’ve done so far hasn’t been a complete waste of time though, as it gives us our directional bias, a time expectation for trade completion, and a guideline for sizing trades, since trades bolstered by higher cycle degree parents should be more reliable. It gives us a criterion when our main guide gives us unexpected signals.

What should we use for the main guide?
When trading manually I would recommend using Jotunheimr. Whenever price crosses the Jotun you can decide if that’s a likely cycle peak or trough and whether this is a temporary condition (worth ignoring) or not. When in doubt, follow the smoothed MA, even if the Hurst phasing says what you are seeing is impossible. All kinds of reasons could account for it, not least a phasing error. “There’s always a bigger parent cycle.” - Qui Gon Jinn (infinite variance)

Meaning you can (almost) always find a reason why a peak or trough didn’t push price in the direction you were expecting. Maybe the cycle is just running long or short - sometimes it will be very far off, and in cases where it’s running short it can be hard to detect.

For this reason I recommend using double-agent entries just prior to expected cycle changes.

When trading with an EA I would use Around the World to establish when to change from long only/short only or long/short (in ranging periods).


Sample procedure 1.
Be sure to read Jotunheimr if you haven’t already. I would stick with H1 timeframe for this instead of M15.
The only other changes are to increase the channel ATR multiplier to 1.8 and reduce the period to 30.

Basically we’re trading Jotunheimr but with variations that make use of data from the Hurst chart.

 

1. Wait for a signal from the Jotun line (colour change) and enter as usual, as per Jotunheimr.
   However whenever price crosses the Jotun, check what’s indicated by the Hurst chart. If it’s an insignificant cycle change, (or no discernible reason) stay in the trade. If it coincides with a tradeable peak/trough then obey the Jotun. Enter early, before the line actually changes colour, if it’s supported by what you know from the Hurst chart.
2. Another difference is the method for adding on to trades. Whereas Jotun would add whenever a swing approached the Jotun line we’ll use a different method.
   1. Gauge the potential variance of the trade you’re trading/considering by checking a similar section of past price history as described above.
   2. Use that gauge to determine when price has retraced sufficiently to add-on to the trade.
   3. Wait for a swing in the current trade that shows price is retracing.
   4. Put the gauge at the swing and place a pending trade at the other end of the gauge line.
   5. Repeat this as often as permitted until you get close to the target. Err on the side of caution since an add-on trade is a sure loss if price has unexpectedly peaked/bottomed.

   
   

3. A further difference is in how positions are sized. Whenever you get an entry signal from Jotun determine the peak/trough that has formed or is already in progress.
   1. 5d/10d = 1 unit
   2. 20d = 2 units
   3. 40d/80d = 3 units
   4. 20w = 4 units
   5. 40w = 6 units
   6. 18M/54M = 8 units
   7. 9yr/18yr = 10 units

   
   

Don’t enter these units as a block but add-in to them gradually, especially on the higher degree cycles which may take some time to turn the corner.

1. Exits can be handled by Jotunheimr with the ATR channel arrow signals especially if you have some add-ons in play (just take one off). However you can also elect to take exits using the Hurst chart instead. Simply place a TP level just shy of the far end of the goldenrod line that you created using the method as described above. And put SL just below the most recent swing hi/lo. If you have a range of amplitudes found using past data you could elect to take profits at each successively more profitable level until you are at expected cycle peak/trough or you run out of active add-on trades. Even if you are trading using the Hurst targets it would behoove you to watch the ATR alerts for signals that price has peaked.

Attached Image (click to enlarge)

First test of this idea seemed to be sub-optimal - more testing required

As Parisboy says, life is hard for the cycles enthusiast! This is the sort of thing David Hickson would pay me to demonstrate on his site, if not for the fact that I recommend against using Sentient Trader.

Riskovore

Attached Image (click to enlarge)

Triger + Trading in and out of risk events.

Trade in the direction of the market’s prevailing sentiment (read news and go with the existing 40 wk trend when prudent) and exit before the news announcement or at least set stops and watch carefully.
Use Triger to gauge when sentiment has changed or it’s time to get out early? It might not even be needed or a use for it might become apparent as you test this.
The main difference between this and Around the World is that cycles are mostly not considered, and issue selection is done via the calendar.

Now this could totally work, if volatility comes back to the market during news releases.

Beetlejuice


Documentation not found! I can only find this in a reference within a different system.

Trading the nominal model cycles at consistent intervals guided by expectations for each level (family member) of a cycle lineage using the Averages Table. Demonstrates the power of the table of averages and the (custom) nominal model. Use of channels (?)

But as I said, the nominal model is shaky, the idea of a cycle lineage is therefore built upon quicksand. The averages table can be supplanted with bands.

Clemmo17 do not confuse the Nominal Model as concept and the rigid use of some hurstonians followers.

Hurst has used 3 different Nominal Models - one in his book Profit Magic (1970) and 2 in the course (1975)

Edward dewey had already one of 10 Cycles in 1944 then of 11 cycles in 1956

Bandersnatch

“Beware the Jabberwock, my son!
The jaws that bite, the claws that catch!
Beware the Jubjub bird, and shun
The frumious Bandersnatch!” -Dodgson/Carroll

3 target bands
Inner (green), outer (orange), stop (pink) -4 deviations.

M1 TF

Initiate short trades only in the top chamber of the outer band, long trades only in the bottom chamber. (because price tends to revert to the center, if you start a trade when it’s in the other chamber it’s statistically more likely to be half done its journey already) (true)

Enter when price is at or very near the opposite (to the direction you’re trading) edge of the inner band. So short trades enter when price is touching the top inner band side, long trades enter at the bottom side of the inner bands.

Best entries are where green band is piercing orange. Wait for the green band centerline to change direction, and/or just wait for price to re-enter the orange bands. Pyramid (add to position) when price touches the other side of the green band and then swings back (makes a swing high or low doing so). Always use a reduced size for add-on trades. If this isn’t possible because account is too small to use a larger initial size, then don’t pyramid.

Follow the direction of the inner band centerline to know whether to stay long or short. If you do that you get chopped up by twisting in the mid-band. Enter and exit only at double band piercings, use this principle for trading on any timeframe. Consider a band-to-band double piercing the basic unit of the initiatior (Mandelbrot). You’ll recall this is a trendline upon which the zigzags of market movement can be hung. Specifically you want to chart band to band double piercings where price makes its final re-entry back into the container band. If this makes the trading horizon too long, reduce the band heights.

Exit everything except the anchor trade at the orange bands.Exit anchor when the green band direction changes. Exit if price crosses the pink band the wrong way (trail the stop along the pink band).



I mean, this all seems very logical; why didn't it work? Well, for one thing the bands I'm using here are a shifty illusion. For another thing I was expecting fair treatment.

After a while I made these observations:

Observation: After a big move the bands remain distended and much good reversal activity is missed near the end of the day because nothing is reaching the band edges. Solution? We’d have to adjust the band period to a smaller value or use an arbitrary channel to make the trades. (or quit at the end of the day)

Observation2: Many of the loss trades would have won if the stop had been more generous. Try making the pink band equal to the orange band at 4 deviations. (but the size of your average loss will then just go up)

Question1: Is trailing the stops using this stalking technique more profitable than simply leaving the stop at BE and letting it gun for orange? (my research suggests not)

Question2: Should this system trade the trending moves? They seem obvious to me now that we see they occur when price aggressively refuses to come back from a band. (shades of upstairs/downstairs) (one or the other only)

Observation 3: Try as I might it is difficult to avoid trends. Monitoring on higher TF might be necessary no matter what.
(Just stick to certain daily times then.)