Here's something I didn't know before:



Did you happen to notice this pattern, Pup?

Unfortunately, being a very old technical tool it doesn't work as everyone knows it.

Like many charting techniques they all work great on past charts but do not predict the future.

I had never heard this one either. I will have to study it a little bit.

I am with Bobbyald for the moment.....

After studying this a little, my new opinion is that it works!!!

I know this is a shock.

I went through the daily crude oil NYMEX price database from the EIA for the raw data, and defined "market event" as follows: When the close on a given day exceeds the 100-day moving average by greater than 2-sigma (twice the 100-day standard deviation) then that's an "event" (this is derived from Deming and the statistical process control types). It kicks out the random variation and focuses on "underlying market changes".

There have been 22 "events" since Jan 1, 2000. Of these, 9 events fall on a "fibonacci" number +- one day, or about 45%. An "extended event" i.e. a "trend" can sometimes last for a week or two, and that shouldn't necessarily count as several "events". If the "event" lasts a couple of weeks, I identified the highest high (or lowest low, if it was the low event), and defined this as the "event". If the "event" was just a day, I counted it as an event.

If you count the number of fibonacci numbers between 5 and 144, and allow for +-1 on either side, that's 24/144 or about 17%.

So, the "market events" fall on a fibonacci number about 3 times more often than random.

Now, I will tell you why it works:

There are 21 trading days in a typical month (21 is one of the numbers). so this is just picking up the monthly cyclicality. If, for example, some non-fibonacci related report (such as crude stocks) comes out every month on the same day (plus or minus one or two for holidays) and drives the market crazy, you will pick it up on the 21 day chart, if you are using the daily data. Also, all of these crude oil contracts have an expiration date once a month and the market gets crazy when one month expires and you have to move out to the next month.

If you are using the monthly data, same thing with 13 weeks being a "quarter" or a "season", 3-5 weeks being a month, and so on.

All of these commodities have similar expirations, option expirations, and market reports that come out all the time. The weekly crop report in August-September is notorious for screwing up the soybean and corn markets. Some similar thing comes out on livestock.
For the stock market it's the Greenspan meetings(six weeks, 34 trading days), or the unemployment statistics (weekly), or quarterly earnings reports (13 weeks).

I think better just to figure out the seasonality and go with that.

By the way, if you trade the "event reversals" as defined above, you will be correct about 6 out of 10 times. Example: If a "market peak" event of at least five days (per Deming) occurs, on the day after the price returns to within 2-sigma of the moving average (in other words, the event ends), short the market. 60% of the time, the market will fall at least $1.50 from the day you shorted it, so you cash out, and wait for another opportunity. Similarly, if there is a market bottom, the day after the event occurs, go long and it will gain at least $1.50 60% of the time. Don't get greedy.

We might be having an "event" right now. I will have to compute the daily prices and moving averages and see.

Update:

If the market closes above 55.57 today, this will be another "event", in which the price closes more than 2-sigma greater than the 100-day moving average. The 100-day moving average is 47.92 and the standard deviation (sigma) is 3.82.

The last time the market entered this state was on September 22, and it lasted until October 26, 22 trading days. The market went from 48.41 to 56.37, $8 per barrel.

The time before that, July 28-Aug 20 (17 days) the market went from 42.81 to 48.66, $6 per barrel.

Therefore here is the official pup55 prediction:

The upcoming "event" will last 20 days and cause the price of oil to go up $7 per barrel (the average of the two previous "events").

The price on April 6 will be $63.00 +/- $0.50

If this prediction is correct, the moderators need to start charging a subscription to this website because there is a lot of money to be made by greedy speculators anxious to know the future.

If it is wrong, I will refine my predictive method and try again later.

I get relief. It did not close above 55.57. I will recompute a new forecast when the "event" finally occurs.

Subscribers to the Elliot Wave theory of economic activity are predicting a grand cyclic turn down at about this time using fibonacci numbers. Perhaps you were referring to Elliot wave pertaining to oil too, in which case, I'm being remarkably redundant.

The Golden Section and Elliott's Wave Principle

On the previous pages of our Museum we have considered many different applications of the "golden section" and Fibonacci numbers - in the Solar System, in the Egyptian pyramids, in the art works, in plants and animal morphology, in the cardiac mammals activity, in the modern computers. But on this page we will tell about the rather exotic application of the Fibonacci numbers. The question is about the surprising regularities found by the American engineer and accountant Ralf Elliott still in the 30th years of the 20th century at the market fluctuation research. These fluctuations are called in modern science as the "Elliott Waves".

http://www.goldenmuseum.com/1609Elliot_engl.html

I'm going to step in here and declare that this is really just a bunch of numerology. Put another way - in a system with a lot of noise (market activity for example) there are bound to be theories which fit the data well enough to give one confidence that they are correct - but in fact are wrong. Further I don't see a comprehensive analysis here that indicates there is well defined way in which the markets produce fibonacci ratios - except pups analysis which is primarily due to the 21 day cycle apparently (can you break down the number of events etc by which fibonacci number they fall near).

One thing that may illuminate this a bit is to note that the ratio between consecutive fibonacci numbers approaches ALPHA = [1+rad(5)]/2 (if memory serves) as we move along the sequence. Hence the sequence is an additive approximation to an exponential function. Now can anyone here give justification as to why this ratio should have more influence in the market than any other? Perhaps you could make an emprical argument that ALPHA is special due to it's frequent approximation in nature . . . but so is rad(2) - or `e' for that matter. Or PI.

So what? The fact is that the systems in which ALPHA and/or fibonacci numbers emerge biologically are not really analogous to markets or human society - they are far more orderly. The # of sprirals in a sunflower going one way - and then the other are consecutive fibonacci #s (55 and 89 for the giants). The ratio of the long knuckle to the middle and the middle to the short etc etc. But these are pretty obviously emerging based on how biological systems use enzymes to calculate and control growth - and it really would not be surprising if these systems were simple additive systems and hence ratios like such emerged. Unlike markets the way that the body decides what tissue grows where and how much is under central control.

So in short - I think this is nonsense.

Xelat,

Do you have your Curriculum Vitae handy?

CV probably isn't necessary. All a person needs is good enough sight to read the gas prices.

By closing at 56.46 today, the nymex market closed at 2-sigma higher than the 100-day moving average, which is our definition of an "event":

The official pup55 prediction is as follows:

Oil will be at $65.21 +-.50 per barrel on April 15th.

The last occurrence of such an "event" was last October. In 20 trading days after the "event" occurred, the oil price went up about 14%.

Also, in four out of the last 6 years (99, 00, 02 and 04), there has been a similar seasonal runup in prices in March. These have averaged about 20 days in duration, and have resulted in an average 14% increase in prices.

If this prediction is correct, and the current overheated market is true to form, the market will have increased 82% in the 200 days prior to April 15th. This runup in oil prices will be comparable to the market-crash-inducing 80% increase of May 1987, and the 91% recession-causing increase of September 1990. We will check back to see what happens.

By the way, in each of the previous March cases, the market ran out of steam in about 20 days, and corrected back down to the base price or lower.

Go, Pup, Go.

Xelat may pitch in at some point.

Pup,

Take a look at this:
http://www.kucinich.us/phpBB2/viewtopic ... rt=20#8489

Sorry to upset you, but I did study Maths at college and I know a lot about the Fibonacci numbers. One thing I know is that they have nothing to do with market fluctuations. People keep seeing patterns in them, but it doesn't mean that all the patterns mean anything or have any predictive value.

Excellent!

I will have to browse through this site for entertainment.

You know, of course, that if we have a hell of a recession between now and 2020 (ref: 1979-82) Randall might be able to collect without the peak actually occurring, but some one will no doubt point this out. He should probably have made the claim about 4 years in length.

But still, a fine effort.

By the way, in light of the fact that my prediction of March 15th appears to have been a stinking failure (I still have 3 days), I am working on a new one which I will reveal on Friday. Today's inventory report should cause more weakness in the oil price.