questions from a future quant | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

The Birth of a Quant Trader 107 replies

Statistical mean of the market [quant corner] 191 replies

Quant trading & QuantLib 6 replies

SAS for quant analysis 0 replies

Quant Corner 12 replies

Attachments: questions from a future quant
Exit Attachments

questions from a future quant

  • #1
  • First Post: Jun 29, 2005 9:40am Jun 29, 2005 9:40am
Since this is my first post, I should probably give a brief introduction of myself. I just graduated with a BS in physics and I'll begin my work on a PhD in physics this fall. Since I will now be earning a salary for the first time in my life (as opposed to an hourly wage), I plan on setting money aside to invest.

My mother took me to a 4x Made Easy class last year. It didn't take me too long to figure out that 4xME was just a waste of money. But, on the bright side, I did learn about the Forex market during the seminar. I've been paper trading an account with CMS over the past 3 months, as well as doing a lot of reading on money management and technical analysis. But so far, I've mjust hovered around breaking even.

There seem to be a number of knowlegeable traders on this forum. I was hoping some of you would be kind enough to answer some questions for an eager-to-learn future speculater.

1) I seem to get much better results when I simply draw trend lines rather than use indicators. Basically, for a long position, I just draw a support trendline, enter the trade on the third touch and exit the trade when either the price goes a little below support or reaches a point just below resistance (I flip things around in the obvious manner for a short position). My question is, is there anybody here that trades off the just the charts (i.e. just drawing support and resistence and looking for bar patterns), or do nearly all succesful traders use some sort of indicator or combination of indicadors?

2) If you do use an indicator, how do you combine what support, resistence, and bar patterns "say" with what your indicator tells you to do, while at the same time mantaining strict discipline in your trading?

3) For long term trades, do you thing it is important to know what percent of people are holding buy and sell options on your currency? Or is that information already reflected in the charts? (I don't trade options, so I may have the terminoligy wrong here, I think put and call options might be the correct term, in any case, let me know if the question isn't clear)

4) Since I will begin graduate school this fall, I don't anticipate having a LOT of free time to trade. So that either means setting up Visual Trader to automatically trade for me, or just focusing on long term trends. Does anybody know if it is possible to write a program on Visual Trader that will raise stops along a pre-defined trendline (i.e. I would draw the trendline, fingure out the slope, and have the stops and limits move up at a rate equal to the slope of the line).

5) I've read a lot about the importance of the close in technical analysis books. And I can understand why that would be the case in a market that is only open a certain number of hours per day. But, since the forex is a 24 hr/day market, does the close of a period in ANY time-frame (other than the close of a week) hold any special significance?

6) Does anybody here make trades based on bar or candlestick patterns alone? I aske just because it seems like that might be easy to program, and would have the potential for a lot of trades, but perhaps those trades wouldn't be so successful.

7) Lastly, I read a number of money management techniques that suggest staggering your stoplosses. I guess I can't follow the logic on why this would be advantageous. If I decide I am willing to lose x dollars on a trade, why would I then say I am willing to lose y dollars on that same trade? Does anybody here stagger their stops? If so, can you explain the reasoning behind it?

Thanks in advance to anybody that chooses to answer one or more of my questsions. Hopefully I will be able to return the favor by becoming a contributing member of this forum some day (as opposed to somebody with a bunch of questions and no answers) .
  • #2
  • Jun 29, 2005 10:14am Jun 29, 2005 10:14am
Quoting mattlorig
2) If you do use an indicator, how do you combine what support, resistence, and bar patterns "say" with what your indicator tells you to do, while at the same time mantaining strict discipline in your trading?
Hi Matt and welcome to ForexFactory!

While this does not directly address your question of reconciling indicators with support/resistance/barpattern, you may find it useful to read the "Chart Thread" thread in the Forex Discussion subforum. Also take a look at the "GBP/JPY trading using daily charts" thread in the Trading Systems subforum.

Basically, look at anything posted by James16 in the last month or so.

Diallist
sxaxlxvxaxtxixoxnxbxyxgxrxaxcxexdxoxtxoxrxgx
  • #3
  • Jun 29, 2005 10:18am Jun 29, 2005 10:18am
I, personally, would not recommend trading using candlesticks alone. The gentlemen who first introduced this method to the market himself had poor results trying to utilize only the candlesticks. They can, however, aid in the determination of trend changes and who is controlling those trends at any given time (bulls or bears).
Your correct in your assumption that the charts can indicate volume movement. The number of people holding or shorting any given currency is irrelevant. The market moves back and forth a majority of the time based on emotional responses in small swings, but long term trends are what's important. How you determine those trends and what tools you utilize is highly dependant upon each individuals preferences and their ability to understand those tools.
I have always enjoyed using EMA's, SARS, RSI, and Stochastics in a combination that I understand and has had good results for me personally.
Capital Preservation is key to long term wealth accumulation
  • #4
  • Jun 29, 2005 1:59pm Jun 29, 2005 1:59pm
hi mat and welcome to the factory. you have some great questions, many of them you may never find the answer to. but let me see if i can add insight to a few of them...


1) I seem to get much better results when I simply draw trend lines rather than use indicators. Basically, for a long position, I just draw a support trendline, enter the trade on the third touch and exit the trade when either the price goes a little below support or reaches a point just below resistance (I flip things around in the obvious manner for a short position). My question is, is there anybody here that trades off the just the charts (i.e. just drawing support and resistence and looking for bar patterns), or do nearly all succesful traders use some sort of indicator or combination of indicadors?


you can successfully trade off of just the charts and with no indicators. over the past year i have been gravitating toward indicator-less systems. while i do trade some systems on indicators, i find that they rely too heavily on recent data. these days i measure market behavior of independent events instead of using indicators, and this seems to be working great.



2) If you do use an indicator, how do you combine what support, resistence, and bar patterns "say" with what your indicator tells you to do, while at the same time mantaining strict discipline in your trading?


this is one of those questions you may never know the answer to



3) For long term trades, do you thing it is important to know what percent of people are holding buy and sell options on your currency? Or is that information already reflected in the charts? (I don't trade options, so I may have the terminoligy wrong here, I think put and call options might be the correct term, in any case, let me know if the question isn't clear)


well, i dont have a great answer for you, but if you look at market theory, all available knowledge is built into the market instantaneously.



4) Since I will begin graduate school this fall, I don't anticipate having a LOT of free time to trade. So that either means setting up Visual Trader to automatically trade for me, or just focusing on long term trends. Does anybody know if it is possible to write a program on Visual Trader that will raise stops along a pre-defined trendline (i.e. I would draw the trendline, fingure out the slope, and have the stops and limits move up at a rate equal to the slope of the line).


this would be a VERY difficult system to program, but it is definitely possible with TradeStation. im not sure about visual trader, but judging by the complexity of programming your idea, i would be shocked if VT let you do it.



5) I've read a lot about the importance of the close in technical analysis books. And I can understand why that would be the case in a market that is only open a certain number of hours per day. But, since the forex is a 24 hr/day market, does the close of a period in ANY time-frame (other than the close of a week) hold any special significance?


the cool thing about Forex is that there is no standard open and close time that everyone follows. people use all kinds of different times for the open and close, i generally consider that the market opens and closes at 5PM EST.



6) Does anybody here make trades based on bar or candlestick patterns alone? I aske just because it seems like that might be easy to program, and would have the potential for a lot of trades, but perhaps those trades wouldn't be so successful.


candle patterns are not something i subscribe to because my research has shown they do not offer predictability. i do find that bar measurements can tell you a lot about the future, however.



7) Lastly, I read a number of money management techniques that suggest staggering your stoplosses. I guess I can't follow the logic on why this would be advantageous. If I decide I am willing to lose x dollars on a trade, why would I then say I am willing to lose y dollars on that same trade? Does anybody here stagger their stops? If so, can you explain the reasoning behind it?


you generally want to keep the amount at risk (risked on each trade) a constant amount. you dont know if the next trade will be a winner or loser, so you really have no way of adjusting your amount risked on a per trade basis. you may want to move your stoploss up or down due to support resistence levels, but the max $$ amount you can lose is what you want to keep constant. you may want to read this for a better explanation.



Thanks in advance to anybody that chooses to answer one or more of my questsions. Hopefully I will be able to return the favor by becoming a contributing member of this forum some day (as opposed to somebody with a bunch of questions and no answers) .


hope i have offered some help! good luck with trading and i look forward to seeing you around here!
Relax and be happy.
  • #5
  • Jun 30, 2005 1:45pm Jun 30, 2005 1:45pm
Thanks for the suggestions guys. By the way Merlin, I just employed your don't-buy-a-breakout-unless-it-follows-a-false-breakout rule and it worked great. I guess trading is a lot about building up a bag of tricks.
  • #6
  • Jul 1, 2005 5:19pm Jul 1, 2005 5:19pm
Quoting mattlorig
By the way Merlin, I just employed your don't-buy-a-breakout-unless-it-follows-a-false-breakout rule and it worked great...
glad you like it

i got that idea from Richard Dennis.
Relax and be happy.
  • #7
  • Jul 1, 2005 11:44pm Jul 1, 2005 11:44pm
Quote
these days i measure market behavior of independent events instead of using indicators, and this seems to be working great.

Hi Merlin,

Would you take the time to give me a more detailed description of what you mean in that statement? I always thought market behavior was a lot of interdependencies ... but of course that is true... so an "independent event" would have to be a subset of all these thousands of interrelated events that make up the market. I suspect you mean a definable event that can be isolated as something that took place within a certain set of parameters that isolate the "event" you are describing. However it seems that even an indicator simply measures an isolated kind of activity. Therefore I'm completely intrigued by your statement. Please elaborate.

Thanks,

FT
  • #8
  • Jul 2, 2005 3:45pm Jul 2, 2005 3:45pm
Quoting fijitrader
Hi Merlin,

Would you take the time to give me a more detailed description of what you mean in that statement? I always thought market behavior was a lot of interdependencies ... but of course that is true... so an "independent event" would have to be a subset of all these thousands of interrelated events that make up the market. I suspect you mean a definable event that can be isolated as something that took place within a certain set of parameters that isolate the "event" you are describing. However it seems that even an indicator simply measures an isolated kind of activity. Therefore I'm completely intrigued by your statement. Please elaborate.

Thanks,

FT
fiji fiji fiji, only you would see that the true depth of that statement i probably shouldnt have even posted that, because it digs deep into my secrets

but since you called me on it, i will do a little explaining without giving away to much (although i think you already understand what im doing)...

say you have a moving average looking at the last 20 days. you would in turn be trying to predict future prices using the past 20 days. thats how most indicators works. but what if data from a year ago can help you predict future prices? you wouldnt want your indicator to look at the past year because that would give you too much useless data and dilute its predictive value.

in this circumstance, where data from a year ago can help you predict future prices, you can define the current situation that you are trying to predict, then ONLY consider past situations that meet that same criteria. in this way, you are only considering past events that are just like the present one. beileve it or not, i have found that if an event is uniquely definable, it will have low variance and will follow a standard distribution over past and future data. and anytime you can find a standard distribution in price data, you can take advantage of it because it is PREDICTABLE!!!

the code on these types of systems is horrendous, unfortunately. there are no platforms that make it easy. thats how i know its a good idea
Relax and be happy.
  • #9
  • Jul 2, 2005 8:30pm Jul 2, 2005 8:30pm
Quoting merlin
fiji fiji fiji, only you would see that the true depth of that statement i probably shouldnt have even posted that, because it digs deep into my secrets

but since you called me on it, i will do a little explaining without giving away to much (although i think you already understand what im doing)...

say you have a moving average looking at the last 20 days. you would in turn be trying to predict future prices using the past 20 days. thats how most indicators works. but what if data from a year ago can help you predict future prices? you wouldnt want your indicator to look at the past year because that would give you too much useless data and dilute its predictive value.

in this circumstance, where data from a year ago can help you predict future prices, you can define the current situation that you are trying to predict, then ONLY consider past situations that meet that same criteria. in this way, you are only considering past events that are just like the present one. beileve it or not, i have found that if an event is uniquely definable, it will have low variance and will follow a standard distribution over past and future data. and anytime you can find a standard distribution in price data, you can take advantage of it because it is PREDICTABLE!!!

the code on these types of systems is horrendous, unfortunately. there are no platforms that make it easy. thats how i know its a good idea
I didn't think you were going to say much more about the subject. Understandable I guess. You've shed a sliver more light though pretty much through confirming a generality. Though really you've just described any pattern recognition process.

Clearly the secret is in the type of events you are going after and how you quantify and qualify them. The greater the number of qualities and quantites you use to define an event the more elaborate the code required to find and differentiate them from false positives. With the information you've given this does not seem (speaking generally) much different than any other type of indicator except in the complexity and methodology used to define an event. Obviously this is the area of inquirey I had hoped to elicit. If you can take it any further without handing us the farm please do.

Thanks,

FT
  • #10
  • Jul 3, 2005 1:02am Jul 3, 2005 1:02am
[quote=merlin
.....you can define the current situation that you are trying to predict, then ONLY consider past situations that meet that same criteria. in this way, you are only considering past events that are just like the present one......
[/QUOTE]

Merlin, I was thinking about this kind of thing for a while already -
one could use autocorrelations and things kind of markov chain models. Just like you predict weather (no weather prediction id based on indicators - they are all based on thing such as those described above).
  • #11
  • Jul 3, 2005 8:22am Jul 3, 2005 8:22am
Quoting merlin
fiji fiji fiji, only you would see that the true depth of that statement i probably shouldnt have even posted that, because it digs deep into my secrets

but since you called me on it, i will do a little explaining without giving away to much (although i think you already understand what im doing)...

say you have a moving average looking at the last 20 days. you would in turn be trying to predict future prices using the past 20 days. thats how most indicators works. but what if data from a year ago can help you predict future prices? you wouldnt want your indicator to look at the past year because that would give you too much useless data and dilute its predictive value.

in this circumstance, where data from a year ago can help you predict future prices, you can define the current situation that you are trying to predict, then ONLY consider past situations that meet that same criteria. in this way, you are only considering past events that are just like the present one. beileve it or not, i have found that if an event is uniquely definable, it will have low variance and will follow a standard distribution over past and future data. and anytime you can find a standard distribution in price data, you can take advantage of it because it is PREDICTABLE!!!

the code on these types of systems is horrendous, unfortunately. there are no platforms that make it easy. thats how i know its a good idea
Hi,

An interesting idea and it reminded me of something 'similar' had I read before. I had a dig around.

The Geometry of Stock Market Profits - Micheal S. Jenkins, p134, Trading tip 6 - Mirror Image Foldback Patterns. He say that patterns can repeat in reverse order in the future. At major highs and lows, you can 'fold' the chart in half and what has happened in the past will repeat in the future (time movement may be speeded up or slowed down). A simple example is the head and shoulders pattern. If you fold the left shoulder and head, it will repeat to right shoulder in the future. Micheal Jenkins was saying you can predict 3-9 or more months ahead if you can spot the right place.

Similar but not the same. Food for thought

Regards
Alan
  • #12
  • Jul 3, 2005 1:55pm Jul 3, 2005 1:55pm
Quoting ivenger
Just like you predict weather (no weather prediction id based on indicators - they are all based on thing such as those described above).
is that really how they predict the weather??? how do you know that?

man, i might be studing weather patterns for the next couple weeks!
Relax and be happy.
  • #13
  • Jul 3, 2005 2:02pm Jul 3, 2005 2:02pm
Quoting fijitrader
I didn't think you were going to say much more about the subject. Understandable I guess. You've shed a sliver more light though pretty much through confirming a generality. Though really you've just described any pattern recognition process.

Clearly the secret is in the type of events you are going after and how you quantify and qualify them. The greater the number of qualities and quantites you use to define an event the more elaborate the code required to find and differentiate them from false positives. With the information you've given this does not seem (speaking generally) much different than any other type of indicator except in the complexity and methodology used to define an event. Obviously this is the area of inquirey I had hoped to elicit. If you can take it any further without handing us the farm please do.

Thanks,

FT
yeah i guess you are right here, its related to pattern recognition. inicators are not about pattern recognition, thats really the difference.

when you say pattern recognition i think of "head and shoulders" or "flag" formations. thats not at all what im talking about though. im looking at patterns created by the market's infrastructure (ie settlement dates, GDP report dates etc.).

the reason i say this stuff is hard to code is because you have to write all of the arrays. its really not that tough, but compared to indicator based systems it sure is.
Relax and be happy.
  • #14
  • Edited 5:57pm Jul 3, 2005 5:52pm | Edited 5:57pm
Quoting merlin
yeah i guess you are right here, its related to pattern recognition. inicators are not about pattern recognition, thats really the difference.

when you say pattern recognition i think of "head and shoulders" or "flag" formations. thats not at all what im talking about though. im looking at patterns created by the market's infrastructure (ie settlement dates, GDP report dates etc.).

the reason i say this stuff is hard to code is because you have to write all of the arrays. its really not that tough, but compared to indicator based systems it sure is.
In my thinking I usually work from the general to the specific. To me any attempt to find a repetitive occurance is a pattern recognition process. There clearly are subsets of "pattern recognition" that most people think qualify or don't qualify as "patterns" but since the term is so general I cannot find expansive thinking within those confines. When people think first in terms of the subsets they've already defined as patterns and reject what they don't consider to be patterns I find they tend to overdefine their thinking and therefore inadvertantly eliminate certain lines of thinking because they run out of words to describe certain kinds of phenomena and therefore become inefficient in their thinking.

Whereas I agree that indicators do not recognise patterns, but rather filter activity and provide us with a lense through which to observe and or recognise patterns, I'd still say that indicators are also used to find patterns. For instance: "oversold" is a pattern because it is a level we set where the oscillator is less likely to continue indicating a trend. We surmise that it is unlikely to continue because we have seen a pattern of reversals happening at that level.

Basically speaking anything that can be statistically modeled as an event would qualify as a potential pattern because the word pattern does not have a more basic subset that is definite in its definition. Words like "foggy notion" or "vague impression" are too indefinite for pattern recognition in trading. Any event that might be more specific such as "head and shoulders" or "oversold" are themselves subsets of the more general term "pattern" and are too specific to be called something less specific such as "foggy notion". They are, however, more specific than the patterns "up" or "down" but that does not mean the term "up" is not a pattern regardless of how simple or basic it seems. In fact when we combine the event "up" with the pattern "trend" we can get completely bogged down in unbelievable complexity just defining such a compound pattern. Obviously various subsets of "pattern" can be combined with other technical or fundamental events to postulate more specific and/or sophisticated patterns.

Therefore while I understand your comments and realize you are speaking in subsets when you refer to various events/patterns that can be qualified as sufficiently signficant as to be profitable -- it is those specific types of patterns that I'd love to learn more about. Granted that you are perhaps wisely not in the habit of divulging such secrets I'm still sufficiently intrigued to try and extract any more tidbits you'd care to let slip out.

Thanks,

FT
  • #15
  • Jul 3, 2005 6:06pm Jul 3, 2005 6:06pm
Quoting ivenger
Just like you predict weather (no weather prediction id based on indicators - they are all based on thing such as those described above).
What does this mean? If no weather prediction id based on indicators then what is a barometer, a thermometer, a weather chart mapping barometric pressures and temperatures, an altimeter, a wind speed indicator???

What is weather prediciton based on if those are not indicators? When I took meteorology for flight school and also for various sailing certification levels all of the courses were about reading a host of indicators, mostly laid out on charts or heard in reports, including cloud types. A cold front is a pattern of pressures, temperatures, and air movement velocity, all derived from reading indicators.

FT
  • #16
  • Jul 4, 2005 12:50am Jul 4, 2005 12:50am
Quoting fijitrader
What does this mean? If no weather prediction id based on indicators then what is a barometer, a thermometer, a weather chart mapping barometric pressures and temperatures, an altimeter, a wind speed indicator???

FT
No, those are not indicators in the sense we use them. What we use when we say "indicators" in trading signifies some kind of derivation from the price data.
But pressure, wind speed etc. are correlated, but different components of weather. When you do a numerical prediction of temperature at time t for example - you wish to know all those parameters of the system at time (t-1) (pressure X, wind speed Y etc.). In order to make the prediction you compare them to the historical weather data and see that in some past point (or several different points) in the past (k-1) the pressure was X, the wind speed Y and the cold front was Z km. away and you KNOW that at time k the temperature back then was 25 degrees. You predict that this time the temperature also will be the same. Of course, usually you are not basing your prediction on just one time point in the past, but on several of them and check the variance of the temperatures. You also look at time series from (k-n)...(k-1).
In trading data we have the price (which is compared to the temperature) and the only other parameter I know of which is continuous is volume. But you could try to get additional data from say different currency crosses. In addition - I think we could incorporate the discrete data regarding some economic events.

Ok, it was a bit unclear - I was in a hurry. Hope I am being understood - any more questions?
  • #17
  • Edited 4:22am Jul 4, 2005 4:10am | Edited 4:22am
Quoting ivenger
No, those are not indicators in the sense we use them. What we use when we say "indicators" in trading signifies some kind of derivation from the price data.
But pressure, wind speed etc. are correlated, but different components of weather. When you do a numerical prediction of temperature at time t for example - you wish to know all those parameters of the system at time (t-1) (pressure X, wind speed Y etc.). In order to make the prediction you compare them to the historical weather data and see that in some past point (or several different points) in the past (k-1) the pressure was X, the wind speed Y and the cold front was Z km. away and you KNOW that at time k the temperature back then was 25 degrees. You predict that this time the temperature also will be the same. Of course, usually you are not basing your prediction on just one time point in the past, but on several of them and check the variance of the temperatures. You also look at time series from (k-n)...(k-1).
In trading data we have the price (which is compared to the temperature) and the only other parameter I know of which is continuous is volume. But you could try to get additional data from say different currency crosses. In addition - I think we could incorporate the discrete data regarding some economic events.

Ok, it was a bit unclear - I was in a hurry. Hope I am being understood - any more questions?
Clearly you have missed my point which is that indicators are used in weather predicting. All the instruments used to measure weather related activity are indicators and all they do is measure events which we use for pattern recognition which we use for prediction.

Technical indicators as used in trading measure movement relative to certain constants whether they be time, volume, frequency, price, speed, distance, or combinations thereof and that does not begin to cover the more sophisticated smoothing algorithms used to filter noise nor does it address any of the fundamental indicators that may be used in trading. But no matter how you slice up the data that is used to forecast market conditions the tools you use are indicators. They are indicative of historical events but not predictive without employing pattern recognition and statistical evaluation to establish probability. The same can be said for weather prediction.

Indicators used for atmospheric conditions measure movement relative to certain constants. With them we measure pressure, temperature, humidity, velocity etc. and use pattern recognition to establish probablity of future events. But all the tools we use for measurement are indicators. We even call them indicators, like I mentioned before, wind speed indicator.

An indicator is simply an instrument that provides feedback that may be used to quantatively observe change. When statstical evaluation is applied to historical events that we have observed with the help of these indicators then we can establish probability for recurrence. Clearly there are more components with weather if you compare only technical data in trading to a combination of both technical and fundamental data in weather forcasting. However when you also bring fundamental indicators into trading the parallels are so fundamentally similar it is hard to argue that they are different except semantically or qualitatively.
  • #18
  • Jul 4, 2005 4:28am Jul 4, 2005 4:28am
Maybe I wasn't clear on this - you are right that the different weather indicators are the same kind as technical indicators. However, the algorithms used to predict future weather are such as those described by Merlin - looking for certain patterns and predicting based on those patterns in the past.
  • #19
  • Jul 4, 2005 7:14pm Jul 4, 2005 7:14pm
Quoting ivenger
Maybe I wasn't clear on this - you are right that the different weather indicators are the same kind as technical indicators. However, the algorithms used to predict future weather are such as those described by Merlin - looking for certain patterns and predicting based on those patterns in the past.
Indeed! The same is true of any predictive methodology. Observing, defining, measuring, documenting, and analyzing historical events to discover probabilities of future ones. Also since Merlin did not post any algorthms we are talking in pretty general terms. Anyway I was hoping to elicit a bit more specific information from Merlin about the kind of events he was talking about but I suspect this diversion has provided him sufficient cover to avoid responding.

FT
  • #20
  • Jul 6, 2005 1:52pm Jul 6, 2005 1:52pm
Quoting fijitrader
In my thinking I usually work from the general to the specific. To me any attempt to find a repetitive occurance is a pattern recognition process. There clearly are subsets of "pattern recognition" that most people think qualify or don't qualify as "patterns" but since the term is so general I cannot find expansive thinking within those confines. When people think first in terms of the subsets they've already defined as patterns and reject what they don't consider to be patterns I find they tend to overdefine their thinking and therefore inadvertantly eliminate certain lines of thinking because they run out of words to describe certain kinds of phenomena and therefore become inefficient in their thinking.
gotcha. the "subset" of patterns i am looking are patterns caused by events. i find that when a pattern is caused by a event it is rooted in crowd psychology and will persist over time (ie it is predictable). i do trade 123 formations, and those are not caused by events (they are just patterns floating out there), so i cant really say this is the only subset i look at. but in the past few months i have completely dedicated my research to patterns caused by events.

Quoting fijitrader
Whereas I agree that indicators do not recognize patterns, but rather filter activity and provide us with a lense through which to observe and or recognise patterns, I'd still say that indicators are also used to find patterns. For instance: "oversold" is a pattern because it is a level we set where the oscillator is less likely to continue indicating a trend. We surmise that it is unlikely to continue because we have seen a pattern of reversals happening at that level.
most indicators to me are an attempt to find an algorithm that predicts the market. usually indicators give you no more than a different shade of sunglasses which to look at price. personally, i do not believe there is an algorithm out there that predicts the market, putting me in the "believer" category of the efficient market hypothesis.

i disagree that "oversold" is a pattern. oversold is a statement made by an indicator based on that indicators algorithm. also, patterns are found intermittently in the data, whereas oscillator indicators look at uninterrupted data over a prior period.
Relax and be happy.
0 traders viewing now