Multi timeframe indicators, what do you think? - page 5
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
Hello @Nikolai Semko
Beyond your pretty animation, as I have seen many very artistic ones on your profile I would like an explanation of what you could call "dynamic timeframe", please.
Honestly guys, I've trained and compared 2 AI models to predict the future price of an asset.
Good morning
interesting.
Can you tell us more about the strategy that was applied? Could it be related to the fact that the performances were identical with or without timeframe?
What is a Timeframe?
that's for sure something nice if you can zoom between a summarized view of the market and a "real time" view instead of pressing timeframe buttons, but maybe would be more suited for manual trading? For automation it seems a little tricky
Good morning
Hello @Nikolai Semko
Beyond your pretty animation, as I have seen many very artistic ones on your profile I would like an explanation of what you could call "dynamic timeframe", please.
I don't even know what to add to the animation I sent, which very clearly demonstrates what a dynamic timeframe is.
Okay, I'll try another way.
The price has a fractal nature, like a coastline. No matter what height you look at the coastline from (from the height of 10 meters or from space), these lines will look alike.
Here is an example of the hierarchical structure of parabolic channels on EURUSD at the current moment in time. There are 13 channels in the hierarchy. The length of the junior channel is 3 hours, and the senior one is 21 years. All channels are alike.
Option with a nested junior channel (which is higher in the channel table, you can see it by its color)
Now imagine the most primitive trading system on one MA indicator on the M1 timeframe with a fixed MA period. Obviously, the efficiency of such a trading system will be zero and it will work extremely unstable, since the market is constantly changing.
But if you change only one thing - let the fixed period of the MA become dynamic and change every M1 bar according to some formula (for example, let it be a Fourier series of 20 harmonics, where each harmonic contains three parameters - amplitude, phase and frequency), then there is such a set of coefficients of this formula for a certain time range of trading (in this case 60 parameters), that the trading system on one MA will be super effective.
The market is not a static system in which you need to use fixed periods or fixed timeframes (sets of fixed timeframes), but a multidimensional dynamic system, as can be seen from the hierarchy of channels above.
It is important to note that the concepts of TF (timeframe) and the period of any indicator are related to each other. For example, MA with a period of 100 on M1 is equal to MA with a period of 20 on M5. 99% of all trading systems have a whole set of input parameters-periods that require optimization on historical data. Optimization is a beautiful word, but it would be correct to use another term - "adjustment to historical data". Obviously, since the market is a dynamic system, then a customized set of these parameters for one history will not work as effectively in the future. Therefore, any parameters related to TF or period are evil.
A correct trading system should not depend on the timeframe and period. For example, my trading system does not depend on the timeframe and is period-independent. All parabolic channels are recalculated each time the High or Low of the current M1 bar is updated. This provides complete independence from the choice of period, since this period is calculated constantly.
All fixed TFs are conventions and limitations. For example, there is a period, H1 and H2. Why not H1.0334 or H1.765? There is an infinite number of periods, as well as TFs. 4 years ago I gave an example (https://www.mql5.com/ru/forum/321704/page3#comment_13133599) on the Russian forum of how much bars with the H1 timeframe (3600 seconds) differ from the H1.00028 timeframe (3601 seconds).
that's for sure something nice if you can zoom between a summarized view of the market and a "real time" view instead of pressing timeframe buttons, but maybe would be more suited for manual trading? For automation it seems a little tricky
This dynamic timeframe visualization is of course created for the trader. The robot (EA) does not need visualization. It only needs an array of M1 bars or tick history. In my trading system, I use service index timeframes to increase the speed of calculations and visualization, but these timeframes are not at all like the standard, familiar timeframes. These are 7 timeframes: M1, M4, M16, M64, M256, M1024, M4096. Such timeframes are much more convenient for calculations and they do not have time collisions, which are found in standard TFs due to weekends and missed bars.
@Nikolai Semko
I understand your reasoning, but the application does not seem to follow through on the idea.
You say
" It is important to note that the concepts of TF (timeframe) and the period of any indicator are related to each other. For example, MA with a period of 100 on M1 is equal to MA with a period of 20 on M5 "
It is as true as it is false and for me even more false.
Indeed, for an SMA average, it can do it, but it is not as precise.
Moreover, doing it like this implies a symbol that is open 24/7.
because on a symbol that will only open for a few hours 5/7, the gaps will create a shift.
If you take an exponential that will give more weight to the last bar, it will be even more false, because the weight of the last M1 is not worth the weight of the last M5 bar.
If we take a smoothing which will take into account a volume like a WMA, VEMA, etc. the same and it is even more false.
I'll take an example where this is totally false.
An RSI 14 on M5, cannot be replaced by an RSI 70 on the M1.
We don't have the same results at all, even from a very long distance.
Other indicators must have the same concern. They will not be replaced by a times 5 or other to go from an M5 to M1.
Where I agree with you that TFs are not necessarily the best representation of a symbol.
But everyone gets caught up in it. Traders, investors, algos etc.
The TFs are therefore legitimate.
Besides, as I showed above, a lot of indicators are calculated on these TF bases. We can't reinvent everything and it's not recommended, because personally I want to follow the biggest ones, and the biggest ones must make their decisions on a TF definition
I also agree that some bars are missing...
I can understand your approach, but it is off topic for this discussion.
Good luck
Now imagine the most primitive trading system on one MA indicator on the M1 timeframe with a fixed MA period. Obviously, the efficiency of such a trading system will be zero and it will work extremely unstable, since the market is constantly changing.
You may be a fan of parabolic channels, but in my opinion saying a fixed period on X timeframe is unstable is not really true depending on the trading system. If you would trade based on the direction of the moving average then yes - inefficient. But a market crossover of a longer period MA is an efficient trading system even on the M1 timeframe. It's even more efficient if pending orders are used close to the MA line instead of market orders.
The Market crossing a 200 day MA is going to give a clear signal of a breakout or pullback.
You may be a fan of parabolic channels, but in my opinion saying a fixed period on X timeframe is unstable is not really true depending on the trading system. If you would trade based on the direction of the moving average then yes - inefficient. But a market crossover of a longer period MA is an efficient trading system even on the M1 timeframe. It's even more efficient if pending orders are used close to the MA line instead of market orders.
The Market crossing a 200 day MA is going to give a clear signal of a breakout or pullback.