Moving Averages

[Odalv]

Just listen to the bulls around here! They'll tell you to blindly buy at any price no matter what. After all, it has only gone up since it's inception, so it will continue to go up for eternity. You can't lose!
Apparently that is all the indicator you need 
Never mind HODLing through a bear market because you aren't really losing money. You will just lose the purchasing power, which isn't what I would call a winning strategy.

What kireinaha said is a good way to look at trading MA's. Sure you will lose on some, but you will ALWAYS be on the right side during big, impulsive moves. In when it's rising and out when it's falling.

As for period lengths, the more time that the MA is averaging, the longer the period you should be using. For a longer period (4 hour, Daily, Weekly), you will trade less, and with less noise, but if the MA is too long, the price will bottom before the MA's cross down, and top before the buy signal is triggered.
Too short, and you will run out of money in sideways movements where there is a lot of noise in underlying indicators.

If you sell at the bottom this will not increase your purchasing power.

That's kinda obvious no?

That's why you don't sell at the bottom! What are you thinking?
Granted, if you are that terrible at trading that you constantly lose money, then trading isn't your forte. You might as well take the easy way out and lose your money the slow painful way (in a bear market dreaming of higher prices). Hope will buy goods, right? Didn't think so...

Can we agree at: Sell at $1000+  buy at $500,$600,$700 ? Because this is my strategy :-).

[empoweoqwj]

Just listen to the bulls around here! They'll tell you to blindly buy at any price no matter what. After all, it has only gone up since it's inception, so it will continue to go up for eternity. You can't lose!
Apparently that is all the indicator you need 
Never mind HODLing through a bear market because you aren't really losing money. You will just lose the purchasing power, which isn't what I would call a winning strategy.

What kireinaha said is a good way to look at trading MA's. Sure you will lose on some, but you will ALWAYS be on the right side during big, impulsive moves. In when it's rising and out when it's falling.

As for period lengths, the more time that the MA is averaging, the longer the period you should be using. For a longer period (4 hour, Daily, Weekly), you will trade less, and with less noise, but if the MA is too long, the price will bottom before the MA's cross down, and top before the buy signal is triggered.
Too short, and you will run out of money in sideways movements where there is a lot of noise in underlying indicators.

If you sell at the bottom this will not increase your purchasing power.

That's kinda obvious no?

That's why you don't sell at the bottom! What are you thinking?
Granted, if you are that terrible at trading that you constantly lose money, then trading isn't your forte. You might as well take the easy way out and lose your money the slow painful way (in a bear market dreaming of higher prices). Hope will buy goods, right? Didn't think so...

Can we agree at: Sell at $1000+  buy at $500,$600,$700 ? Because this is my strategy :-).

Sssh - never reveal your strategy ..... unless its fake

[RyNinDaCleM]

Just listen to the bulls around here! They'll tell you to blindly buy at any price no matter what. After all, it has only gone up since it's inception, so it will continue to go up for eternity. You can't lose!
Apparently that is all the indicator you need 
Never mind HODLing through a bear market because you aren't really losing money. You will just lose the purchasing power, which isn't what I would call a winning strategy.

What kireinaha said is a good way to look at trading MA's. Sure you will lose on some, but you will ALWAYS be on the right side during big, impulsive moves. In when it's rising and out when it's falling.

As for period lengths, the more time that the MA is averaging, the longer the period you should be using. For a longer period (4 hour, Daily, Weekly), you will trade less, and with less noise, but if the MA is too long, the price will bottom before the MA's cross down, and top before the buy signal is triggered.
Too short, and you will run out of money in sideways movements where there is a lot of noise in underlying indicators.

If you sell at the bottom this will not increase your purchasing power.

That's kinda obvious no?

That's why you don't sell at the bottom! What are you thinking?
Granted, if you are that terrible at trading that you constantly lose money, then trading isn't your forte. You might as well take the easy way out and lose your money the slow painful way (in a bear market dreaming of higher prices). Hope will buy goods, right? Didn't think so...

Can we agree at: Sell at $1000+  buy at $500,$600,$700 ? Because this is my strategy :-).

That works! Until it doesn't...
The thing about using MA's is that even when the range bound market is gone or we get in uncharted territory, it still continues to work.

[TERA]

EMAs are good for predicting spurts of bot movement on mtgox.

I actually find MACDs to be more useful.

[empoweoqwj]

EMAs are good for predicting spurts of bot movement on mtgox.

I actually find MACDs to be more useful.

Yeah, you should put the "bot" in bold since all bots I have seen just work on EMAs, which makes there behaviour all too predictable

[arepo]

Technical analysis is astrology for speculators. It is all about confirmation bias. Moving averages tell you about what happened in the past, but they tell you nothing about the future. They can't even tell you about the present.

Its a sentiment-based market, technicals of any kind struggle to cope.

this is nonsense. the smoother and more continuous the price function is, the better these kinds of indicators work. remember calculus? the slope of a curve at a point is very, very close to the slope of a small line segment defined by points close to that point. EMAs basically track the relative slopes of differently-sized line segments, attempting to quantify the instances in which the rate-of-change in the short-term diverges from the long-term, or average, rate-of-change. other indicators, like oscillators, do even more fancy tricks and can be used to predict reversals as well.

--arepo

[odolvlobo]

Technical analysis is astrology for speculators. It is all about confirmation bias. Moving averages tell you about what happened in the past, but they tell you nothing about the future. They can't even tell you about the present.

Its a sentiment-based market, technicals of any kind struggle to cope.

this is nonsense. the smoother and more continuous the price function is, the better these kinds of indicators work. remember calculus? the slope of a curve at a point is very, very close to the slope of a small line segment defined by points close to that point. EMAs basically track the relative slopes of differently-sized line segments, attempting to quantify the instances in which the rate-of-change in the short-term diverges from the long-term, or average, rate-of-change. other indicators, like oscillators, do even more fancy tricks and can be used to predict reversals as well.

The price function is neither smooth nor continuous. You can make it smoother and more continuous by removing information. Removing information makes predictions worse, not better.

The truth is that these indicators do not work because they are flawed. The major flaw is this: a moving average introduces a lag that depends on the size of the window (this is why the moving average graphs always appear shifted to the right). When you compare two moving averages with two different lags you get noise that people misinterpret as signals. If you fix this by shifting the moving averages to remove the lag, then the best you can do is predict the past.

[TERA]

I've found that a crossed MA doesn't neccessarily tell me WHEN to buy but WHERE to buy. I don't need to buy or sell immediately as soon as they cross. Rather, if an MA crosses up, then next time the price approaches the MA, it is probably a safe bet to buy at the MA. Similarly, if an MA crosses down, then next time the price approaches the MA, it is probably a good idea to sell there. But if other indicators are also telling me that a large movement is imminent, I might not wait.

[arepo]

this is nonsense. the smoother and more continuous the price function is, the better these kinds of indicators work. remember calculus? the slope of a curve at a point is very, very close to the slope of a small line segment defined by points close to that point. EMAs basically track the relative slopes of differently-sized line segments, attempting to quantify the instances in which the rate-of-change in the short-term diverges from the long-term, or average, rate-of-change. other indicators, like oscillators, do even more fancy tricks and can be used to predict reversals as well.

The price function is neither smooth nor continuous. You can make it smoother and more continuous by removing information. Removing information makes predictions worse, not better.

The truth is that these indicators do not work because they are flawed. The major flaw is this: a moving average introduces a lag that depends on the size of the window (this is why the moving average graphs always appear shifted to the right). When you compare two moving averages with two different lags you get noise that people misinterpret as signals. If you fix this by shifting the moving averages to remove the lag, then the best you can do is predict the past.

you're correct, the price function is stochastic, which is quite far indeed from the smooth functions standard calculus is used to deal with:



the further in you zoom, the more and more it resembles a random walk. however, contrarily, the further out you zoom, the less and less it resembles a random walk. this is not an accident, or an artifact of the "removal" of information.



case in point: the obvious uptrend is actually perturbed by the "noise" of market inefficiency. bubbles, selloffs (greed and fear, respectively), and other anomalies are affecting the price and masking an underlying trend. in this way, the main goal of the technical analyst is to strip away the noise to get at the information about the market that is hiding in price and volume data.

the trick, of course, is determining what is noise and what is data. this is the art and science of market analysis, the discussion of which must be saved for another time.

that being said -- and hopefully clearing up why it is almost always beneficial to "smooth" price data in some way -- i do believe you are missing the point of moving averages. you can change the size of the "window" to produce any kind of crossover you wish, and for traders who don't realize this and blindly trade using that one signal, you're not going to have a good time. however, this does not invalidate the fact that moving averages are an extremely useful graphical interpretation of the different rates of change hiding in the price function whose interactions mark tops, bottoms, and inflection points.

-- arepo

[RyNinDaCleM]

this is nonsense. the smoother and more continuous the price function is, the better these kinds of indicators work. remember calculus? the slope of a curve at a point is very, very close to the slope of a small line segment defined by points close to that point. EMAs basically track the relative slopes of differently-sized line segments, attempting to quantify the instances in which the rate-of-change in the short-term diverges from the long-term, or average, rate-of-change. other indicators, like oscillators, do even more fancy tricks and can be used to predict reversals as well.

The price function is neither smooth nor continuous. You can make it smoother and more continuous by removing information. Removing information makes predictions worse, not better.

The truth is that these indicators do not work because they are flawed. The major flaw is this: a moving average introduces a lag that depends on the size of the window (this is why the moving average graphs always appear shifted to the right). When you compare two moving averages with two different lags you get noise that people misinterpret as signals. If you fix this by shifting the moving averages to remove the lag, then the best you can do is predict the past.

you're correct, the price function is stochastic, which is quite far indeed from the smooth functions standard calculus is used to deal with:



the further in you zoom, the more and more it resembles a random walk. however, contrarily, the further out you zoom, the less and less it resembles a random walk. this is not an accident, or an artifact of the "removal" of information.



case in point: the obvious uptrend is actually perturbed by the "noise" of market inefficiency. bubbles, selloffs (greed and fear, respectively), and other anomalies are affecting the price and masking an underlying trend. in this way, the main goal of the technical analyst is to strip away the noise to get at the information about the market that is hiding in price and volume data.

the trick, of course, is determining what is noise and what is data. this is the art and science of market analysis, the discussion of which must be saved for another time.

that being said -- and hopefully clearing up why it is almost always beneficial to "smooth" price data in some way -- i do believe you are missing the point of moving averages. you can change the size of the "window" to produce any kind of crossover you wish, and for traders who don't realize this and blindly trade using that one signal, you're not going to have a good time. however, this does not invalidate the fact that moving averages are an extremely useful graphical interpretation of the different rates of change hiding in the price function whose interactions mark tops, bottoms, and inflection points.

-- arepo

Nice post!

[MAbtc]

You can make fiat trading EMA but you'll probably have less and less coins with time.

IIRC, Goombo provided backtesting on his daily EMA trading system and it outperformed "buy and hold" (by what margin I do not recall). This suggests that, if fiat gains are reinvested, that you will have more and more coins with time.

[odolvlobo]

that being said -- and hopefully clearing up why it is almost always beneficial to "smooth" price data in some way -- i do believe you are missing the point of moving averages. you can change the size of the "window" to produce any kind of crossover you wish, and for traders who don't realize this and blindly trade using that one signal, you're not going to have a good time. however, this does not invalidate the fact that moving averages are an extremely useful graphical interpretation of the different rates of change hiding in the price function whose interactions mark tops, bottoms, and inflection points.

Moving averages are extremely useful at interpreting the past, but they contain no information about the future. You can look at a trend and believe that it will continue, but there is nothing in a trend that will tell you how long it will last.

As I pointed out before, the conventional way of comparing moving averages is flawed and produces nothing but noise. When you fix the lag problem, indications then occur too long ago in the past to be useful.

[arepo]

that being said -- and hopefully clearing up why it is almost always beneficial to "smooth" price data in some way -- i do believe you are missing the point of moving averages. you can change the size of the "window" to produce any kind of crossover you wish, and for traders who don't realize this and blindly trade using that one signal, you're not going to have a good time. however, this does not invalidate the fact that moving averages are an extremely useful graphical interpretation of the different rates of change hiding in the price function whose interactions mark tops, bottoms, and inflection points.

Moving averages are extremely useful at interpreting the past, but they contain no information about the future.

this is patently false because the price function is not a random walk. look for my future post about triangle consolidation patterns for an example of how autocorrelations in the price function can be surprisingly consistent, and how, because of market inefficiencies, past performance can sometimes predict future results.

--arepo

hint: the price function, on large scales, does not have many cusps. similarly to a continuous function, the rate of change of the price changes before the price itself changes.

[odolvlobo]

You can make fiat trading EMA but you'll probably have less and less coins with time.

IIRC, Goombo provided backtesting on his daily EMA trading system and it outperformed "buy and hold" (by what margin I do not recall). This suggests that, if fiat gains are reinvested, that you will have more and more coins with time.

I don't know the details, but it is likely that he tuned his parameters to maximize the backtesting results. That's what people do to improve their system. However, it is a fallacy to believe that the parameters that maximize returns over the last N years will work well for the next N years.

[MAbtc]

You can make fiat trading EMA but you'll probably have less and less coins with time.

IIRC, Goombo provided backtesting on his daily EMA trading system and it outperformed "buy and hold" (by what margin I do not recall). This suggests that, if fiat gains are reinvested, that you will have more and more coins with time.

I don't know the details, but it is likely that he tuned his parameters to maximize the backtesting results. That's what people do to improve their system. However, it is a fallacy to believe that the parameters that maximize returns over the last N years will work well for the next N years.

That may be. But it would also be quite inaccurate to say that trading with an EMA system will net one "less and less coins."

[empoweoqwj]

You can make fiat trading EMA but you'll probably have less and less coins with time.

IIRC, Goombo provided backtesting on his daily EMA trading system and it outperformed "buy and hold" (by what margin I do not recall). This suggests that, if fiat gains are reinvested, that you will have more and more coins with time.

I don't know the details, but it is likely that he tuned his parameters to maximize the backtesting results. That's what people do to improve their system. However, it is a fallacy to believe that the parameters that maximize returns over the last N years will work well for the next N years.

So true. People tune their systems by backtesting, and think to themselves what an amazing system they have. When all they've done is optimize for the past.

[arepo]

You can make fiat trading EMA but you'll probably have less and less coins with time.

IIRC, Goombo provided backtesting on his daily EMA trading system and it outperformed "buy and hold" (by what margin I do not recall). This suggests that, if fiat gains are reinvested, that you will have more and more coins with time.

I don't know the details, but it is likely that he tuned his parameters to maximize the backtesting results.

well maybe you should explore the details before advancing ad hoc refutations of an extremely diligent project? i was there during the course of goomboo's thread, and it was a valuable service he did for these forums. i don't know enough about the details to tell you whether or not he tuned the parameters, but then again i'm not the one making the claims.

[empoweoqwj]

You can make fiat trading EMA but you'll probably have less and less coins with time.

IIRC, Goombo provided backtesting on his daily EMA trading system and it outperformed "buy and hold" (by what margin I do not recall). This suggests that, if fiat gains are reinvested, that you will have more and more coins with time.

I don't know the details, but it is likely that he tuned his parameters to maximize the backtesting results.

well maybe you should explore the details before advancing ad hoc refutations of an extremely diligent project? i was there during the course of goomboo's thread, and it was a valuable service he did for these forums. i don't know enough about the details to tell you whether or not he tuned the parameters, but then again i'm not the one making the claims.

And what happened after that thread? Every man and his dog built a bot that copied his EMA system. So now we have thousands of bots running the same "system". That's when systems stop working ....

[arepo]

You can make fiat trading EMA but you'll probably have less and less coins with time.

IIRC, Goombo provided backtesting on his daily EMA trading system and it outperformed "buy and hold" (by what margin I do not recall). This suggests that, if fiat gains are reinvested, that you will have more and more coins with time.

I don't know the details, but it is likely that he tuned his parameters to maximize the backtesting results.

well maybe you should explore the details before advancing ad hoc refutations of an extremely diligent project? i was there during the course of goomboo's thread, and it was a valuable service he did for these forums. i don't know enough about the details to tell you whether or not he tuned the parameters, but then again i'm not the one making the claims.

And what happened after that thread? Every man and his dog built a bot that copied his EMA system. So now we have thousands of bots running the same "system". That's when systems stop working ....

such is the nature of all systematic trading strategies -- they counteract the patterns which they utilise. this is no way invalidates them while the consistency they exploit is still exploitable.

surprise! the price function isn't a random walk! speculating is a better game than a simple coin toss! although i don't know why that is much of a surprise, seeing as you're trying your hand at it. all you naysayers make me laugh because your arguments tend to claim, reductio ad absurdum, that no trading strategy is viable, in which case i have no clue why you're even in this subforum, at all

--arepo

[empoweoqwj]

You can make fiat trading EMA but you'll probably have less and less coins with time.

IIRC, Goombo provided backtesting on his daily EMA trading system and it outperformed "buy and hold" (by what margin I do not recall). This suggests that, if fiat gains are reinvested, that you will have more and more coins with time.

I don't know the details, but it is likely that he tuned his parameters to maximize the backtesting results.

well maybe you should explore the details before advancing ad hoc refutations of an extremely diligent project? i was there during the course of goomboo's thread, and it was a valuable service he did for these forums. i don't know enough about the details to tell you whether or not he tuned the parameters, but then again i'm not the one making the claims.

And what happened after that thread? Every man and his dog built a bot that copied his EMA system. So now we have thousands of bots running the same "system". That's when systems stop working ....

such is the nature of all systematic trading strategies -- they counteract the patterns which they utilise. this is no way invalidates them while the consistency they exploit is still exploitable.

surprise! the price function isn't a random walk! speculating is a better game than a simple coin toss! although i don't know why that is much of a surprise, seeing as you're trying your hand at it. all you naysayers make me laugh because your arguments tend to claim, reductio ad absurdum, that no trading strategy is viable, in which case i have no clue why you're even in this subforum, at all

--arepo

Never said it's a random walk, never said its a coin toss, simply that using the same EMA trading system thousands of bots are using is not a smart play. But use that knowledge to your advantage, sure