Linear Regression Bands II

BarroMetrics Views: Linear Regression Bands

In yesterday’s blog, I introduced the subject. In this blog, I’ll show how I draw the bands, and show its use in the recent S&P price action.



Figure 1 shows the March 2009 low. I have overlaid two Linear Regression Bands: one marking the 18-day swing  price action (red) and on marking the 5-d swing action (blue).

What anchor points do I use for the bands?

To answer that question, let’s look at the 5-day Linear Regression Band.

  1. I place Barros Swings of theTrader Timeframe (in this case, the 5-day swing, [blue lines; weekly trend]) and the First Higher Timeframe (in this case, the 18-day swing, [red lines; monthly trend]).
  2. I start the Bands from the most recent First Higher Timeframe extreme, the 18-day swing low on July 8 low.
  3. I end it on the most recent high that will define whether the S&P price action has accepted below the lower Band.
  4. NB: I do not include all the data – just the data from the Higher Timeframe Low to the most recent defining swing high. This will be at least a 5-day (i.e. a Trader’s Timeframe swing high)


Finally notice how well the Bands warned the trader of a possible rally.

On Oct 2, the ES formed a small range day below the bottom of the band. In my free weekly video/daily forum service I warned that a rally from these levels would probably mean a challenge of the 1075 (basis Dec) high. So far this is what has transpired.

The ES has hit the minimum target, the 78.6% of the Boundaries of Congestion (see Nature of Trends). In all probability we’ll see a challenge of the highs. The one dark cloud against this are the warning patterns of a possible 18-day swing Change in Trend in the US$. If that eventuates, we’ll probably see the ES tumble.

Linear Regression Bands

BarroMetrics Views: Linear Regression Bands

Firstly, Ana, thank you for taking over the past two days; I really appreciate your efforts.

Secondly, my apologies to those whose e-mails I have yet to reply to; rest assured I shall be attending to replies and should answer all by the weekend.

I have received a ton of mail following this week’s video regarding the use of Linear Regression Bands.  I’ll answer the queries in this blog.

One of the patterns I use for changes in trend is the Higher Timeframe Line Turn (see Nature of Trends).  However there were times when a higher time frame line change did not visually signal a change in trend in  the Trader Timeframe.  For this reason, I brought in the Linear Regression Bands, 3 stdev rule.

The rule says that a change in trend in the Trader Timeframe is signaled when:

  1. The Higher Timeframe line turns and
  2. We see acceptance beyond the Linear Regression Band extremes.

Sometimes, the application of the two conditions results in a later signal; but I have found that the use of the Bands leads to more reliable indications of the change in trend.


FIGURE 1 Crude Oil

Figure 1 shows Crude Oil: after the reaction low on July 29 2008. We note that:

  • The line turn price for the 13-week (quarterly trend) is coming in at 119.82.
  • The 18-day swing low is 85.42. Normally without one of the Lagging Change in Trend Patterns (see Nature of Trends) we need to see a breach of 85.42 before we can say the 18-day uptrend has ended.
  • But with the Higher Timeframe Line Change Rule,  we’d say the 18-day trend is deemed to have changed when the 13-week line turns at 119.82.
  • The new rule would require acceptance below the bottom ( 3 stdev)  of the Linear Regression Band.

Figure 2 shows  the equivalent of the 13-week line turns on Aug 4. At this stage, under the new Rule, we have a warning of a possible change in trend in the 18-day.


FIGURE 2 Crude Oil

With that warning, I would place longs on hold. Generally, once the Higher Timeframe line turns, we’ll see prices reach the appropriate extreme of the Band (in this case, the bottom band).


FIGURE 3 Crude Oil

Figure 3 shows that Crude bounced off the bottom Band. Longs would have been placed either on Aug 20 or Aug 21 and exited on Aug 22. (last bar in Fig 3).


FIGURE 4 Crude Oil

On Sept 5, with acceptance below the bottom Band and the Maximum Extension, the trend would have been deemed down and I’d have instituted shorts.  Note that had the bar I have labeled ‘Neutral Bar’ in Figure 4 been a bearish conviction bar, I would have deemed the trend down on Sept 4.

I’ll conclude this tomorrow.

Questions from Readers (2)

BarroMetrics Views: Questions from Readers (2)

This is not so much  a question from a reader; it’s a question from a student. But I thought the question and answer would be of interest to the readers here. I have observed that many newbies judge the correctness of a decision by the results of the decision. Since the trading is a probability game and therefore the same action can produce two very different results, such a decision making process is inherently flawed.

In the answer to the student, I have set  out what I believe is a better process. The question posed is in the attached Word document.

My answer:


Thanks for post. I’ll answer this on two levels:

  1. A generic answer applicable as a general principle
  2. An answer based on the specifics of your trade


Firstly, you need to set a policy BEFORE you take a trade. Otherwise, you will judge a trade by the results and that is not what you want to do – as Aussies say, you will be on a hiding to nothing.

The policy you set is based on your personality and values. Your trading needs to be a reflection of these values. This is why I ask students at the BarroMetrics Barros Swing course to create simple statements reflecting their trading philosophy. This trading philosophy will set the policy.

For example, I am by nature ‘risk adverse’. So, my first statement is: “Preservation of capital” first and foremost. This means I am only willing to take a trade if   my perception of a trade’s risk falls within my risk parameters – i.e. I am willing to miss a move if it does not. Taking this further….

….You will recall that as a matter of policy I don’t take breakout trades because they have a low win rate. Note however that in years where there are strong trending markets, breakout trading will enhance my bottom line. But in years where there are ‘whippy’ breakouts, my Expectancy Return drops dramatically. As a result, by excluding breakout trades (in most situations), I have a smoother equity curve in return for a lower net return based on my results since 1990.

Turning to real-time example……

In the chart attached, you have the GBP/JPY with an Upthrust Change in Trend Pattern. The pattern was triggered on Aug 11 09. However, since I like to place my stop above the Maximum Extension, an entry at 157.80 and a stop at 165.77 was not possible. My money management allowed for entry above 160.2. I chose 160.47.

The chart shows I missed my entry by 13 pips on Aug 13.

I did not chase the market. I won’t say I did not feel regret as the market headed south. But I also know that the miss is the price I pay for my the plan I have.

What about the current situation? The same policy applies:

The market has broken below B. I now have to wait for a WPC etc. Could the market:

Explode to the downside resulting in another missed trade? Or
After waiting for the WPC, zone, setup and entry bar, could I be stopped out?


But just as is possible, the WPC  (Whole Point Count) etc could form and I could wait for the zone, setup and trigger to enter.After my entry, the GBPJPY could then move strongly in my favour.

In other words we will never know what the future (right hand side of the chart) will bring. All we can do is formulate a set of rules and execute them as consistently as we can, given our state and temperament at the time of execution.

That’s the generic answer, let’s turn to the specifics of your trade.


Firstly let me say that I use a tool that you may not use: I keep a set of stats for impulse and corrective swings of the various timeframes.

In the chart you sent, the questions I’d ask myself are:
Statistically, has DP (DuPont) move greater than mean +1 (if it has, the theoretical probability of a bounce is 97.5%)? I may take the trade without a bounce on reduced size.
Statistically, has DP move greater than mean +2 (if it has, the theoretical probability of a bounce is 99.5%)? I don’t take a trade without a correction.

I would also note:

  • My entry is at the Death Zone. This increases the probability for a rally to the top of Value (67%).
  • DP was down for 5 consecutive days within a congestion zone. The consecutive moves may increase the probability of at least a 1-d rally.

Given the above, I would have waited for a rally to sell (looking for a zone and setup) knowing and accepting, that by doing this, missing a move is on the cards.

Hope this helps.


FIGURE 1 Student DuPont Chart





Comparing Analysis Approaches

BarroMetrics Views: Comparing Analysis Approaches

The current GBPUSD provides a clear illustration of the difference between traditional technical analysis and my approach.

Traditional technical  analysis tends to be a closed system. For example: if we see a H&S topping pattern with a breach of the neckline, we assume a change in trend from up to down. My understanding from the books I have read is that once the pattern appears, I lock in on the one scenario unless the stop is elected. This is the Richard Schabacker approach that was adopted by Edwards and Mcgee.

The approach I take is modeled on that of Richard Wyckoff and Peter J Steidlmayer (originator of Market Profile): rather than focusing only the patterns, they seek to understand the principles behind them and seek to keep an open mind on the alternative scenarios.

In the trad approach, the GBPUSD has signaled a clear change in trend in the 18-d swing (monthly trend). Figure 1 shows the H&S topping pattern with the breach of the neckline. The projected target is 1.5140

On the other hand, if you are using the material in Nature of Trends, the change in trend at this stage still needs to be proven.

Figure 2 shows the Barros Swing approach:

  1. There was an Upthrust Change in Trend Pattern with the possibility of adding to your shorts (assuming the core profit contract was taken at the Primary Sell Zone using the Rule of 3 – Nature of Trends).
  2. I discard the ‘neckline’, focusing instead on the price at low 1.5980 marked ‘B’. For me, at this stage, the change in trend is still in flux. There are two possibilities:
  • A Whole Point Count will form under 1.59809 confirming the change from up to down in the 18-day.  Or
  • We will see acceptance above 1.61184 in the form of a bullish conviction bar (open no lower than bottom third of range, close no lower than top third of range and at least 50% of body above 1.61184). This would suggest a move to at least the Primary Sell Zone and probably the resumption of the 18-day uptrend: note that although the 13-week does not have an uptrend structure (higher swing highs and higher swing lows), the 12-month (yearly trend) has triggered a buy signal to the Primary Sell Zone 2.1160 to 2.0220. A 12-month move to 2.1160 must mean the 13-week is in an uptrend. (See Figure 3)

The filters I use are not only price filters (Maximum Extension); I use a momentum filter (Line Change Count – see Nature of Trends) and the Whole Point Count (WPC, a time filter). The WPC for the 18-day is 9 consecutive days where the high is at or below 1.5980. Of the three, the WPC, the time filter, I rate as the most reliable and most important.

Overall I have found that by using confirming filters and keeping my  mind to alternative scenarios provides me a more profitable bottom line.




FIGURE 2 18-day Barros Swing


FIGURE 3 12-month Barros Swing

Understanding Timeframes When We Trade

BarroMetrics Views: Understanding Timeframes When We Trade

Last Friday,  September 18,  Scott Morrison (Managing Director, Tech Wizard) and I appeared on Trading Matters with CNBC’s Oriel Morrison.  Oriel has always brought out the best of  me: she has an easy going manner to which I respond; moreover, she is always well prepared.

But for me, Friday’s show was something special. At the end of it, all of us were amazed that the 30 minutes had flown by so quickly. You could say that we were all caught up in the flow of the moment.

As may be expected, Scott and I had our points of disagreement.

One area was whether “the AUDUSD would get to .8800 without correcting”.

As I was commenting on the question, I realized that our apparent disagreement lay with the fact we were probably talking about different timeframes. Scott took the view that the AUDUSD would correct before reaching 0.8800 whereas I took the contrary position. Scott expected a correction to .8500 before we’d see .8800. On Thursday, Sept 17, the high had been .8775 with .8509 being the low of Wednesday. (Figure 1).

The correction to .8500 was, in my language, a 2-day swing correction.

But as an 18-day trader, I usually look for corrections in the 18-day and occasionally in the 5-day swing. Yesterday, the market corrected to .8589 (Figure 2). So if the AUDUSD now moves to 0.8800, can we say that it corrected? Scott may well say ‘yes’ given that we did have a 2-day swing down; whereas I’d say no – we didn’t even half a 5-day correction.

This difference is not a mere academic exercise for our trading.

We need to identify the timefame we are trading. In this way we can avoid the dangers of hindsight bias. For example, if I say, the 13-week is overbought, I expect to see a 13-week correction; and in turn this means that the 18-day swing probably will change its trend. This allows me to assess statistically a time and price window for the correction – using 13-week corrective stats and 18-d impulsive stats as well as Fibo targets.





The Importance of the Maximum Extension

BarroMetrics Views: The Importance of the Maximum Extension

A reply to a number of e-mail queries: what is the importance of the Maximum Extension in my approach?

The Maximum Extension (see Nature of  Trends) is a price filter that provides advance warning of a change in trend from a sideways mode to a directional breakout. Those that have read the Nature of Trends know that the Maximum Extension is the greater of:

  1. Ten percent of the range of the most recent impulse swing (provided it is at least impulse means)  added (in an uptrend, reverse for downtrend) to the end of the impulse (10% of XA added to A) or
  2. Twenty percent of the range of the boundaries of congestion added to the end of the upper boundary of congestion (in an uptrend, reverse for downtrend) (20% of CD added to C in Figure 1)

In Figure 1 I show the AUDUSD with a 5-day swing (weekly trend). The bar for Fri Sept 4 is incorrect, the AUDUSD closed at .8550.

I’ll use Figure 1 to illustrate my use of the Maximum Extension. The AUDUSD was in a 5-day sideways mode beginning at ‘A’ (Aug 4). There was a high probability of it forming a HorizontalTerminal (see AUD/USD Setup?). But starting Friday we saw a possible breakout with last night’s bar being a bullish directional close. The question therefore arises: has the ADUS broken sufficiently above ‘C’ to say that the Horizontal Terminal idea can now be abandoned.

To answer this question definitively I use a time filter Joseph Hart’s calls a “Whole Point Count (WPC)”. In this case we would need to see 3 consecutive bars at or above ‘C’ (.8477) for a WPC to have occurred. The Maximum Extension provides an early warning that a WPC will most likely form: if I were to see acceptance beyond the Maximum Extension (.8555), I’d take the view that the Horizontal Terminal idea can be abandoned i.e. even if the market returns below the Primary Sell Zone of ‘ABCD’, I would not treat the pattern as a Horizontal Terminal.

What constitutes  ‘acceptance’ in the case of a breakout? I need to see 2 consecutive closes beyond the Maximum Extension and one of those closes must be a bullish directional close i.e. the open must be no higher than in the bottom third of the range, and the close no lower than in the top third of the range.

So, in the case of the AUDUSD, since we saw a directional bar yesterday, we need see only a close above .8555 today to have acceptance.



The Wide Range Breakout Day III

BarroMetrics Views: The Wide Range Breakout Day

In today’s blog, I’d like to draw some conclusions from the various strands of the past two blogs. But before I do that, let me say that the approach I advocate is for the more experienced traders; novice traders are better off ‘seeing a pattern, executing the pattern’.

What distinguishes the novice from the experienced trader is the willingness to admit a mistake and being totally open to any information provided by the market. Novice traders tend to distort, deny and generalize information that fails to support their preconceived ideas. Experienced traders keep an open mind and are ready to act on the preponderance of the evidence while keeping an open mind on the ‘correctness’ of their choice.

Let’s turn to the conclusions:

The blogs, The Wide Range Breakout Day I & The Wide Range Breakout Day II, seek to establish 2 points:

  1. Define and test the trading patterns you use; and
  2. Be aware of the assumptions and context behind the patterns. Use them to decide whether or not the pattern applies in any specific trade.

Computers have made it easy for us to test our patterns; but as a result of this, modern traders sometimes misplace the value of context.

Thursday’s ES breakout provides a good example of what I mean by ‘context’.

In the previous blogs, I showed that the breakout bar on Thursday July 23 fulfilled the conditions of a WRB but because of the context, I determined that it was not a signal I was willing to take. Instead I am looking to go long on a retest of the breakout zone. Recall that a WRB suggests the retest will not take place because the underlying conditions are so bullish (or bearish on a downside breakout).

Can I be wrong about this – perhaps Thursday’s bar was a valid WRB? Perhaps; and I would be the first to admit that my assessment could be wrong. But this is true of every trade that I take. Ultimately, it is important that we have confidence in our judgment of the probabilities as well as in the approach we take.

The Wide Range Breakout Day II

BarroMetrics Views: The Wide Range Breakout Day

The assumption behind a Wide Range Breakout Day (WRB) is this: current bullish conditions are so strong that  we don’t see the normal retest after a breakout.

In yesterday’s blog, The Wide Range Breakout Day, we saw that the breakout bar on Thursday met the definition of a WRB; but I also said that things were not as clear as I would like them to be. This blog examines why.

Figure 1 shows the Normalised Volume from the March lows. The volume within the red bands shows ‘normal volume’ when our sample runs from March 9 to date; the volume within the purple lines runs from the May high (Left Shoulder of the aborted H&S) to date. You see that  since May, ‘normal volume’ contracted.

The same story can be told about the range. Since 2000, the ‘normal range’ for the S&P has been 20 to 25 points, with a standard deviation range of 8 to 10 points. Since May, this has fallen to 16 to 17 points, with a standard deviation of 6 points. The lower standard deviation tells me that since May, the S&P has had consistently lower ranges than the historical norm.

The breakout on Thursday had a range of 26 points and the volume just made ‘normal’ if we use the March lows as our starting point. What I am saying is this: but for the severe contraction since May, Thursday’s bar would not have qualified as a WRB. Therefore we can expect a retest of the breakout zone.

Figure 2 shows a 30-minute chart. For reasons I go into in the Forum analysis (available if you register for the free BarroMetrics Weekly Video), I have opted for that pattern to be ‘3 Drives to a Top’.

Figure 1 shows that while the daily range dropped on Friday and yesterday, yesterday’s volume was greater than yesterday’s. Add this to the fact that yesterday the buyers were in control (MarketDelta Chart and the higher high, higher low and higher close) and we have all the earmarks for a 1-day reversal swing.

The intra-day patterns support a retest of the 5-day breakout zone 952 to 947 (basis Sept). But there is one more point to add:

Note that we have not accepted above the 18-day Maximum Extension. Basis the daily chart (Figure 3), this comes in at 981 (basis Sept); the 30-minute pattern has extended it to 983. That being the case, the possibility of an 18-day Upthrust is still alive and will be triggered on acceptance below 937. I rate this a lower probability than a successful retest of 952 to 947.

Tomorrow I draw some conclusions on pattern analysis using the WRB as an example.


Figure 1 Normalised Volume

Charts through the courtesy of Market Volume


Figure 2 30-Minute Chart


Figure 3 ES 5-d and 18-d Swings

Charts through the courtesy of  Market-Analyst

BLOOMBERG VIDEO -Chinese IPOs -Ray with Bernie todaybernie-lo.jpgray-on-bloomberg-with-bernie-lo.jpg

The Wide Range Breakout Day

Barrometrics Views: The Wide Range Breakout Day

On Thursday, on the S&P (basis cash), we saw a day with a  26 point range and a Normalised Volume Reading of 5,107,846,144. Since May 8 (Left Shoulder of aborted H&S), the daily range has been 16 points with standard deviation of 6 points and Normalised Mean Volume of 4,025,661,196.

I tell you this because over the next 2 blogs, I’ll be considering a bar I call a Wide Range Breakout Bar.

A wide range breakout bar is one where:

  1. the range is at least greater than mean and preferably, mean + 1 to below mean +2. And,
  2. the volume is also with greater than mean and preferably, mean + 1 to below mean +2.
  3. the structure is one where at least 50% of the bar is above the breakout point and the bar exhibits buying conviction (for an upside breakout) i.e. we have a close now lower than in the top third of the range and the open no higher than in the bottom third of the range.
  4. the breakout is NOT concurrent with a major news event e.g. a CPI release.

If I see a wide range breakout bar, I will assume I shall not see a pullback to the breakout point – a response I consider to be the norm.

If we examine Thursday’s bar, we see that it has qualified on all 4 grounds (Figure 1).


FIGURE 1 WRB Thursday July 23 with Normalised Volume.

Charts courtesy of Market Volume

With a WRB what I expect to happen is for the market to test the 50% level of the directional move and then resume the move in the direction of the breakout. Figure 2 shows that this is what occurred on Friday.


Figure 2 30-minute Chart with Market Profile

Charts courtesy of Market-Analyst

So on this basis, can we conclude that ES will resume its upmove tonight? Unfortunately, the situation is not as clear as I’d like it to be. More tomorrow.

By the way, you can find timely comments as markets unfold at the Video/Forum/Twitter service. To join, register for the weekly video.


Shanghai Composite and more



The Perception of Volume Patterns

BarroMetrics Views: The Perception of Volume Patterns

At one stage, one of the ‘hot’ tools was Richard Arms ‘equivolume charts‘. Then in 1992, Gregg Morris developed “candle volume charts“; he called it CandlePower. His candlevolume charts were candlesticks drawn the equivolume way: where the horizontal axis was volume rather than time. The software I had that drew the charts ceased to work when Y2K came around.

At the time, I thought ‘that’s a pity’ because Candle Volume Charts suited the way I trade. I am a visual trader and my first port of call is always the graphics and then the numbers.

Recently I came across Insider TA which draws Candle Volume charts as well as Equivolume and its moving averages. It does much more, but I am using only the CandleVolume function. What CV charts allow me to do is see at a glance the ebb and flow of volume in respective time periods.

Figure 1 (Australian All Ords Index) shows what I mean. Once I get a feel for the periods when there are noticeable differences, I can then apply my statistical analysis.

Let’s take a look at another example.

Figure 2 is the cash S&P. The first thing I note is that the volume in period 1 is much greater than period 3. Period 2’s volume is about the same as period 1 but the ranges appear smaller. With those first impressions, I can use Market-Analyst’s Probability Box to confirm or reject the initial hypothesis.

(Usual disclaimer: I receive no referral fee for the products mentioned. I use them in my own trading and I find the items useful).