For Wednesday I chose a 10 min chart, I do not want to cheat here, but Wednesday's range was small and in the end it turned out to be an inside day. So if you look at a 30 min chart, you will not see any support and resistance lines, while a 10 min chart may catch soem of the smaller swings to display Fibonacci lines.
By the way: all charts use the same rules and default configuration to establish Fibonacci lines - there is no optimization to produce charts that explain price action in hindsight. I usually use 30 min charts to locate support and resistance, and smaller time frame charts to place the trade.
But as an exception, I will look at Fibonacci lines generated on a 10 min and 3 min chart, to check the validity of the concept.
Two Fibonacci lines contained the price action of the entire day. There is something I had observed three days earlier. If there are two lines, say a pivot line and a Fib line, prices often stop at the second line. Maybe the reason is that there are not enough traders to follow pivots, but the combined forces of pivot and fibonacci traders create enough support and resistance.
Now if you look at the 3 min and 10 min chart, you will observe that the upper fibonacci line was shifted down on the lower timeframe chart. This is not magic, but going to a lower timeframe will add lines calculated from smaller swings and not look for distant swing highs and lows. This may create a shift by adding weight to a confluence zone, which will then "defeat" the other confluence area using smaller ammunition.
10 min chart : upper fib line at 1169.50
3 min chart: upper fib line at 1168.25
Expected after an inside day: Breakout or False Breakout and Reversal
The following 4 users say Thank You to Fat Tails for this post:
Looking back from Thursday to Wednesday, there was only a single red bar causing a shift in perceived value. One bar for the day. This shows that most of the time is spent in trading ranges.
One single green auction bar (see chart) defined the night session of Wednesday, another single red auction bar (see chart) defined the day session of Wednesday and the night session of Thursday! The breakout only occured at 7:30 in the morning.
What could be observed then, is a Fibonacci ladder. Each fib line caused temporary resistance, then the lower line would offer supportand price would climb to the next level. 1176.50 was terminus and prices came back ansd closed below yesterday's close. The down hill action also stopped at all fib lines retraced back to the higher levels and then down again.
Confirmed: Combined Fibonacci/Pivot levels are excellent entries for pullback trades.
The day became a classical reversal day, a very bearish sign. Again these are my exact definitions for a reversal day and this is what I coded for the Auction Bar indicator to display the yellow bars that can be seen on the chart:
(1) Range must be larger than 80% of ATR (14)
(2a) For a reversal down: Close and open within the lower third of the bar
(2b) For a reversal up: Close and open within the upper third of the bar
(3a) For a reversal down: Close <= Open + 0.2*range
(3b) For a reversal up: Close >= Open - 0.2*range
(4) Ideally the tail should be outside the range created by the last auction bar, the close within.
To summarize: For a reversal up, I want to see a healthy range bar with a long tail down and an upclose. For a reversal down, I want to see the opposite. This is the setup. Often a reversal bar is followed by a balancing bar. The trigger occurs, when the high of the reversal bar up is breached.
Let us check all daily reversal bars that occured since January 2009:
Reversal bars up
January 15, 2009 : next day upclose, then bearish engulfing pattern, was a first entry and therefore too early
February 12, 2009 : although successful retest of a higher low, was a failure -> after the failure the market declined by 20% during 3 weeks
November 3, 2009: reversal bar followed by 6 upcloses in a row -> those who had entered at the high of the reversal bar and exited at the close on the auction bar would have made over 50 points on the ES
February 5, 2010: started the rally of this year -> over 100 points for the ES
February 12, 2010: excellent entry -> over 80 points
February 25, 2010: excellent entry -> over 50 points
Reversal bars down:
February 19, 2009: excellent entry, market lost 15% within 3 weeks
March 16, 2009: The setup was never confirmed, as prices did not break below the low -> no trigger
April 30, 2009: False breakout to the down side, then up move -> failure
May 20, 2009: 20 points down next day, but no follow through
June 5, 2009: excellent entry -> over 50 points
June 11, 2009: excellent entry -> over 50 points
January 25, 2010: good entry -> but stop loss should have been 3 to 4 ticks above the entry bar, as it was retested
March 25: This was last Thursday, it was followed by a balancing day.
Observations: High probability trade with excellent reward to risk ratio
- There were 14 reversal bars since the beginning of 2009.
- 7 of them started moves that lasted between 2 and 7 weeks.
- 3 of them produced an engulfing pattern on the first day and started extended moves in the opposite direction
- 3 of them would have stopped you out above or below the entry bar
- 1 is still under observation
And volume ?
March 18 was a weak churn bar, March 24, was a strong churn bar. For a topping pattern churn bars usually precede the actual reversal, so we may still see a test of Thursday's highs or a limited move above. For example the January 5 churn bar closed at 1127 and the actual high was made 4 days later at 1144, which is 1.5% above the close. But the January churn bar occured well above the trend, whereas Wednesday's churn bar is at the trend, a second churn bar and shows a down close.
I think we have clearly entered a distribution pattern. If Thursday has not been a major high, the high might be near. Be aware, however of any bullish engulfing pattern, this might trap a lot of traders.
The following 6 users say Thank You to Fat Tails for this post:
The volume profile indicator is Dean's dValue for NT7 that I have downloaded from the NinjaTrader forum. I have coded the other indicators, partly it is still work in progress. I have already put the pivots indicator on this forum: SessionPivots V1.6 for NT7.
The following user says Thank You to Fat Tails for this post:
It was my intention to review the full week, but I ran out of time, so it was incomplete. So I will now comment last Friday.
On Thursday I made an analysis of reversal bars, so I added them to the chart now. There were two tradeable reversal bars on Friday. Both showed long tails that were outside the prior trading range. The second one was a 2B retest of the lows.
Thursday was a down reversal day, so the first duty was a retest of the low, which happened during the night session. Afterwards the price moved up but stayed within the trading range created by the auction bar created by the last 30 minutes of the cash session.
Learned: An Auction Bar can easily set the trading range for an entire session. There was no perceived shift in value on Friday!
The whole of Friday was therefore contained within this last bar of Thursday. as you can see this trading range is as tradeable as Fibonacci lines or pivots. Watch out for multiple support and resistance. Let us have a look on these for Friday:
- Lower border of range created be auction bar and yesterday's low (with fibline and pivot below for stop triggering expeditions)
- Fibline near Midpoint of auction bar and main pivot (prices stayed here for about 7 hours)
- Yesterday's VWAP, Fibline, Yesterday's midpoint and upper border of range created by auction bar.
Trading opportunities indicated by the 30 min chart:
(1) Test of low during night session : Excellent risk/reward, stop below the low of the session, target at next Fibonacci lines
(2) Breakout in the morning: After a symmetrical triangle had formed, there was a doji - hammer or hanging man - that confirmed equilibrium just above the fib line. It was possible to trade the upside breakout with a stop just below the hammer. The breakout failed, so this would have left you with a scalp / breakeven situation.
(3) After the failure of the breakout you could have traded the opposite with an entry 1 tick below the failure bar. With yesterday's close as target and a stop above the failure bar, the risk/reward ratio would have been 1:1
(4) reversal trade was cleared by the trendline break (on a 5 min chart) and the retest of the high. You could have entered below the green bar. First target would have been the fibline with a risk reward of less than 1:1, but after a triple failure of the bulls the trade was worth taking.
(5) The reversal bar showed that the bears were not strong enough to drive price out of the trading range. You would then expect a balancing day with a close near the middle of the range / today's VWAP.
Let us see how the Thursday's reversal bar on the daily chart will play out!
The following 4 users say Thank You to Fat Tails for this post:
Fattails really nice work, I always use auction bars to set the expected range, and also the volume from that bar to judge the intensity of the break out.
Also volume on the edges of those bars should be analyzed as they touch. volume vacuums etc for fading
In you first post you mentioned working out probabilities on fib retracements and extensions, is it possible for you to post that indicator, i would like to understand the method how you do that.
Are you also going to release the range indicator, that looks very nice, i already do this by sight as i trade, it would be handy for me to look at someone else's thinking and see if i can adapt it to share
The following user says Thank You to tinkerz for this post:
thanks for the comment. I absolutely agree on your statement that it is necessary to observe volume on the edges of the trading ranges. Volume spread analysis is my current favorite, as it confirms support or resistance. Below an example for ES about an hour ago. You can see how price hit the double support (fib confluence and floor pivot R1). This was the ideal entry point for a pullback trade. However, I needed some volume confirmation, which I got from two different volume indicators.
(a) my own better volume indicator, which showed a higher volume expansion bar (light blue) followed by a strong churn bar (yellow)
(b) The signal from the Volstop indicator (thank you Cory), which showed rising volume on lower lows with a neutral/upclose for the third bar.
With the volume confirmation this was a low risk trade with a first target just at the next fib line and the second target arounf the midpivot.
The following 6 users say Thank You to Fat Tails for this post:
The fib confluence indicator is a probability game. Different traders watch out for different levels, so if you find an area, where several of these levels add up, there is a higher probability of a price reaction. Also not all levels are created equal, but some are more equal than others.
Let us take the example of retracements. This includes expansions, which are just retracements > 100%. Now what is the probability that a line generates a price reaction? To measure this you will take a zigzag indicator (all swings smaller than the minimum deviation are considered as noise) and express the size of a swing as the percentage of the prior swing. You can divide your space for the swing sizes in equal intervals and draw a first distribution. Now this distribution will show a lot of small swings and fewer large swings. The median will be around 100% if the market has no bias. Now the point here is that the relative number of swing sizes within an interval does not represent the probability that price reacts to a fib line. To obtain the latter, you will want to calculate the conditional probability that price once it has entered the intervalretraces the defined minimum distance, before it hits the next interval.
Now if you look at the conditional probabilities, you will find that expansions have a higher probability than retracements < 100%. Once the price has retraced 161.8% or 200% it is more likely to revert than for smaller retracements.
But the probability also depends on the size of the intial swing, which represents the timeframe, so to obtain the specific probability for a line you woud need two coefficients, one for the type and size of the retracement and another one for the size of the intial swing. Assuming that your swings are significantly larger than the tick size you can base your model on scale invariance and multiply the coefficients.
This is done by the fib indicator, which the adds up the probabilities of all the lines within a predefined (depending on volatility) confluence range. If the probability exceeds a minimum threshold, the strongest line of the confluence range will be displayed. The indicator evaulates up to 540 lines in 10 timeframes.
This was lot of work, so I do not distribute it open code, but send me a private message, if you want to test it.
The following 3 users say Thank You to Fat Tails for this post: