Jump to content

Talk:MACD

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Wiki Education Foundation-supported course assignment

[edit]

This article was the subject of a Wiki Education Foundation-supported course assignment, between 4 October 2021 and 9 December 2021. Further details are available on the course page. Student editor(s): Herry3999.

Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT (talk) 03:02, 17 January 2022 (UTC)[reply]

Swing traders

[edit]

MACD is no way liming traders. Spread traders, position traders, day traders of all sorts refer to MACD as a lagging indication of trend. GT

Signal line

[edit]

What the heck is a signal line? Please explain or link to an explanatory page. Bmunden 12:58, 28 June 2006 (UTC)[reply]

[edit]

The link to Ref#3 is broken. As outlined in the article MACD isn't lagging when it comes to divergent pattern signals. As far as I know it's still a very commonly used tool. Maybe need to revisit the Criticism paragraph. —Preceding unsigned comment added by 79.97.10.15 (talk) 15:45, 3 June 2009 (UTC)[reply]

weird —Preceding unsigned comment added by 203.24.1.137 (talk) 01:19, 10 March 2010 (UTC)[reply]

[edit]

On 29 June 2011, Jonny.allason Added a second External link, pointing to a Dig MACD page. This page does nothing to explain the MACD, and only serves as a location where the Dig MACD indicator may be purchased for a very specific trading platform (TradeStation). Unfortunately, still being somewhat new to Wikipedia, I am not completely versed in all the rules and policies, so my question would be: Is the External Links section an acceptable location to provide links to store sites that exist only to allow the purchase of a project. If the answer to this is 'no', I think the link should be removed. However, if the answer to this is 'yes', then I can think of several other MACD related indicators that are sold by several other companies for several other trading platforms. In fairness we should probably provide external links to all of their products as well. Zackron (talk) 21:31, 21 September 2011 (UTC)[reply]

The answer to your question is an emphatic NO. See the official guideline Wikipedia:External links. Links that are clearly promotional should be removed immediately and the editor who added them should be warned and eventually blocked if the abuse persists.
I had already warned Jonny.allason in July when he added a similar link to Tradestation but I missed the one he had added to this article at about the same time. It's gone now. ~Amatulić (talk) 00:02, 22 September 2011 (UTC)[reply]

Is MACD really a velocity?

[edit]

“In signal processing terms, the MACD is a filtered measure of velocity”

Is that correct? Given a constant trend, MACD equals zero constant (it does, but not zero, see below. Mclaudt), but not the velocity. Mclaudt (talk) 03:19, 1 December 2011 (UTC)[reply]

Seems correct to me. Given a constant trend, the difference between two moving averages will not be zero because each moving average has a different lag. The difference in lags is constant, so a constant trend will result in a constant proportional to the slope (velocity) of the trend. ~Amatulić (talk) 04:51, 1 December 2011 (UTC)[reply]
Yes, now I see that it is correct. I played with MACD in Excel, and it's obvious that at constant trend, MACD constant is not zero, but is defined by the trend slope. Thanks for explanation. Mclaudt (talk) 05:47, 1 December 2011 (UTC)[reply]
Yes. See the new reference in the section on signal processing for the exact relationships between MACD and the first derivative ("velocity"), and the histogram as a measure of the second derivative ("acceleration"). Gmstanley (talk) 21:30, 23 September 2012 (UTC)[reply]

Oscillator classification

[edit]

On my opinion this section is completely irrelevant. Plus statement "The MACD is an absolute price oscillator (APO), " is wrong. MACD consist of three elements (MACD Line, Signal Line and Histogram). APO on the other hand is one-line indicator which is calculated as MACD Line. Correct statement would be APO is one of the MACD's components. In any case, on my opinion, "Oscillator classification" section should not be in the MACD article.~Hghlight (talk) 9 May 2013 (UTC)

Signal processing theory

[edit]

the "Signal processing theory" section on my opinion is not general in the nature. It is very specific and id describes a specific technology. If this deserves to stay here then thousands of various trading strategies based on MACD and filters applied to MACD should be here as well which I do not think is wise.~Hghlight (talk) 9 May 2013 (UTC)

That section describes in mathematical terms what MACD really does - providing measures of first and second derivatives (with lags). The overall article talks about how to calculate and plot the MACD components, some notable events like line crossings, and rules traders have made up to use them. But it never really says what the resulting calculations mean. This section is the one that opens up understanding MACD to a much larger group of people than active stock traders - anyone who was even briefly exposed to calculus or physics. The section is not in any way a "specific technology" any different than MCAD. It's just providing a simple explanation of MACD does, from the much more general perspective of mathematics, rather than narrow perspective of a trader who uses it perhaps even without understanding what it really means.

Consider the plight of a typical person whose primary job is (or was) not involved in actively trading, who starts getting active in stock trading. He or she quickly discovers tools (whether a stock trading platform from a company like E*trade or Schwab, or from widely-used free sources like Yahoo finance) that offer MACD calculations or plots. They wonder "what is that, and what does it mean?". Wikipedia should answer that question for this general reader, and you can't get much simpler than saying the MACD calculations provide information on the derivative (rate of change/ velocity) and second derivative (acceleration) of the price. Without that basic understanding, any associated trading rules sound like mysterious mumbo-jumbo with no obvious basis.

Traders are free to invent trading rules based on these components, some of which are described in the article. The "Signal Processing Theory" section says nothing about these trading rules. It just explains what the basic MACD components measure. So there is no reason this section implies the need to bring in discussions of the "thousands of trading strategies" - that's a completely different thing, and inclusion of those strategies is an independent decision. As an analogy consider an explanation of a car speedometer and tachometer. You explain what they measure, which is the most basic thing. Independently, other sections can optionally describe strategies for using these tools for safety, winning car races or minimizing fuel consumption. Gmstanley (talk) 20:20, 12 September 2013 (UTC)[reply]

Timing

[edit]

in the "timing" section, a statement "One popular short-term set-up, for example, is the (5,35,5)." is very specific. MACD setting vary from one security to another security, from one time-frame to another time-frame, from one market condition to another market condition. It would be wrong if novice trader read it and lose money on this statement.~Hghlight (talk) 9 May 2013 (UTC)

Removed dubious statement

[edit]

I removed this statement because it seems meaningless without specifying thetime constants used in the various moving averages:

However, in this regard the MACD does not lag as much as a basic moving average crossing indicator, since the signal cross can be anticipated by noting the convergence far in advance of the actual crossing.

--Jorge Stolfi (talk) 16:50, 23 April 2014 (UTC)[reply]

Removed unhelpful code

[edit]

I removed the following code fragments because they are not very helpful:

  • MACD – signal = 0
  • EMA[fast,12] – EMA[slow,26] = 0
  • Sign (relative price extremumfinal – relative price extremuminitial) ≠ Sign (relative MACD extremumfinal – MACD extremuminitial)

The first two test only for exact equality, and do not distinguish true crossing from touching and then separating without a crossing. The last one does not define the variables in the formula, which voids the purpose of giving code -- namely, making the definitions absolutely precise. --Jorge Stolfi (talk) 19:01, 23 April 2014 (UTC)[reply]

Minor comment on mathematical interpretation

[edit]

Overall, it was good to consolidate the "Mathematical Interpretation" and "Signal processing" sections. One minor nit in the resulting "Mathematical interpretation":

"MACD estimates the derivative as if it were calculated and then filtered by the two low-pass filters in series... "  was changed to
"MACD estimates the derivative as if it were calculated and then filtered by the two low-pass filters in tandem"  

The "in tandem" usage usually implies operations in parallel which are then combined, and that is not the case. While the filters do their calculations in parallel, and then the difference is calculated, the point of the sentence was that the overall mathematical effect is to estimate a derivative just as if the derivative were calculated ideally, and then filtered by two filters in series, not in parallel. ("In series" can be taken as "followed by".)

There's a distinction there between how the calculations are implemented (which involves 2 filters in parallel), and the net mathematical equivalent result.

The same problem is in the sentence on the average series. It really should read "The average series is also a derivative estimate, with an additional low-pass filter in series for further smoothing (and additional lag)". Gmstanley (talk) 20:20, 23 April 2014 (UTC)[reply]

Lagging indicator

[edit]

The last introductory paragraph just before the table of contents is misleading for several reasons. First, while it is true that moving averages of price are lagging indicators of price, it does not necessarily follow that MACD is a lagging indicator of the price, which is how most people would probably interpret this. But MACD doesn't directly provide an estimate of price. As noted in the mathematical interpretation section, by looking at the difference between two lagged values with different time constants, MACD is calculating and highlighting changes in the first and second derivative estimates. It is only accurate to say that MACD provides lagging indicators of the derivatives, not the price itself.

If the stock price is trending, and the trend (derivative of price vs time) doesn't change, the MACD estimate doesn't change either. For instance, that means that if the price is increasing at a constant rate, MACD remains constant, suggesting that the price will continue to increase at the same rate -- in a straight line. That's an implied linear prediction of the future for the price which is changing, and it will be correct without any lag if the trend is truly constant. That price itself is not directly predicted, but if you used the derivative estimate, you could do that, and there would be no lag in the case of a constant rate of increase or decrease. There's only lags when the trend changes.

Second, if the stock is trading in a range, generally with unpredictable actions, that's just another way of saying that the trend (derivative) is zero. There's always a derivative estimate. In effect, there's simply a lot of noise overwhelming any small magnitude or short term trends that might be occurring. The use of the word "hence" is especially inappropriate. When the trend changes to something other than 0, MACD will pick it up. No trend lasts forever, and MACD simply highlights when changes occur in the trends. There isn't a notion of a "completed" trend in MACD. The derivative of the price vs. time plot always exists and simply changes over time, sometimes slowly, sometimes rapidly. If stocks are trading in a range with no significant (nonzero) trend, MACD says so by its almost constant value, and that doesn't make is less useful or wrong. It's just describing what's happening, which is useful.

I'd suggest simply deleting that paragraph. Gmstanley (talk) 15:46, 25 May 2022 (UTC)[reply]

Another way to look at this: you could do a linear regression using recent price values (including the current value) to fit a straight line through the price data. That is, you get estimates of the value and its derivative at any time past or future. So that straight line can be used to project future price values, and it's predictions would be accurate without lag as long as the price kept increasing or decreasing at the same rate. These future values are weighted linear combinations of the recent values. Any weighted linear combination of past and current value by definition is a (weighted) "moving average" in general signal processing terms. So though the predictions are based on past values, they are not lagging indicators. Just because some calculation includes using past data, it doesn't follow that it must be "lagging". That's because in effect the past data are used to create a model (a linear one in the case of either regression or MACD). Gmstanley (talk) 16:07, 25 May 2022 (UTC)[reply]

Unfortunately, the sources cited throughout the article (even though they are very weak sources in my opinion) call MACD a "lagging indicator". If you find a good source that supports your statements, then we could do something here. Vgbyp (talk) 08:07, 26 May 2022 (UTC)[reply]