PF3 Discussion – 17/12/02

PF3 Discussion – 17/12/02

I called it the PF3 system, wanted to see what profile was like when you had positive slope of 1) 2 period ROC
2) The fast line on the 3/10 oscillator,
3) The slow line on the 3/10 oscillator.....(16 period SMA of the 3/10)

PF = positive feedback.

You rarely get more than 2 PF3 days in a row, they are like short term market impulse. So if we had 2 PF3 days UP...what happened the following day? (was good day to sell any further rally)

BUT also, you could get days where you would get really hurt if you sold the day after 2 PF3s up, so what was the profile of this condition?

Also, what happened if you had 2 PF3s up and then you bought the first pullback, 2 days later?

We posted three chart examples showing the PF3 bars on Beans, bonds and SP’s. We profiled these bars and the days after these bars every which way and back.

When a bar is WHITE, the slope of the 2-period ROC (rate of change) is plus. The slope of the 3/10 oscillator is PLUS (the gray line) and the slope of the longer period 16 moving average, the PINK line is plus.

Quite frankly, the 3/16 fast line is the least significant variable...maybe about a 54% edge there. Most of the power is in the 2 period ROC, combined with the trend of the slow line.

So when we first started testing this, we would look at what happened if you sell the CLOSE OF THE NEXT BAR AFTER A PF3 up, or the opening of the next bar after a PF3 up and then ask, what happens on open and close of the next bar and also day 2 open and close...what is average holding time?

We are looking for something that works across all markets. When testing off daily bars, really best to test enter on close or open, or at previous day's high or low or range expansion function...don't try and optimize this looking for intra-day entry or trigger. We want to look at basic tendency only.

Also tested out looking for entry the other direction...i.e. if we have 2 PF3's up (white bars), what happens if the buy the close of not the next day but the day after that.

Believe me, we could not come up with a mechanical system based on these rules alone. But you see, it does make excellent pattern recognition...it really keeps you from getting too bullish after PF3s up...need that one day correction, and vice versa.

I look at this every night. You notice coming in yesterday morning. We had had one PF3 down, and I did not know if we would get one MORE down, but you can see that the 2-period ROC had already been down two days. I am just really careful after we have an NR7 day, because that has always been an excellent filter for momentum type of stuff, range contraction.

And, if in fact, you look at the left hand side of the SP chart, you will see nasty run to the downside and likewise with the bonds it would not have been a good idea to have been LONG on the close of that second PF3 DOWN (you see the gap). Here was rare run, too...nasty downside, that type of run, where the PF3 FAILS to provide a check after the 2 big bars, that type of run only happens about 10% of the time.


So now we want to ask, what type of conditions PRECEDE that type of run, where we could get 3- 4 days in a row? We sure want to avoid thinking in a countertrend mode in these periods. And the conditions that precede this type of move are:

1) A divergence on the daily charts
2) A break from an extended chart formation...a low volatility area.
3) When the HIGHER time frame (weeklies) is just turning back up or down.

So we can work backwards in a roundabout way to see where the home run trades come from...the runs like we saw in the crude to the upside (which we exited too soon), or the gold, or the EC, or nat gas.

PF3 is a strong signal, so the fact that a FALURE to check back after 2 PF3 days indicates an even stronger force behind the markets movement. Well this is how I construct a model in my head each night of what I am expecting the market to do. 90% of the comments that I make on my trade sheet, that you have been watching get posted for quite awhile now, come off this model.

You can see in the bonds, the chart that we posted, how there was 2 PF3's up on the right hand side of the chart. The next day, it may not look at first like the market was going to check back, but it did.




On the trade sheet that we post, there is a number on the far right column, it says STpvt = pivot. It is the level at which the 2-period ROC will flip up or down.
Where it says Sig, there is a B for the SP’s. This means that it has only flipped up one day.
We should be looking for one more day up today (except that a trend day like we had yesterday often uses up all its bullets in one day)

Better example...coming in this morning...the Canadian Dollar had (s) meaning the 2-period ROC had been down 2 or more days in a row and was poised to flip up again. We see that if this market was above 63.62 today, it would turn the momentum back up again. It is up almost $500 a contract so far.

Now, there was also an NR7 notation. That is one thing that can throw off the rhythm of the 2-period ROC, so we always respect the fact that the market is in breakout mode FIRST, and then look for directional bias second.

Well...maybe this can give you an idea of the twisted little way I have gone about constructing my price models...again, most of the edge is in the 2-period ROC....

Now, if you want to really play mind games with yourself late at night. Now try to figure out the conditions that PRECEED a good PF3 move. Instead of taking our cue as to how the price behaves the day following and the next day following a PF3 what conditions precede the best PF3s, because this is ideally what we want to capture!




Blue bars are PF3 days down. White bars are PF3 days up. PF3 = positive feedback between three indicators...lots of ways to test these patterns!



Big Day Pattern

But I can take a few minutes and explain a novel pattern to you. Was something that came out of our Golf System that we used to trade many years ago and is also something that I go over in our workshop... it is a momentum phenomena.

You will notice that I posted 2 charts to the website that have numeric tables...
One is of the bonds. The second is on the beans because I wanted to show you this pattern on another market too.

There are specific patterns that setup when Momentum has extreme readings they do not work 100% of the time but they work enough of the time that I have a lot of respect for this pattern...it works particularly well in the SP’s.

It sets up when there are two consecutive HIGH extreme readings in the 2-period ROC. It sets up a sequence of 5 trading days which we call the BIG DAY pattern which is a pattern of 5 bars...1-2-3-4-5.

Days 1 and 2 are the extreme readings...If you look at the bond table that we posted, you will see 2 readings under the header MTM -1'16 and -1'02. That is day 1 and 2 and then day 3 is the countertrend reaction. A very aggressive trader can go LONG on the close of day 2...looking to exit Day 3. (Day 3 was Friday in the bonds)
.
But, you can also go SHORT on the close of day 3...for the great majority of the time you will gap DOWN and not see the highs of day 3. You will see how this makes sense when I show you several examples.

This is one of the reasons why I was trying to stay short the bonds on Friday, we just shorted too early in the day on this trade. But, you can see that the highs of Day THREE were not exceeded...and ultimately, the market has tendency to retest back down, make lower lows, and then possibly setup short term divergence pattern in the 2-period ROC.

Bonds - Sell short day set-up – 14/05/02

Close of Day 3 is rarely seen on Day 4 of Big Day pattern.

When we used to trade the golf system in the SP’s, one of the rules was never to INITITATE a trade on the close of day 3. I know there are some older members in this room who have heard my lectures before on the GOLF system, and it is something that I will go over again, in the future, even though we do not trade is mechanically now.

But it was studies in the 2/period ROC - and this is also where we learned not to initiate Golf trades on NR7 days. At any rate, let me clarify with a another example....

Beans - example of Big Day count – 13/05/02

We have both an example of a Buy and a Sell on day three of the Big Day pattern

If you look at the next chart of the BEANS you will see 2 large MINUS readings on the table.... -13'6 and –11. This is day 1 and day 2. Now Day 3 we expect countertrend reaction, but it is scalp ONLY on the buy side. This is where Scalp ONLY mode comes into play - you would not carry any LONG position home overnight on Day three in the beans after 2 high minus readings.

Sure enough, the CLOSE on 4/30, which was day 3, was not exceeded the following day and the market continued lower.

Now, we also had pattern to the PLUS side in the beans. Last Wed and Thursday, we had +9'2 and +10'6…2 STRONG readings. Friday was Day 3...we corrected back down, but you would not have held any short positions overnight.

This dovetails with the Taylor method. So, you can see that this morning we gapped HIGHER.

You will see this pattern over and over in many different markets. And perversely, this is what we had coming in this morning on the NAZDAQ futures. We had two strong back-to-back readings Wed and Thursday in the 2-period ROC…115.5 and 90.0. Friday's close was day three...we closed on the lows but gap HIGHER this morning. Funky little pattern.

There was a good question that was posed regarding the Big Day pattern. How do you quantify what constitutes a big reading...is there a specific look back period?

Moore Research center did the testing on this at the time, and we used a fixed amount in the SP’s...this was many years ago, and was not proper way to do it as we should have used ATR function.

The best answer that I can give is that by posting these numbers by hand, you are quickly able to see what constitutes a big reading. And you can have a Big Day pattern and then have another one 2 weeks later.
.

We talked a bit about Big DAY pattern in the Bonds the other day......today would be Day 5 on the pattern, and any shorts would be exited on close....usually do not get much downside follow through after Day 5. We exited earlier after morning number came out, but there was also Big Day pattern in the Nazdaq futures... model says that the market should close near the upper end of its range but NOT to take longs home overnight.

That is always a good way to think about your "roadmap" for the market's action....do you want to be long or short in the afternoon, playing for overnight follow through or do you want to exit on the close or by the close.

On a trend day, or when in breakout mode.....we usually exit by the close. There is not much play after trend day. When we look at the 2-period ROC, this is often what we are looking at, in one day on fresh flip, close out the position the following day.