Elliott Wave Analysis

[bitcoinBull]

Wow.  Epic thread.

I count Netrin making three stunning calls:

1)  On Oct 26 he called a ~$4 top.  The price popped then turned down on Oct 29.

2)  On Nov 12 he predicted predicted a crash from ~$3.20.  Price fell from $3.00 to $2.10 two days later on Nov 14.

3)  On Nov 16 called a top at $2.6.  Price had already started declining.

Someone else should double-check the timestamps to verify these weren't made after the fact.

[bitcoinBull]

I have doubts that we're hitting a top around $3, before descending to test the lows.  My non-Elliott reasoning leads me to bet that, this time, the trend has changed.  The bid walls, trade volume vs. bitcoin supply, and a hunch that it will be easier for the heavy hitters to bank on a $2 reversal and push it upwards from here rather than downwards).

"Netrin, you've presented three different variations, but all of them result in a downward arrow. Why are you so bearish?"

I have to reiterate this.  I see that you're charting a triple three combined correction (WXYXZ).  I suppose because you're aiming for that $1 pricepoint where "A correction usually finds its lowest point in the area of the fourth wave of the preceding impulse wave of the same degree."

Though I'm betting against it, if this rally peaks in the $2.8-$3.45, it would retro-actively suggest the "Swannell wave" count I shared last week. It would imply that the previous $3 sideways correction were a fourth wave, the drop a fifth wave, and our current rally a correction of a higher degree. Anything above $3.45 (before testing and possibly breaking below $2) invalidates all of my counts for the past month or two (reversal, month long triangle, downward impulsive failure).

I attempted an alternate, which counts a reversal at $2.  I believe it still honors your past counts, because most of the waves coincide (impulse to impulse, corrective to corrective).  I get a double three (WXY) rather than a triple three, by counting the first drop to $2.04 on Oct 19-20 as the third wave down, rather than the fifth and final.  This way the third downward wave is the longest ("whereas upward fifth waves tend to be longest in commodities it is less true on down trends, where third waves tend to be longest for all asset classes.")

The alternation looks good this way too (corrective wave II is sideways, and IV is sharp) [13.4 Principle of Alternation].

The fifth wave down just extends only $0.10 past the third (to the low of $1.9449), but I believe that's valid.  Notice any violations?  It's not an ideal shape, but I hope it does the job.

You could probably come up with something prettier that doesn't test $2 again.

[netrin]

1)  On Oct 26 he called a ~$4 top.  The price popped then turned down on Oct 29.

2)  On Nov 12 he predicted predicted a crash from ~$3.20.  Price fell from $3.00 to $2.10 two days later on Nov 14.

3)  On Nov 16 called a top at $2.6.  Price had already started declining.

I like to think of these as all part of the same form (from 26 to what I call the 30 October count):







However, I was quite wrong about a late November peak at around $2.8:



Which unambiguously invalidated the 30 October count, when prices surged above $3 just recently, leaving only the "Swannell wave" as a valid count:



BitcoinBull, you raise excellent points. I had been quite happy this north-summer with the single zig-zag since the June peak, but S3052 pointed out some unavoidable errors in my count. I am open to a count such as your own, as well as a double zig-zag followed by some bottoming out like a triangle.

My only-human sentiment sensor tells me people already feel euphoria. As if we're in a new first wave (III.i), but I'm not sure how to count that, as I'm very confident in the October count (cyan a-b-c and its previous downward impulse). Could mid-November represent a failure (your C.v.5)? It would certainly explain my own egregious counts in the past few weeks.

I'll look further into higher scale counts. Thanks for your posts!

[netrin]

Whether or not we've already reached a peak at $3.14, it's worth noting for yourself, whether the downward motion is three corrective (or variation) or five impulsive waves and on what scale. Elliott doesn't just show what's likely to come, but provides numerous methods of validation (and failure) along the way.



For example, at the time of this writing (2 Dec, 01:40 UTC), I see three waves down, followed by a single rising wave up that though still below $3.14, would invalidate a previous impulsive three wave downward count. We could be in the B wave of a sideways (flat or triangle) correction or we might still be in the first wave of a large downward impulse. It's just too early to say.

[ElectricMucus]

I don't think you can see patterns on this timeframe. The 'noise' (bid-ask-spread) is a good fraction as large as all the actual price movement.

Too much are the result of individual decisions, you don't breathe after an Elliott Wave do you? 

[netrin]

I tend to agree about the noise on this timeframe. I don't generally even post the 'green' scale but I do count even lower because confirmation on subwaves gives strength to counts on higher scales. I hadn't thought of it, but your comment about spread sounds right; The prices in the spread are just a vacuum of quantum noise. But what's that about Elliott breathing?

Scale	Impulsive	Corrective
-------------------------	-------------------------	-------------------------
Yearly	1 2 3 4 5	a b c
Seasonal	I II III IV V	A B C
Monthly	i ii iii iv v	a b c
Weekly	1 2 3 4 5	a b c
Daily	I II III IV V	A B C
5-Hourly	i ii iii iv v	a b c
Hourly	1 2 3 4 5	a b c
15 Minutely	I II III IV V	A B C
4 Minutely	i ii iii iv v	a b c
Minutely	1 2 3 4 5	a b c

[netrin]

In the impulse down from $3.8, 29 October to $2, 19 November, using the Swannell count, we may have seen a fifth wave failure (ending triangle) $3.1, 12 November to $2, 19 November.

Modifying the mid-November count slightly, we can count an extended third from $2.2, 21 November to $3.14, 1 December, beginning a fourth triangle or flat (A $2.99 late 1 December, B $3.14 early 2 December, perhaps C later today), or more likely, a fifth wave failure, followed by a drop.

Though I'm betting against it, if this rally peaks in the $2.8-$3.45, it would retro-actively suggest the "Swannell wave" count I shared last week. It would imply that the previous $3 sideways correction were a fourth wave, the drop a fifth wave, and our current rally a correction of a higher degree. Anything above $3.45 (before testing and possibly breaking below $2) invalidates all of my counts for the past month or two (reversal, month long triangle, downward impulsive failure).

I attempted an alternate, which counts a reversal at $2.  I believe it still honors your past counts, because most of the waves coincide (impulse to impulse, corrective to corrective).  I get a double three (WXY) rather than a triple three, by counting the first drop to $2.04 on Oct 19-20 as the third wave down, rather than the fifth and final.  This way the third downward wave is the longest ("whereas upward fifth waves tend to be longest in commodities it is less true on down trends, where third waves tend to be longest for all asset classes.")

The alternation looks good this way too (corrective wave II is sideways, and IV is sharp) [13.4 Principle of Alternation].

The fifth wave down just extends only $0.10 past the third (to the low of $1.9449), but I believe that's valid.  Notice any violations?  It's not an ideal shape, but I hope it does the job.

You could probably come up with something prettier that doesn't test $2 again.

Thanks again BitcoinBull for your counts. It's given me much to think about. We agree on the W count, and your Y count is good, though I am not fond of the 'look' of your zig-zag, specifically that Y.c so dwarfs Y.a. Your black IV (my red y-x) seems to be ignored on the higher scale. However, your Y.c count is solid, both black and cyan. Your fifth wave ("only $0.10 past the third") including a sub-fifth wave failure (adjusting your mid-November magenta count) would strongly suggest a reversal.

Alternatively, I'm looking at my final third as a potential ending triangle with a $2 base and many more large bounces (3-3-3-3-3). Three waves from $3.8, 29 October down to $2 might be a cleaner fit than my numerous attempts at five waves (30 October count, Swannell, et all).

[netrin]

This count is strictly valid (at least as far as the numbers labeled), but what is valid on these tiny scales, with huge spreads and light volume? Magenta iii ($3.14, 1 Dec) might have been the peak as in my previous green count (and would be be re-labeled II.z.C.v.5), otherwise we can expect a small jump up for the real magenta v (anything over $3.14) before returning to $2.9 and likely falling lower.

The price drop to $2.9 had not been very informative. It falls nicely within the previous suggested fourth waves (green and gold (and magenta) bars). So either way, if we do see a new v peak or prices immediate collapse under $2.9, we can expect the price to first correct to about this same price territory. A price below $2.65 at any time would seriously discourage hope of a rally, continue the monthly downward trend, and I would expect prices to drop to $2 and perhaps below.

[meelba]

Fake Volumes on bitcoincharts.com

this is not the first time....

[netrin]

@Meelba: I did not watch it live, but 40K was roughly the size of the bid wall at $2.9. If you are presenting a conspiracy theory, it's certainly interesting and I'd like to discuss it elsewhere, but I'm afraid it's completely off topic here (Elliott waves).


The latest drop certainly looks impulsive. The first wave ($3.13 - $2.9) is roughly the same size as the third ($2.99 - $2.75). Therefore, one can expect either the third or the fifth to extend much lower. Of course, the drop may have only been corrective (perfect zig-zag), and then our fourth wave is really the beginning of a rally to new heights. That's also a reasonable position because even after three waves, bitcoin is still in the price territory of the fourth wave of the previous impulse of the same degree; In other words bitcoin has not yet dropped below the green bar. However, the pattern should be familiar to any bitcoin trading veteran, sharp second wave, a long sideways fourth triangle, followed by a collapse. The gamble is yours, but if it's bed time, you might want to place your stops and orders below $2.65 or above $2.98, depending on your bet.

On bigger scales, I have counted a major expanded 3-3-5 flat correction since ($2.1, 14 Nov): A ($2.6, 16 Nov), B ($2, 18 Nov), C ($3.14, 3 Dec) although wave C, while certainly impulsive, only counts three strictly valid waves (extended third beginning ($2.17, 21 Nov). A drop below the magenta wave i ($2.5, 20 Nov) without a progressive correction would paradoxically invalidate yesterday's valid count, but honestly, I'd just accept the invalid count from earlier in the week and call $3.14 the fifth wave peak since 14 Nov.

[netrin]

Long bulls, who believe we are currently experiencing a correction in a major progression since $2, must accept that the drop from $3.13 yesterday displays an impulsive five waves. As of this morning (4 Dec), the fifth wave is truncated (the subwaves during the same day overlap). This implies either an extension down or reversal. None the less, zig-zags are the only corrective waves that begin with five subwaves. I estimate a B wave up to $2.9 (a half retracement to a very well established resistance), followed by another roughly equal five waves down, to about $2.4, certainly below the wave A low of $2.61.

Unfortunately, it's not a good time to make a definitive case for either the bulls or bears going forward. The bulls can make a good case for a truncated fifth and bottom ($2, 14-18 Nov), an impulsive first wave ($3.14, 2 Dec), and an on-going zig-zag second wave correction this week. Only a price below $2 would truly invalidate this scenario. A rise above $2.9 without first crossing below $2.61 would be a rally signal.

Rather than a zig-zag, bears would expect an extra two waves down. The A-B-C of the zig-zag would be indistinguishable from the 1-2-3 of the regressive impulse. Regardless of your labeling, both of these counts carry the same short term message (5 down, 3 wave correction, 5 down) and should be profitably traded. However, the wave that follows would be very different. The bulls would expect an impressive rally, while the bears would expect a small correction followed by another low toward and likely below $2.

[netrin]

Mt. Gox hit exactly $2.9. Unfortunately other markets came nowhere near that level. I expect prices to fall now below $2.5 in five waves.



This chart does not add much in terms of predictions, only that the waves have so nicely fit within the suggestive bars thus far. If the price falls below the magenta bar $2.5, most likely completing a zig-zag, then the impulsive magenta rally (since B) has failed. The black scale does not technically fail until a drop under $2, but a drop below the first magenta wave (i) leaves absolutely no other valid count.

A rise above the gold bar at $3.14 would confirm the continuation of the magenta fifth wave. However, given the extended third, magenta waves i and v should be about equal before a correction. That doesn't leave much room for v to rise above iii. There's much reason to believe the C.v already peaked at $3.14 even if I can't draw a valid count. The drop we've seen to about $2.6 already looks sufficiently corrective.

Zig-zag B wave:

• subdivides into zig-zag, flat, triangle, or combination thereof
• never moves beyond the start of wave A ($3.14)
• typically retraces 38-79% of wave A ($2.81-$3.03)
• if running triangle, will retrace 10-40% of wave A ($2.67-$2.82)
• if zig-zag, will retrace 50-79% of wave A ($2.88-$3.03)
• if triangle, will retrace 38-50% of wave A ($2.81-$2.88)

Zig-zag C wave:

• always subdivides into an impulse or diagonal
• almost always subdivides into an impulse
• about the same length as A
• almost always ends beyond the length of wave A ($2.61)
• the ends of A-C are often parallel to the starts of A-B

[netrin]

Pretcher's 200 year chart of the US economy published in 1977 (I'm not sure how the data was created) merged with 100 years of Dow Jones (source Wikipedia). The charts don't line up perfectly. Sorry.

• 1929 - (III)
• 1932 - (IV)
• 1935-ish - I
• 1942 - II
• 1955-ish - III
• 1974 - IV
• today - (V).V (peak)

[netrin]

Zig-zag C wave:

✔	always subdivides into an impulse or diagonal
✔	almost always subdivides into an impulse (*)
✘	about the same length as A (*)
✔	almost always ends beyond the length of wave A
✔	the ends of A-C are often parallel to the starts of A-B (*)

The above count fulfills the first requirement and four of five guidelines (there may be more). However, I think one could more easily count the 'c' impulse (five subwaves) to mid-October, which would fulfill all five guidelines. I think one must ask oneself "did I create the count because I want it to be true?"

I am unable to provide anything more convincing. Late October produced a picture perfect a-b-c zig-zag. I'm still not happy about my counts in early November from $3.8 to $2 and am entertaining the idea of three waves. The rise back up, peaking early December at $3.14 only counts three waves. Thus far, it is ambiguous.

I would not be surprised if bitcoin bounces between $2 and $3 a few more times before... I don't know.

It is entirely possible that the drop this past week to $2.6 is the fourth wave followed by a future fifth to new heights, precisely as I've counted the magenta and green waves earlier. The problem is that I either count five gold subwaves down from $3.14 to $2.61 or a zig-zag with a double-length c-wave which is most uncommon. Assuming the drop is five subwaves and bitcoin is in the fourth of five wave rally, then I'd expect another c-wave drop before rising on the fifth wave. Alternatively, assuming the same five subwaves, but that bitcoin is still correcting, then I'd expect the same drop not to return above $2.6 for a long time.



$2.5 is the deciding line (below the lower magenta bar). A symmetrical zig-zag assumes but does not guarantee a 'c' wave below $2.5 just as a third impulsive wave is nearly always longer than the first wave. However a drop below $2.5 would invalidate the magenta count since black B ($1.99, 20 Nov) - invalidating hopes of a rally, signally a return to $2 or below.

The dark channel vaguely represents a possible 'c' exit point. If a zig-zag is confirmed then the price will likely end within the dark channel near the bottom of the magenta bar. Green 'b' may rise up to but the zig-zag must not cross above $3.14. This would make room to complete a prettier (though flatter) sub-zig-zag iv.b.C (unlabelled gold C wave since 6 December, within green a-b since 4 December, within magenta iv since $3.14, 2 Dec). The higher green 'b' travels, the flatter the zig-zag will likely become, providing sufficient space for green 'c' to bottom out in the lower magenta bar, safely above $2.5.

A price breach above $3.14 magenta would invalidate the green December count. Green 'a' would be relabelled magenta iv as before. Green 'b' would be relabeled somewhere in the third wave, and we'd expect a magenta 'v' peak well within $3's but not likely to reach $4 without a grand correction. I expect that simply crossing above $3.14 could trigger short stop loss thus igniting an extended fifth wave up.

[dancupid]

I think the psychology of Elliot waves is fascinating - it's a black art like reading tarot cards. This is not a criticism - it seems to be a device that allows the natural pattern recognition abilities of the human mind to be harnessed - but maybe there is too much room for projection of subconscious desires. If it could be truly automated than maybe we could test it properly, but perhaps that misses the whole point of it.

[netrin]

Swannell has automated it and produced some nice statistics as well as challenged some of Elliott's original rules and guidelines. (Look back a few pages, I have a link to his 2003 paper). Pretcher has also refined Elliott's work. I think we generally consider Pretcher, Frost definitive.

It is a black art yes. One should be on guard against creating charts based on personal desire to prove an extant position or to create unnecessary doubt. Elliott does provide confirmations and invalidation, which are excellent for pulling out of a losing position with minial loss. In my own short experience, it's been very useful to analyze patterns greater and smaller than the scales I actually intend to trade. This is true with any methodology, but Elliott requires the scaled view and nicely ties multiple scales into the same forecast, which I have not found to be true of other analytic methods.

[bitcoinBull]

I don't think you can see patterns on this timeframe. The 'noise' (bid-ask-spread) is a good fraction as large as all the actual price movement.

Too much are the result of individual decisions, you don't breathe after an Elliott Wave do you? 

I have to agree with ElectricMucus.  Because Elliott Waves are principles of mass psychology, you wouldn't expect them to work on the smaller timescales.  Mass psychology tends to change on a day-to-day or year-to-year basis, not minute-to-minute and hour-to-hour.

While the fractal nature of Elliott Waves is appealing, I doubt it scales more than a few orders of magnitude either way (bigger or smaller). 

This question is especially relevant to the 200-year dow chart you posted above.  We're supposed to be at the peak of the fifth wave of a supercycle or a grand-supercycle, but it doesn't appear we've yet seen the first (impulsive) reversal wave down.  Its mostly bounced back from the 2008 crash (mortgage crisis) and the 2011 one in October (euro crisis).  Maybe IF the China "bubble" pops, that would be the tipping point.

I'm not sure I buy Elliott Waves on such a long timescale (again, time long-past should have little effect on mass psychology). 

If we are at the beginning of a fifth wave down, I probably shouldn't be considering opening a short position on TLT, as cypherdoc recommended in his gold thread.

[netrin]

Each move within the spread is a single act just as easily toward the ask or bid. As unpredictable as quantum noise within a vacuum. Trends are generally easier to read on each greater scale assuming we could stand far enough away. I have no reason to believe this would not include the expansion of the universe if we sat in God's private theatre booth.

I may have doubts about the data but fewer about the theory. I am skeptical connecting the pre-industrial economy through the Dow Jones Industrial Average and I do not know how to reconcile the Mt. Gox hack and June sell off.

Each bitcoin day has been informative in general and each week quite tradable. I hesitate to label overlapping waves within a single day but can most often discern distinct subwaves of entire waves within a week's duration. Fortunately resolution is enhanced with volatility not strictly with time.

While the fractal nature of Elliott Waves is appealing, I doubt it scales more than a few orders of magnitude either way (bigger or smaller).

If I shared your doubt I would not be counting Elliott at all. The greater the number of scales in a lattice of confirmation, the greater my confidence in a count. I think you may be asking too much to call the day of the multi-century peak, but the decade? Well, I won't deny stocking up on dried food and bullets. The stock markets are unusually volatile. Just as bitcoin is more difficult to predict at a major bottom, the global markets are difficult to predict at the top.

I don't know if Cypherdoc is correct, but his argumentation certainly opened my eyes. I think it's well worth everyone's read.

[netrin]

As an example of uncertainty not typically seen on greater scales:



DOWN5: ($3.14, 2 Dec) to ($2.61, 4 Dec) counts five clean waves.

DOWN3: ($3.14, 2 Dec) to ($2.61, 4 Dec) counts three wave zig-zag. Some vague impulse ($3.14) to ($2.9, 3 Dec), correction ($2.99), and five invalid waves down (the fourth wave ($2.69-$2.8, 4 Dec) enters the price territory of one and two)

UP3: extended three waves. The first wave ($2.61-$2.9, 4 Dec) is followed by a correction to the previous fourth ($2.72, 4 Dec). First subwave ($2.93, 5 Dec), sideways correction (~$2.88, 6 Dec), third subwave ($3.05, 6 Dec), followed by a multi-day sideways correction ($3 +/- 3%, 6-8+ Dec), and a fifth subwave up, thus finishing the extended third wave.

UP5: five wave rally. Same as UP3, with a fifth impulsive wave breaching well above $3.14.

✔✔	DOWN5 and UP3: Both are part of an abnormal flattened 5-3-5 zig-zag correction. A drop in five waves to ~$2.6 can be expected
✔	DOWN3 and UP3: Both are part of a flat 3-3-3 correction. A drop in three waves to ~$2.6 can be expected (same message as above)
✘	DOWN5 and UP5: These can not both be five waves. One is wrong
✔	DOWN3 and UP5: Confirmation of a major rally since a $2 bottom. After the fifth well over $3.14, we can expect a large second wave correction of a higher scale back into the high $2's (this suggested fourth $2.61-$3.14). An impressive third wave rally would definitively confirm a major reversal since June rally since October

Bigger scales:

Since June, bitcoin has trended down with unambiguous impulsive waves until recently hitting ($2, 19 Oct). 19 October began a perfect zig-zag correction to ($3.8, 29 Oct) and continued down from there. I anticipated five waves, a correction mid November and a continuation of the trend below $2. That never happened and my counts fell apart.



Suppose we never completed the major correction since October? There is no mistaking the late October zig-zag, but that ($3.8, 29 Oct) zig-zag could represent only the first 'a' wave of a 3-3-5 flat (or 3-3-3-3-3 bounded triangle). The 'b' ($2, 19 Nov) is followed by the 'c' wave we experience today (7 Dec). Even a rise back up to $3.8 in five waves would not invalidate the flat. In fact, according to Pretcher, wave 'c' is most often 100-165% the height of 'a' ($3.8-$5). And what follows would be a continuation of the trend, a fall below $2, into the suggested price range of the March iv, about $1.

[netrin]

UP3: extended three waves. The first wave ($2.61-$2.9, 4 Dec) is followed by a correction to the previous fourth ($2.72, 4 Dec). First subwave ($2.93, 5 Dec), sideways correction (~$2.88, 6 Dec), third subwave ($3.05, 6 Dec), followed by a multi-day sideways correction ($3 +/- 3%, 6-8+ Dec), and a fifth subwave up, thus finishing the extended third wave.

This count continues to 'look good' but has been invalidated in so many ways that one would be unwise to bet heavily on it. While the fourth wave (~$3, 6 Dec thru present) looks like a double zig-zag, the drop to $2.9 moments ago invalidates the entire count (again) as it enters the price territory of the first ($2.93, 5 Dec) and second waves.

Alternatively, we could count the entire rally ($2.61 4 Dec) to ($3.05, 6 Dec) as a single wave, though we'd have trouble (it's impossible) counting five subwaves. The sideways correction since would be a legal second wave, though one would have expected a sharp second wave (see the bitcoin yearly chart for perfect examples: (ii, Nov 2010) and (II since June)).

So, again while this little rally 'looks good', it is not Elliott, and likely wishful thinking. The chart at this scale is at best only informative but at worst it's random noise. Whether up or down, the breakout is likely to be 15% or more.