God Indicator - page 3
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Ok, great points there thanks! (except number 4, that's taken for granted! :P)
I like that you mentioned noise, even though it's very obvious, it is often given little consideration. Sideways markets could be considered a type of noise too, for example, a SMA of 10 might filter out typical noise, but not the side ways market, but a SMA of 100 might filter out the sideways market. So, noise can be seen on different resolutions. Counter-trends could also be considered noise. If, trading is done short term, then a counter trend could be seen as a new trend, but, with a long term outlook, the counter trend is noise. Noise is very general, and indicators that can perform at different resolutions such as MAs are a good way to accommodate all of this.
The issue I have a fixation on in terms of noise, is lag. For example, the SMA is lagged, it smooths really well, but because of the lag it looses it's usefulness fairly quickly in that regard.
How do you picture number 3 being improved? Can you pick a popular indicator (SMA, MACD, etc...) and explain how it might function?
Brain storming is always fun!Psychological* typo correction
Anyways....
The indicator should be self-adjusting, self-learning even.
It should increase and decrease in sensitivity, depending on the hour of the day, day of the week, month of the year, and proximity to daily level.
It should also keep a record of its changes in sensitivity, and use that information toward future adjustments.
It should keep a record of how many ticks occurred in each bar, their direction, and combined average with previous candles; then adjust itself accordingly.
It should also "watch" other cross pairs and be aware of any major moves in prices on those other pairs; and thus adjust accordingly to the statistical odds of that other currencies affect on the current currency. Timing also a factor with this. For example; how long before the 25 pip jump on the G/U affects the E/U, and if at all?
It should keep a record of patterns and learn new ones as they form. It should react accordingly, adjusting itself according to the odds of a pattern forming and repeating.
It should have fibo recognition (calculation of and plotting), and be able to predetermine the most probable fibo level the price is going to stop and reverse, or if at all.
It should also be able to play Beethoven's 9th in D-minor on piano.
Psychological* typo correction
Anyways....
The indicator should be self-adjusting, self-learning even.
It should increase and decrease in sensitivity, depending on the hour of the day, day of the week, month of the year, and proximity to daily level.
It should also keep a record of its changes in sensitivity, and use that information toward future adjustments.
It should also "watch" other cross pairs and be aware of any major moves in prices on those other pairs; and thus adjust accordingly to the statistical odds of that other currencies affect on the current currency. Timing also a factor with this. For example; how long before the 25 pip jump on the G/U affects the E/U, and if at all?
It should keep a record of patterns and learn new ones as they form. It should react accordingly, adjusting itself according to the odds of a pattern forming and repeating.
It should have fibo recognition, and be able to pre-determine the most probable fibo level the price is going to stop and reverse, or if at all.
It should also be able to play Beethoven's 9th in D-minor on piano.hahaha, the music maybe a bit over the top but the rest isn't as crazy as I assume that sarcasm makes it out to be. My field is in advanced machine learning techniques (aka, artificial intelligence) that are used when a direct approach is unknown or isn't possible.
Determining the most probable stop and reverse levels is a problem I'm working on at the moment.
All these are great ideas, however, combining them into a single indicator is no easy task. The best way, would be to design an indicator for each one and then build a new indicator incorporating the "sub" indicators. I would imagine it would mostly rely on statistical probabilities from what you've detailed.
Do you have a background in statistics or mathematics or computer science?
hahaha, the music maybe a bit over the top but the rest isn't as crazy as I assume that sarcasm makes it out to be. My field is in advanced machine learning techniques (aka, artificial intelligence) that are used when a direct approach is unknown or isn't possible.
Determining the most probable stop and reverse levels is a problem I'm working on at the moment.
All these are great ideas, however, combining them into a single indicator is no easy task. The best way, would be to design an indicator for each one and then build a new indicator incorporating the "sub" indicators. I would imagine it would mostly rely on statistical probabilities from what you've detailed.
Do you have a background in statistics or mathematics or computer science?Late post edit entry...
"It should keep a record of how many ticks occurred in each bar, their direction, and combined average with previous candles; then adjust itself accordingly."
Anyways...Stats yes. comp science no.
Currently I am working on a formula that forecasts the daily high and low for each currency pair.
If you take a look at my thread, you'll see a few examples of the accuracy of the formula, but also the inaccuracy.
As I have learned, this formula is something I must adjust on a regular daily basis. (On the days I was right, the original formula was proved successful. On the days I was wrong, the formula needed adjustments and tweaks to match market conditions and requirements. ) And it's those conditions that I am trying to line-up and use accordingly.
I took a break from posting, because I need to ensure I post forecasts that are at least 70%-80% accurate per week.
I just have a knack for doing things the right way.
Late post edit entry...
"It should keep a record of how many ticks occurred in each bar, their direction, and combined average with previous candles; then adjust itself accordingly."
Anyways...Stats yes. comp science no.
Currently I am working on a formula that forecasts the daily high and low for each currency pair.
If you take a look at my thread, you'll see a few examples of the accuracy of the formula, but also the inaccuracy.
As I have learned, this formula is something I must adjust on a regular daily basis. (On the days I was right, the original formula was proved successful. On the days I was wrong, the formula needed adjustments and tweaks to match market conditions and requirements. ) And it's those conditions that I am trying to line-up and use accordingly.
I took a break from posting, because I need to ensure I post forecasts that are at least 70%-80% accurate per week.
I just have a knack for doing things the right way.Sounds like you're doing a good job. Here is an idea for you, why don't you try to extend the forecast horizon, so, instead of forecasting tomorrows high and low, you forecast the high and low 5 days ahead, so, if today is day one, you produce the high and low for day 5. That way, you can get a look at the high and low for tomorrow as well as how things may change as days go on.
Sounds like you're doing a good job. Here is an idea for you, why don't you try to extend the forecast horizon, so, instead of forecasting tomorrows high and low, you forecast the high and low 5 days ahead, so, if today is day one, you produce the high and low for day 5. That way, you can get a look at the high and low for tomorrow as well as how things may change as days go on.
I'll take this into consideration.
Obliged
I'll take this into consideration. Obliged
One idea I have filed away for a day when I have more time to do analysis is almost like that. To extend the forecast horizon to paint a more clearer and long term view of the market trend.
forecast horizon --- what are you talking about
how about this one, could be easy -- seems to be @
forecast horizon --- what are you talking about
The forecast horizon is just the number of steps ahead that is being predicted. Such as, if I was to forecast tomorrows data given the information up to and including today, that's a forecast horizon of 1. If I were to predict the values 5 days into the future (and we're working off of daily data), that's a forecast horizon of 5.
I'm researching ways to model stochastic processes (pure mathematics) and as part of the PhD, I'm going to need an application for my work, the major applications I'll be using will be stock and foreign exchange. There'll also be interest rates, options, hydrological data, astronomical data, the list goes on, just the major ones will be financial. I'm not making a trading system, I'm not expecting this to be used for a trading system, I'm only doing financial modeling with the possibility of it being used in other areas (such as astronomy).
Many research is now in main stream use such as the models used to price derivatives such as options. They are not trading systems, they are financial models, which is what I'm working on.
.Hi Chewyraver,
Do a google on Black-Scholes model.
It will give you a lot of information and will give you a direction where to look further.
It discribes an econometric model that needed to predict what would happen economicly and the major instruments used where options.
You gonna find interesting stuff that you can use in your studies.
It was considered as a model that would never fail. But it proofed also the moment that you think you can predict then there will always be new elements that devellop or show up that will proof that it is impossible to predict.
They are looking at this moment to find the gods particle but up till now and there is no gods indicator or you need to devellop it. It is also know by the name cristal ball or madame soleil.
Friendly regards...iGoR.
PS. the more you studie the more you will get proof that it is impossible to predict with a certainty higher then 55% on whatever kind of event. Not the direction of tomorrows bar, not the next hour not the next minute. Whatever that people will try to make you believe
Hi Chewyraver,
Do a google on Black-Scholes model.
It will give you a lot of information and will give you a direction where to look further.
It discribes an econometric model that needed to predict what would happen economicly and the major instruments used where options.
You gonna find interesting stuff that you can use in your studies.
It was considered as a model that would never fail. But it proofed also the moment that you think you can predict then there will always be new elements that devellop or show up that will proof that it is impossible to predict.
They are looking at this moment to find the gods particle but up till now and there is no gods indicator or you need to devellop it. It is also know by the name cristal ball or madame soleil.
Friendly regards...iGoR.
PS. the more you studie the more you will get proof that it is impossible to predict with a certainty higher then 55% on whatever kind of event. Not the direction of tomorrows bar, not the next hour not the next minute. Whatever that people will try to make you believeYes, I know about the Black-Scholes model. It was created to price options. It was not intended to be infallible, it has changed many many many times to accommodate various other financial derivatives.
btw iGoR, you can predict with more certainty than 55%. I can assure you that it is done all over the place, all the time. The question is how you use the forecasts, not their percentage of correctness. Also, most forecasts cannot be assigned a percentage, they are regression problems, not classification problems.