Clemmo's Trading Odyssey

3 Nights in Paris

Based on this post by [redacted for privacy] (Parisboy)



So it’s fairly simple. Using his bands (or yours - any will do, likely) simply find setups where

1. 3 bands are pointing up or down, united (but wait for price to be on the non-trend side of the third band - buy the dip)
2. Price has crossed your (preferably Hurst-valid, but...who cares?) trendline
3. SL at last relative extrema as advised
4. Then enter.
5. Target ---> The same distance as price moved to cross the trendline - again, as per Hurst.

How important is it that the innermost or outermost bands are turning? If you wait for a droop in the innermost you’ll be past the trendline break and if you wait for a droop in the outermost you’ll be way too late, and vice versa for longs.

Trendlines require a point in a wyckoff range maybe 2019-11-25 07.22.31

More questions:
Is it better to use the band crossing points as supports for trendline ‘bridges’ rather than actual peaks (my research suggests yes). (interesting)

Peaks may be better associated with other (higher) TF not represented by the current band. Then again you often get earlier entries with steeper lines (associated with hinging on peaks).

Attached Image (click to enlarge)

Even if I have got some details wrong, (missing the FLD for example) this is a far more sensible way to trade than anything I've shared yet, and the only reason I share it openly is because the maestro has already given us everything we need at his thread which you should be reading instead of this rambling by a journeyman.

Extrematron

You likely don't recognize the name but I turned this into an EA and was testing it for a long while. My thoughts after the system rules.

1. Relative Extrema (RE) are valid only when they have a certain number of bars before and after the break. Some significance test of the depth of the peaks/troughs might also be needed but so far on higher TF I haven’t needed one. The number of bars varies per TF and instrument. So far on XAUUSD I’m using 3-MN, 5-W1,
2. Set a pending entry (or alert) at the level where a break of a relative extrema (RE) + protrusion zone (PZ) distance would occur. The distance of the PZ will vary based on the instrument and TF. In my first test on XAUUSD, I’m going to use 24 pips for MN, 20 for W1 and expect the distance to decrease with the TF. This distance should be adjusted for volatility, possibly using ATR.
3. Set a SL at the end of the PZ opposite the pending entry
4. After entry, if price CCCs the entry line, make the entry a phoenix (break-even) line and re-enter when the phoenix line is re-crossed. You can hardly ‘overdo’ this so just be diligent/careful or automated.
5. Exit profitable trades when an adverse PZ break occurs - so that would be a floor break for longs, a ceiling break for shorts.
6. Set a break-even+n pips after price moves a distance = the range of the bar broken by the RE level. In other words when price moves past 100% of the broken bar range, set the break-even level. The n pips depend on the TF so that these (failed breakout) trades are at least cost-free. They should not be ambitious though as we are after big trends on each TF.
7. Add-on new trades with every new RE+PZ break until you receive a ‘brush-off’ (close call). See below. Add-on trades should have their weight decreased each time.
8. If the floor/ceiling breaks but does NOT reach or exceed the far side of the PZ I call this a ‘brush-off’ and it signals that add-on trades should be paused (due to potential trend weakness) until … reset by an opposite floor/ceiling break?
9. Trade volume varies with TF and with trend maturity. Every add-on should be lower weight than its preceding trade if possible. The weight per TF needs to be tested. Right now I’m using 0.01 for MN, 0.02 for W1, 0.04, for D1, and so on. The weight should be smaller for longer term trades that might go significantly underwater for prolonged periods.
10. Pending orders should be set at every RE+PZ on all TF, or as low as possible/practicable. Trading multiple instruments manually is likely too difficult. An EA should excel with the clearly defined rules set forth.
11. So far optional: peaks/troughs with elbow (inside path) depths less than M points are not significant and are not REs
12. Loss recovery rule - in some bad ranges you will get multiple failures of the breakout that traverse the height of the PZ and endure for long periods. If you get a profitable trade that meets the 100% of the broken bar (signal bar) distance (or the broken bar is itself enough to recoup losses), exit immediately to recover your losses. The problem is if price continues what is the re-entry procedure? Wait for another RE+PZ break I suppose. For now the original RE+PZ level remains active until ceiling/floor break.

Original notes/ idea: A ‘dollar-cost-averaging’ idea but you add-on not based on time, but on RE significance and INVERSE to the size/length of the prior/current trend. So if the trend is young, and you just got on, trade heavier, at the risk of false breakouts, and if the trend is old, mature, trade lighter in anticipation of reversals.

There should be a cutoff/minimum elapsed amplitude before countertrend trades can be entertained on either side. Similarly, there should be a cutoff where with-trend trades can no longer be added for fear of reversals.

Aspects of underhill/overhill and Law of the Jungle.

Fun to test this on all TF, using big and small trades. Even better to automate this and test that.

Right, so I did (automate it). Now, keep in mind my EA is not completely true to the spirit of this system. I particularly had trouble with rules to define the 'significance' of REs (swings/fractals) and that may be the linchpin right there, but you can see from the link I posted that this EA was MASSIVELY profitable for the return/equity. So what's the problem? It was also massively NOTICEABLE and as soon as my account was 50-100% in profit after a certain time (<3 months) It would just get assaulted. The moment I entered a trade, it would go immediately to the SL. Do not pass go, do not collect $200. It was uncanny. Of course, it might have been bad luck. More testing required, as I say. However in my opinion, this isn't a system that can fail, it can only be assassinated. Look what they did to my beautiful boy.

Early days, going strong.

Nothing can stop us now.

Oh shit!

Never say die.

We're gonna make it after all.

Ah screw this.

That's when I realized, there's no sense in playing anything that can be 'worked on' over more than a day or two. 'Let your profits run' is for position traders and suckers. Also, defence is just as important as offence, in the dirty-fighting underworld of retail FX.

Triple Swap

Just place huge trades at quiet times of the day before triple-swap day and make interest on positive swap trades. Possible??

I haven't had the guts to try this. Anyone?

Build It Up

Like Tear It Down but in reverse. We’ll follow the momentum of the big candle instead of expecting an immediate reverse, or if it does reverse to half, we don’t mind. In fact, we’ll expect it and add-on on a retrace.

Ask for 100 50% of the signal candle (or up to the band edge) and prefer entries that take place in the far side of the band.
So for a long trade, prefer bull candles in the lower half of the bands or lower, and for short trades perfer bear candles in the upper half of the bands or higher.
SL just beyond the open of the momentum candle, after all, if the momentum suddenly reverses that’s a confused market - better suited to TearItDown (or MicroMacro).

Trade TID and BIU together with TIU at ½ volume? Very interesting.

Formal description

Whenever a tall candle (defined as a candle that is n pips tall OR n% taller than x candles looking back) forms enter immediately in the direction of the signal candle move. Expect a retrace at 50% of the signal candle height so have a virtual pending order at that level also.

Take-profit at a fixed profit of y pips or y% of the signal candle.

Stop-loss at a fixed points value or a multiple of the signal candle height.

Optional: break-even after price reaches ½ way to target or Z pips.

What gives me pause -just looking at a chart of highlighted tall bars doesn’t fill me with any confidence. Random at best. The reason that MicroMacro is doing poorly at the moment is because prices are trending. It’s more important to distinguish between trending/non-trending markets than to try and find a momentum play. (emphasis added by 2023 me)

Extrema Trading

This is the first step in a tripartite trading system (this became 'Truth, Beauty, Justice')

Free trade simply by placing pending orders at every Relative Extrema and use phoenix lines to control adverse moves.
This could use a script to automatically close every trade that reaches a certain profit in $ or pips. Or it could simply move the SL to BE+n at that level.
It would also be handy to be able to hide the levels from the broker until the last minute.
Also consider integrating something like MadDash so that you pick the fastest moving pairs or the most likely to jump (based on inactivity), or the most likely to retrace.

I described this system to Gharti and he made an EA from it. Check it out here.
https://www.forexfactory.com/showthr...7#post12737407

Also if you're wondering what I mean by 'relative extrema' in all these docs, here is an explanation.

Formal description
Definitions:
Relative Extrema (RE)
Sometimes called a ‘swing’ as in a swing high or a swing low.
The term is borrowed from mathematics, specifically from calculus.

Attached Image (click to enlarge)

Fig. 1 relative extrema example in linear functions

Attached Image (click to enlarge)

Fig. 2 relative extrema in a bar chart

Any bar sandwiched on either side by bars that are lower or higher is an RE.
Simple, yes?

Here’s another definition.

The RE has to look like a reversal. Meaning, if it is a peak it must (at a minimum) have one brother on the left that is lower, and one on the right that is lower, or if it is a trough, one sister on the left that is higher, and one sister on the right that is higher. Peaks and troughs with more bars of 'separation' between them and the current bar are more significant. We can attach a numerical value to significance by the number of bars that have elapsed since their RE level has been broken. More bars = more significant.

That is how we will define the reliability of any RE level breaks - by the number of bars preceding it and the prior RE as well as the depth of the ‘lacuna’ the empty space between the peaks/troughs. Significant peaks and troughs look ‘sharper’ than insignificant ones.

3 factors that make up significance

1. Number of bars since the level was last broken (if at all)
2. Overall amplitude
3. Depth - number of consecutive bars on either side of the RE with few or no REs (retraces) in the runs. The smoother the peaks/troughs the more significant they appear.

Now place a pending entry at the RE above current price and one at the RE below current price. When one level gets hit, ride it to your arbitrary stop. If price reverses and hits the other RE level break enter 2x heavier and ride it to the arbitrary profit target which should be enough to erase the loss of prior fakeout. If price reverses again and hits the opposite RE (a double fakeout) assume you are in a terrible range and enter a single trade in the new direction. That should bring the total weight to 2:2 on either side and then wait for clarity.

Clarity == a new RE level break occurence. Once that happens close two trades that should be of approximately equal value (breaking even) leaving you with a winner and a slightly larger loser. Enter once in the new (clear) direction leaving you weighted 2:1 in the probable trend direction and remove the losing trade as soon as you can, ride the remaining winner to the TP target.

Fig. 3 A somewhat inexact attempt to quantify the importance of various REs.

There should be a setting that controls the number of bars to lookback and find the tallest peak RE or the lowest trough RE. This should allow us to select the ‘extreme-ness’ of the relative extrema worth trading.

Level Breaks
Whenever price exceeds a prior RE (peak or trough) we can consider the level that was at the location of that RE ‘broken’.

This is basically an adaptation of Laurent Bernut’s ‘floor/ceiling’ idea. https://www.quora.com/How-does-Laure...r-ceiling-work

“We will enter Short on a collapsing ceiling. We will continue to trade Short until a rising floor is detected. We will close all our open Short positions and enter Long.”
Fig. 4 level break examples

Attached Image (click to enlarge)

A level break is what happens when price crosses past the prior RE level and then crosses to the other side of the protrusion zone, initiating a trade entry.

Wait, what’s the protrusion zone?

We need price to break the level but not to bounce off that level and reverse. This can often happen as these levels are considered ‘support/resistance’ by price action traders. The solution is to require that price not just break the level but go past it somewhat. The amount that we need price to exceed the past level is called the protrusion zone. The zone can be illustrated by a rectangle where one side of the rectangle is the level and the far side is the actual trade entry point.

Attached Image (click to enlarge)

Fig. 5 illustration of protrusion zones and entry level

1. Enter on every level break within the selected significance threshold.
2. Take a fixed profit at an arbitrary distance from trade origin. (or set to trailing once target is reached)
3. Set a fixed stop at an arbitrary (likely smaller) distance from the trade origin.

If the profit target is sufficiently large you may wish to set a break-even rule, for example

1. Move break-even to BE+1 once price reaches ½ way to to the TP goal.

More (original?) notes on significance (note that as of June 2020 I’m fairly skeptical about the ‘levels’ formed by any peaks/troughs or their breaks. (Everything is more random than you think.)

“Peaks/troughs need to be defined by amplitude (% move) and also by separation of distance (time) otherwise they’re part of the trend or noise.

A wild spike that lasted only a few minutes would be of concern as long as you ignored it. The pain comes from time under pressure.”

Commander Extrema


Trade Extrema Trading but in baskets (diversification)

1. Mechanized (not calculated) (not entirely sure what I mean by this)
2. Low ambition/low risk
3. Un-banded (unbound and blind)

Offset losses with wins?

It is not clear from what you posted here, but it appears to me that your plan is to catch the reversals. In the Extreme TMA thread, I used a non-repainting TMA indicator. I found that trying to catch the reversals too often results in entering before the actual reversal is underway. That is, there are still some more retracement and continuation swings in the current trend that run the trade into drawdown or even SL.

I traded INSIDE OF the slowest (largest) band that had an obvious direction with the middle band in agreement, and the fastest band made a retracement, with an entry on the subsequent return to the trend by the price. When any of the 3 lines that comprise the middle band changes direction, that is the warning to exit any open trades and stand aside until the direction is re-established: Either the middle band will return to the slowest band's direction, or the slowest band will change to match the middle band's direction.

The midline of the fastest band crossing the midline of the slowest band was also "exit now!" condition, and an alert that the slowest band is likely to change direction soon.

I also would often look at a smaller TF that had its slowest band match the fastest band of the larger TF, enabling trading those price bars of the largest TF that did not move far enough on retracements to take the fastest band into an obvious direction change to show a good retracement. That is, I used the fractal-like nature of the time series and carried out the same system on any TF.

I added some lines to your image.
The red and green verticals are where the slowest band has changed direction.
The orange dashed verticals are where the middle band has changed direction.
The yellow trend lines show the fastest band and price have made a lower top.
The horizontal green and red indicate retracements to consider for trend continuation entry.

ObviTrader



Who would have thought that OnBalanceVolume would turn out to be useful? Not me, that’s for sure, until I tested it out. However, how much of this is owed to the merits of OBV and how much to the benefits of the envelope is an open question. Not anymore. I now think it's not predictive of much at all. Unreliable, like an oscillator.

Chart tested: XAUUSD however should be fine for anything, just modify the envelope width to match volume volatility.

TF tested: M5 but should be interesting on H1 or D1. Test away.

Indicators:
OnBalanceVolume - yes a default MetaQuotes indicator! In fact so is the other one
Envelopes (x2 - one on chart window and one on the OBV subWindow)

Envelope settings (indicator window): - Period 40; shift: 0; MA: Smoothed; Apply to Previous Indicator’s Data; Deviation: 2.5% (adjust to fit width of volume volatility stream)

Envelope settings (chart (price) window): Period: 40; MA:Smoothed; Apply to: Close; Deviation 0.25%

OBV settings: Apply to Median Price (HL/2)
Fixed min/max:unchecked

Rules:

1. Enter when OBV pierces the far side of the OBV envelope
2. Exit on the opposing piercing. All channel re-entries are ignored except as a re-entry signal whose intricacies I haven’t worked out yet but imagine they’d be very similar to Juggernaut.
3. Don’t let wins become losses and use phoenix re-entries
4. SL is the adverse side of the price envelope
5. Set a loss-limiting SL AFTER price clears the first swing high or low. Set to BE +5% or so.
6. Adjust SL at an arbitrary level - maybe after first contact with the favorable envelope edge - to 50% or so.

Attached Image (click to enlarge)

Attached Image (click to enlarge)

Attached Image (click to enlarge)

MicroMacroCosm

Building on Tear It Down 2 (irony?) and refining it

Principles:

1. Tall bars are likely to retrace unless they are part of a trend.
2. Trends begin in between ranges
3. Bands show the point where ranges end (range limits)

Indicators

1. Spikebars - OR highlight signal bars
2. Bands (TMA/Quantile/Donchian channels/VMA)

The bands should enclose price ‘like a sausage’.
Spikebars set to the usual - 10-15 pips on M15, 15-20 on H1.
In addition we need a larger band to enclose the inner band and form a ‘top zone’ and a ‘bottom zone’. The inner band area is the ‘neutral zone’.

As usual when a signal bar forms we expect it to retrace to at least the midpoint of that bar. We will still wait for a % of time to elapse and we are still targeting the ~50% height of the signal bar for the TP.
Entry:

1. When the open & close price of a signal bar is entirely within the neutral zone we’ll take reversion trades both long and short. No need for SL or time-based SL. Reversion trades will happen ONLY in the neutral zone.
2. When the open or close price of a signal bar is inside the top zone, we will take only short entries (retraces from a bullish bar). These trades (trend trades) will be held until price reaches the midpoint of the band and then use a BE stop and trail down.
3. When the open or close price of a signal bar is inside the bottom zone, we will take only long entries (retraces from a bearish bar). These trades (trend trades) will be held until price reaches the midpoint of the band and then use a BE stop and trail up.
4. If price leaves the top zone going up, or the bottom zone going down we will close any unprofitable reversion trades, and WAIT...
5. ...until price returns to the neutral zone and flashes a new signal.

The problem is that this still will leave occasional stragglers that will decimate the profit curve. More important than ranging/trending is the (newly coined) concept of ‘price stream density’.

We can measure PSD by the number of times that price returns to the midpoint of a band within a certain period of time. If several days go by on H1 without a mid-point band line touch then price is ‘rare’ or ‘sparse’. You can see on this chart that price was behaving in a way that’s profitable for mean-reverting Tear-It-Down type systems from about Nov-2018 to about Jan-2018 and then it gets less dense and begins to trend up and down between the quantile bands. Even though there were only a few non-prodigals they are enough to make the system unprofitable (if not loss-causing) during that time. Price doesn’t get ‘dense’ again until this September. Without density wait times for reversion increase and generally this system needs it.

There are pockets of density around the edges of the bands - but this seems like a risky time to trade. We want price to be generally stationary, which means flat bands if possible. Quantile bands are good for this as they tend to flatten to show ranging periods.

The red lines are non-prodigals. The ATR doesn’t tell us much of anything useful that I can tell.
Bars that are taller than average, about double, also tend to be non-prodigals, but I haven’t done a thorough check of the probability. Also, I don’t know the average retrace time expected for ‘average’ bars. However it’s less than 2 weeks for ~90% of all bars - during this period. It should be better for ranging instruments, worse for trending.

Also, it doesn’t seem to pay to make short trades at the bottom of the range, and long trades near the top, as that’s precisely when the non-prodigals will kill you the most.

Generally cyclical price is less dense and ergodic price is more dense.

Update 2020
Remove trend mode and replace with a rule

1. When a trade is > n$in drawdown (or n points away from the TP target) place an opposing trade with x volume

Update 2023: became microMacro EA; which works well until it doesn’t. Like all mean-reversion ideas.

And now my testing pile is looking a bit neater. Looks like the pound isn't going to do anything today.

That is quite interesting; thanks for posting. When is the 'slow' band too slow for practical purposes, do you thinK?

I'm not sure what you mean by "slow". Do you mean what period?

Yes, that's what I mean. What's the largest period to use for the outer band?

Got started late today and couldn't take full advantage. However I still need a way to know when this is higher probability. Unsurprisingly, I think a band will help.

The period for any of the the bands is a matter of personal preference such that they provide a view that is in some way meaningful to the specific trader. I would suggest that whatever period is decided for the fastest band, try a multiplier for the next band of anything between 1.5X and 6X. At 6X, that would give the outer band a period 36X the period of the smallest band, which already a bit extreme for my use. But I also find an MA200 too extreme, so again, it is up to each specific trader to find periods and how many instances of the bands are useful for him.

Continue with the same multiplier for the next band, so that exactly the same ratio is maintained between any 2 adjacent bands. I often had 4 bands displayed on a chart. So in my case, band2 period was nX the period of band1, band3 nX band2, and band4 nX band3.

BTCUSD W1

The initial promise is now looking like a low risk short.

When prices are moving slowly like they are today I feel like a sitting duck. That's partly the function of a distribution range - waiting to see how new flows align themselves and then finally moving in the direction of least resistance. So I wonder if I should sit tight or get out and wait for an opening day when things have a reason to move. Probably that one.

Haste is never more dangerous than when you feel that victory is in your grasp. - Eugene Znosko-Borovsky

I'm not saying it's going to happen, because as the recession deepens, the Brit economy will be harder hit than the US, but this recent double top would make for a delicious bear trap. One little zag and then grind upwards. The USD index has plenty of room to fall back to 100.00. As Britain's news gets worse and worse, shorts enter and are left scratching their heads as to why GBPUSD keeps rising. The market absolutely loves this sort of thing.

Also, compare these two double tops and it should be quite obvious how they are significantly different with respect to the total range traversed.

On the other hand, the market has been gunning for parity. Then everyone will think that's as low as it can possibly go. Then it will go lower.

Time to check in on the FF leaderboard.
Well done goldeyes! Still hanging in there. I see no one else from the start of the year?

Nope. All new faces.

So what's Goldeyes' secret?

Don't trade. Leave any losing trades open for months at a time. LOL.
This would be the same advice I give to most FF denizens. Don't trade. Fortunately for me, this excellent advice will be roundly ignored.

I found a screenshot I had taken that included the FF trades page on the other side of the screen. This was taken 2022-12-19, closer to the end of the year. Let's see if any of these contenders pop back on the radar in the coming months .