News for hackers. Stuff that matters.
Re:Bitcoin drops below $10K after three days of cryptocurren... by amelius on Wednesday January 17, 2018 @09:46AM (#16168232)
Sigh. Those figures, why doesn't the Y-axis start at zero?
Re:Bitcoin drops below $10K after three days of cryptocurren... by philipodonnell on Wednesday January 17, 2018 @10:09AM (#16168480)
The chart does include 'zero' for its intended audience, just not the zero that you are used to looking at.

Financial types are only concerned with the % change. 'Zero' in the financial world means 0% change, i.e., the same price as it started with. A chart with a flexible axis will look identical when evaluated as a % change or as an absolute value, giving it a tremendous amount of utility when looking at the price action relative to various other things, indexes, pairs, sectors, etc...

Re:Bitcoin drops below $10K after three days of cryptocurren... by amelius on Wednesday January 17, 2018 @10:34AM (#16168705)
> Financial types are only concerned with the % change.

Yes, that's the point. If you let the Y-axis begin at zero, then you can intuitively read the change-percentages right from the graph!

Let's say something is worth $10, and tomorrow it's $11. That's a 10% increase in a day. Now suppose you let the graph start at $10 instead of at $0. You only see the $1 increase, and you have no clue about the change in percentage.

Re:Bitcoin drops below $10K after three days of cryptocurren... by philipodonnell on Wednesday January 17, 2018 @10:46AM (#16168810)
That's because no one looks at a chart with only two data points. Once you have a few hundred, you are less trying to determine individual percentage moves and more trying to determine the percentage change _relative to the other percentage changes on the chart_, which are obscured at best, and constantly changing as you compare with different things at worst.

You don't have to take my word for it. Every candlestick chart you will ever see uses a flexible axis unless the time series is expanded to include a period of time when it was at zero.

And its not because they never read Tufte. :-)

Re:Bitcoin drops below $10K after three days of cryptocurren... by amelius on Wednesday January 17, 2018 @10:54AM (#16168870)
If you read the text of the article, you see ALL the percentages referring to the total, not to the changes.

For example:

> Yesterday, Bitcoin was just a few dollars away from dropping below $10,000, but it bounced back around $11k, a 15% drop from two days before.

That 15% refers to the total of around $10,000 and not to some change!

The text is full of these examples.

(The fact that those candle-stick diagrams don't start at zero is because traders use technical analysis which indeed doesn't care about absolute values, but imho this shows how crazy it all is.)

Re:Bitcoin drops below $10K after three days of cryptocurren... by philipodonnell on Wednesday January 17, 2018 @11:03AM (#16168958)
Not disagreeing with you about the utility of technical analysis, and not saying that was the best chart selection to illustrate the author's point, just pointing out that the reason the author didn't start at zero is not the same reason most charts don't start at zero (the chart person doesn't know the rules).

A metaphor might be if you listen in on a conversation between two experts and they are using jargon that you know is technically imprecise. If you say "that's not really what that means" then the response is usually "bugger off because we know what it means and there's a good reason we say it like that and we're not changing it just because you are listening."

Re:Bitcoin drops below $10K after three days of cryptocurren... by amelius on Wednesday January 17, 2018 @11:21AM (#16169129)
> there's a good reason we say it like that

But could you explain that in this particular case?

For instance, let's say bitcoin is at $10,000 and it moves a few cents up and down, giving big swings in your non-zero-based graph. Why are those big swings saying anything meaningful to a financial type?

In my graph, the line would be almost flat, and clearly indicates that nothing is really happening (low volatility).

Re:Bitcoin drops below $10K after three days of cryptocurren... by philipodonnell on Wednesday January 17, 2018 @02:00PM (#16170675)
> But could you explain that in this particular case?

Sure, its helps me to think through these things too. :-)

In your graph, you probably don't mind that its almost flat, but if I am someone who makes money by trading every day, I'm interested in very small moves, a couple of pennies, maybe 0.5% over the course of a day. And that's _all_ I'm interested in. Where is the price over the last few days and how does that compare to the price of some other thing?

Knowing the relationship of the current price to zero is really not important because it a) has never been at zero, b) is not going to go to zero, and c) zero is meaningless in pricing anyway.

That last one is kind of unintuitive. You can't buy or sell a share at a $0 price. A $0 price implies that there are no buyers or sellers. No buyers or sellers means means the price is undefined, not $0. No matter how low the price gets, it can never be assumed to trade at $0. No stock has traded at a $0 price (even in bankruptcy there are always some value to the shares right up until they are liquidated and cease to exist). No stock has ever opened at $0, they IPO in the $20-30 range (or somewhere non-zero anyway).

So if zero doesn't mean anything, then you have to say, what should the minimum be? Well, you can say $0.01, but that's just as arbitrary. Assets are infinitely divisible in theory, and lots of trades are made at lower denomination that the currency, take the $0.000000008 / share SEC fee.

If any choice is going to be arbitrary, then you have to go back to the intention, which is to compare current price movements to recent price movements to understand what's happening, so just make sure your chart has all the recent price movements and then add +/- something like 10% on either side to give some padding and emphasize breakouts... a.k.a a flexible axis.

In this particular case, I suppose it is the author's intention. If they wanted to emphasize the volatility of recent movements, a flexible axis is right for that. If they wanted to minimize the volatility, a fixed axis starting at $0 is better.

But, objectively, there is no reason to _insist_ on starting at $0.

Re:Bitcoin drops below $10K after three days of cryptocurren... by amelius on Wednesday January 17, 2018 @03:16PM (#16171378)
> maybe 0.5% over the course of a day

But how does the trader know it's (on the order of) 0.5% by looking at the graph, if it doesn't start at zero? Now they have to look at the absolute value on the Y-axis, estimate the fluctuation in the graph and divide.

PS: appreciate your lengthy reply.

Re:Bitcoin drops below $10K after three days of cryptocurren... by philipodonnell on Wednesday January 17, 2018 @07:08PM (#16173347)
There's usually an option to just convert the axis from absolute $$ to % change from the earliest data point.

Plus, most professionals aren't going to ever look at one particular stock, like an Apple trader will have Apple, a tech sector ETF and the S&P500. As soon as you add a second series the y-axis _has_ to revert to a percentage change. They'll all start at the same point, which will be labeled 0%, and then diverge.

Re:Bitcoin drops below $10K after three days of cryptocurren... by take4 on Wednesday January 17, 2018 @09:50AM (#16168275)
You want a 1 day chart to start Y axis at zero?
Re:Bitcoin drops below $10K after three days of cryptocurren... by LandoCalrissian on Wednesday January 17, 2018 @10:01AM (#16168387)
What other financial investments do this? If the the stock market crashed tomorrow your not going to see people saying "We'll if you invested in '29 you would have been rich!"
Re:Bitcoin drops below $10K after three days of cryptocurren... by oreo81 on Wednesday January 17, 2018 @10:32AM (#16168689)
They said it in 2008...
Re:Bitcoin drops below $10K after three days of cryptocurren... by OzzyB on Wednesday January 17, 2018 @09:49AM (#16168262)
Maybe because not everyone got in at 10¢ and the drop is extremely significant for those that got in recently?
Re:Bitcoin drops below $10K after three days of cryptocurren... by amenod on Wednesday January 17, 2018 @09:54AM (#16168309)
That doesn't matter. Most people are (well, should be) interested in relative corrections, which these charts don't depict. It makes volatility look even bigger than it is.
Re:Bitcoin drops below $10K after three days of cryptocurren... by amenod on Wednesday January 17, 2018 @09:52AM (#16168296)
This! It seems all of the charts start Y axis at lowest value in timeframe, which is simply deceptive. For investors there is no difference if some currency rises from $10.000 to $11.000 or from $1000 to $1100. But there is huge difference between rising from $1000 to $1100 and $1000 to $1001 - which looks the same on these charts. </rant>
Re:Bitcoin drops below $10K after three days of cryptocurren... by qubex on Wednesday January 17, 2018 @09:50AM (#16168273)
Either because the infographic designers haven’t read what The Visual Display of Quantitative Information about accurately depicting time-series, or because they have read The Visual Display of Quantitative Information and are abusing its dicta to make graphs that are as dramatic as possible.
Re:Bitcoin drops below $10K after three days of cryptocurren... by learnstats2 on Wednesday January 17, 2018 @10:05AM (#16168430)
You seem to have understood close to the opposite of the author's intent:

"In general, in a time-series, use a baseline that shows the data not the zero point." - Edward Tufte.

Re:Bitcoin drops below $10K after three days of cryptocurren... by qubex on Wednesday January 17, 2018 @10:12AM (#16168524)
I'm delighted to see you're well-acquainted with the source. I'm absolutely certain he advocates having some kind of evident glyph in the Y axis to indicate discontinuity and the nonzero baseline, I think it was the classic // mark or something akin to it. Am I wrong? That's what I meant by abusing its dicta to make dramatic graphs.
Re:Bitcoin drops below $10K after three days of cryptocurren... by philipodonnell on Wednesday January 17, 2018 @10:00AM (#16168381)
I was at a visualization meetup last night and one of the attendees said something about blowing a sparkline up really large on the page and adding numbers to see it better and I quipped about Tufte rolling in his grave.

People don't read these days. :-(

Re:Bitcoin drops below $10K after three days of cryptocurren... by qubex on Wednesday January 17, 2018 @10:05AM (#16168432)
Apart from any other consideration, as far as I know Edward Tufte is alive and well.
Re:Bitcoin drops below $10K after three days of cryptocurren... by philipodonnell on Wednesday January 17, 2018 @10:13AM (#16168527)
I know, but no one got that joke either.
Re:Bitcoin drops below $10K after three days of cryptocurren... by qubex on Wednesday January 17, 2018 @10:17AM (#16168558)
Well played.
Re:Bitcoin drops below $10K after three days of cryptocurren... by philipodonnell on Wednesday January 17, 2018 @10:20AM (#16168586)
Ha, sorry, unintentionally in your case. My humor does not always translate to the written word. :-)
Re:Bitcoin drops below $10K after three days of cryptocurren... by qubex on Wednesday January 17, 2018 @10:30AM (#16168676)
The most hilarious aspect of this is that some irritable ignoramus has down-voted my “well played” comment. I'm beside myself with laughter.
Re:Bitcoin drops below $10K after three days of cryptocurren... by philipodonnell on Wednesday January 17, 2018 @10:33AM (#16168698)
I upvoted your back. And well played right back at ya.
Re:Bitcoin drops below $10K after three days of cryptocurren... by Retric on Wednesday January 17, 2018 @09:57AM (#16168342)
For a general audience starting at zero is almost always the correct choice, but many exceptions exist such as temperature ignoring 0 kelvin.

It's a detail trade-off. Starting at zero helps the uninformed, but will obscure details.

Re:Bitcoin drops below $10K after three days of cryptocurren... by philbarr on Wednesday January 17, 2018 @10:02AM (#16168394)
> Starting at zero helps the uniformed, but would obscure the bonce back above 10,000

Starting at zero helps the well dressed, but would obscure the person's head back above 10,000?