Great to engage with you in this meaty discussion.
Thanks for joining in! I need someone with experience to keep me in check.
![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif)
I agree with you on points 1)+2), not exactly on rules 3) + 4). Overall, this is an art as you are indicating and no one does this exactly the same.
I should change one thing I said: These are my rules, not The Rules. The market would shake to pieces if everyone started using exactly the same formula everywhere.
![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif)
On rule 3) and 4) Many trendlines, including the 40 year log DJI chart, even work long time after the trendlines have been touched. I have many more examples.
I should clarify 3: I don't have a problem with drawing trendlines out way past their defining points, even decades. I DO care about the ratio of (distance we extrapolate) : (distance between the points). What ratio is acceptable depends on what you're charting, how many points you use, and if there are any changes in the fundamentals.
DJI is a massive index and trends will be valid as long as there are no extreme shifts in national policy. You can extrapolate out 2:1 unless the US decides to dissolve itself and the states are joining the EU. More than that and you have to look at the long-term economy, but it still has some predictive power. This is silly to compare to BTC.
Let's try something more speculative and volatile: Yahoo.
![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FU7OaL.png&t=665&c=YHE-X8mMPRA7Tw)
Let me introduce Rule 5: Trendlines come in pairs. If you can't see the channel, there isn't a channel.
The red lines are how I see it. Bubble, blowoff, normal, good times, losing our way, "I'm not dead yet!". Clear, tightly-defined, multi-year trends.
The purple line is analogous to the lines in the OP: we're taking two points and trying to predict the future despite the dot com bubble, changes of the company's direction and competition, and other significant outside effects happening in there. Do two points in 1997-2002 really predict that Yahoo would blow it on earnings in 2008Q3? If that bottom in 2002 was higher and we had 3 neat points in 1997-2001-2002, and we were watching this develop, should we be expecting Yahoo to screw up in 2007 instead? I think the purple line is silly, and all the things that happened to Yahoo over the years made it completely irrelevant.
The green lines are creepy. It's a decade-long triangle! I think they're meaningless, but they fit surprisingly well.
The way you have drawn the trendlines is not really accurate.
(i) When drawing trendlines you have to always start at the lows and connecting these first. Try to find two significant lows and connect these. (ii) Then see whether you can find a 3rd or 4th touchpoint.
(iii) then draw a parallel trendline connecting highs. if this looks right, you have a great parallel trendchannel.
(iv) if a trendline is broken, then how will you see if this line is important? prices will accelerate the move if the line was really important.
That's how I do it! Look again:
![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FzwLmg.png&t=665&c=bWj89ZMCHcL00w)
I don't insist that the channel is perfectly parallel, though closer is better. I extrapolate backward when I can see that a trend started before the first top or bottom (foolish if I was trading on technicals, but I have the benefit of hindsight). And I draw channels a little narrower (notably August, again with hindsight benefit; and I clip wicks).
Aside from those style differences, do you see a problem with my lines? I think they're a pretty fair view given my assumption that trends are shorter and that more recent changes in fundamentals trump technicals extrapolated from last year.
You see this in the attached. And truly, this trendline is the most valid one, because
(a) prices have 4 or even 5 touchpoint lows
(b) the upper parallel trendline (green) also makes sense. bitcoin prices broke out of this line end April and is now back to that level
(c) prices held this line early october to Oct 15/16 and then
(d) at the EXACT time when this trendline was broken for the first time in more than a year, prices crashed -40% two days ago . This is why this line has been and still will be so crucial.
That's an awfully strong conclusion to draw from a trendline whose slope is defined by the handles of some rather tall hammers that were formed by some block dumps into thin depth and wide spreads... By what possible mechanism would those ephemeral patterns predict something so precisely that far in the future? Did you call this before, or are you too drawing lines with the benefit of hindsight?
You can draw the trend from several other nearby lows to reach other conclusions:
![](https://ip.bitcointalk.org/?u=https%3A%2F%2Fi.imgur.com%2FPNbev.png&t=665&c=U9rYWcxmObjQBQ)
Blue my way, purple your way, and orange showing the hazards of extrapolating long distances from long shadows, especially on log charts where brief blips of downside can throw the slope off.
And what one should never forget: trendlines are there to be held and to be broken! When discussing trendlines, one must be aware of both, chances to hold and break and act (trade) accordingly.
Now, looking into the future, this line will either be reclaimed fast, or it will provide strong resistance. At the minimum, the underside will likely be tested soon. This is currently around 3.7-4$.
Of course! Have a look at my original chart again. On my last trend the current crash is similar to the whole series over the last four months. Flat for a while until we start approaching the resistance line, crash, slight rebound, level out. It slightly breaks my trendline, but from the level-off, I'd say it's just going to regress back to the channel by holding the current price a little while.
I'm looking at the same things you are... I just do it from a background of fundamentals which are telling me the trends are shorter than you see them.