Stage 5 - Wws and H&S on RSI

Much easier to describe what a WW is or an Mm is, when you can actually show people!

When it comes to trading them, it all depends........

The RSI patterns only show the closing position, so a lot could have happened on Price during the candles - this is why backtesting anything based on RSI is fraught with problems.

If the original trade has survived, then maybe it's an opportunity to add another partial or even whole one, or maybe just enjoy the stress free ride.

If the original trade got stopped out, then the next one should stand on its own merits IMO - but most humans will be swayed to go with a borderline one.

Now for the Head & Shoulders (H&S) on RSI. This is not an MM because the initial M would be a rising one i.e. Pt3 above Pt1 and that is not allowed under any circumstances. All we do is trade the second part i.e. the basic M shown.

The Price action that matches the RSI pattern could be a H&S on Price or even a set of rising tops. This does not matter, I am trading on RSI patterns.

cheers theory

Real life example - FTSE 20 min 12/05/09

Notice the RSI window is far bigger than the Price one. The RSI is set on Double thickness Points. The lines of best fit are put in manually with the segment drawer.

This trade shows just the RSI levels and Price movement relative to the Bollinger Band.

The RSI of Pt1 was greater than 70 - Group A.
There was a move above then back under the Upper Bollinger - Group A. The double move is a bonus and not necessary for membership of Group A.

The S/L placement was just above the highest point within the M on Price.
The entry was governed by the RSI hitting 63.5
The risk is the difference between the two values, it was fine!

The initial target was the middle Bollinger line, with a further one of the Lower one. Compared to the risk, even the initial looked very good indeed.

So it was a high probability, low risk, potential high reward trade.

cheers theory

Stage 4 - Context,context,context.

Assuming that you now know you should be looking for tight sharp WMs on RSI - you will be rejecting those which have internal dojis or near dojis on Price. You will soon gain the expertise to find the timeframe on which the dojis disappear and the required shape appears. If there is no timeframe to do this, there is no trade!

Probability of success

Check the RSI level - which group will the shape be in?
Check the Price action compared to the Bollinger Bands and middle line. Which group is it in?
Is there a nearby Resistance/Support level that could well act in your favour? (See the grey and blue coloured lines on the diagram.)
Likewise, is there a trendline of Fib level?
If you use EWT do you think you are looking for an end of Wave 5 or a C?
Use all of these to assess the probabilty of its success.

Assess the risk to your capital

How far is it from the B/E entry on Price to the place where you are going to position your S/L?Anything up to 20, I use whole trades. Once it gets above this I scale it down, so 40 needed will only be a half trade. This limits the max risk to 40*0.5=20.
You have now assessed your risk.

Assess the extent of your initial reward

This means taking into account the nearness or otherwise of a Price level which will get in the way. Typically these will be a previous strong resistance/support level or the middle Bollinger line (which is why I like widely spaced Bollinger Bands).
Is there "room to trade"?
Can you get the trade to B/E and still be a reasonable distance off the Price action?
Is the the distance to the problem area at least twice the initial risk required?
You have now assessed your initial target reward.

The ideal trade?

Timeframe >= 15 mins (guideline)
High probabilty of success.
Low risk to capital.
High initial target reward before you get to a problem area.

What Trigger & I call a HPLRHR trade.

cheers theory

Stage 3 - Use of Bollinger Bands on Price

Bollinger Bands are a well known measure of "unusualness". They are based around the calculation of a standard deviation (SD) from the Price data. The more multiples of the SD the Price moves away from the mean, the more unusual it is.

The problem is of course two fold. Should one even attempt to calculate an SD from the data input and do the conditions remain valid for it to be used with any sort of meaning, given that it had any in the first place!?

Fakeouts are the death knoll of any system based on going Short on an initial move above the Upper Band. Likewise going Long on an initial move through the Lower Band. When the market starts to trend, the Bollinger Bands widen and it's not unknown for the Price to move along the Band in question.

Trades can be split into groups based on the Price & Bollinger Band action. Just like with the RSI levels, Group a has the highest probability of success, B less so………

Group A, the Price moves outside the relevant Band and then returns, if it does so again that is just a bonus.
Group B, the Price touches the relevant Band.
Group C, any other, BUT in particular when the Price bounces off the middle 20ma line.

The wider the distance between the Bands the better it is in terms of potential reward - the middle line and the opposite Band are further away.

Expect the trade to stall at around those levels, if not even reverse.

cheers theory

Stage 2 - RSI levels for trend reversal or continuation.

Trend reversal RSI patterns fit into 4 distinct groups. It is the level of Pt1, and Pt1 alone, which decides the category they are put into.

For a trend reversal W, basically it's the lower the better in terms of deciding the probability of delivering a B/E trade.

Group A, highest % chance Pt1 has RSI less than 30.
Group B, next highest % chance Pt1 has RSI greater than 30 but less than 35.
Group C, even less % chance Pt1 has RSI greater than 35 but less than 40.
Group D, least % chance Pt1 has RSI greater than 40 but less than anything you want.

For a trend reversal M, basically it's the higher the better in terms of deciding the probability of delivering a B/E trade.

Group A, highest % chance Pt1 has RSI greater than 70.
Group B, next highest % chance Pt1 has RSI greater than 65 but less than 70.
Group C, even less % chance Pt1 has RSI greater than 60 but less than 65.
Group D, least % chance Pt1 has RSI greater than whatever you want but less than 60.

Trend continuation WMs are totally different.

At NO stage were they ever the opposite shape and then they changed back within the triggering candle.

In terms of Price action, they are non-threatening retracements within the original trend.

cheers theory

Stage 1 - What are WMs on RSI?

May as well start off with the most basic of the underlying ideas. A W on RSI is a potential buy signal and an M on RSI is a potential sell signal. There are a lot of filters they need to pass through first, before they become confirmed signals to trade.

At this stage I am only going to show W formations. I used to refer to rising Ws, but now I don't bother. The pt 3 MUST be higher than pt 1, if it's level or lower then it's no trade!

Throughout my work, the emphasis is on RSI points and levels - I have the RSI window on Double thickness Points, any line is added by using the segment drawer or ZigZag set on Point=1.

With minimal practice you will soon be able to spot Ms, where of course pt 3 must be lower than pt 1.

The 5, 6, 7... pt shapes are those I use in my WM finder software package - hence I've kept the naming. RSI is ALWAYS 14, so that will not be mentioned again.

The pt 1,2 & 3 labels are those used in the 1-2-3 system on Price, so I have kept them even though I work on RSI.

For an excellent view of 1-2-3 trading, I suggest you read:

Mark Crisp's 1-2-3 trading.

[Some systems will prevent you getting there directly BUT just click on the link you come to, it works.]

The matter of when a W forms is personal - reversal within the same candle is a nightmare!

I allow a 1 pt leeway before entering - see diagram above.

I also like at least 1 pt difference on RSI between, pt 1 & 2, and 2 & 3 - otherwise I am leaving myself open to attack from "market noise". The candles on Price should have "chunky" bodies, so dojis and "almost dojis" also get rejected by using this criteria.

cheers theory