Topic: Can predict indicator - need trading idea

Hello, fellow traders
I've got the following case.
Recently I'm working with ANN in order to use them in Forex trading. Currently I've received some intermediate results - I can predict the movement of indicators with 75%+ accuracy (on some indicators up to 90%) - for example, my ANNs can forecast the moves in sma or rsi one-two days ahead pretty accurately.
But here I've stuck with the following problem: how this knowledge can be transformed into any useful strategy? The question is: whether its possible to transform future values of indicators (which can be predicted quite accurately) in some reliable trading signals?

http://www.darkmindfx.com - download free COT report and US economic indicators for your platform

Re: Can predict indicator - need trading idea

Perhaps using different time frames, one for the forecast and another for the entry.....  (a wild guess)

My 'secret' goal is to push EA Studio until I can net 3000 pips per day....

Re: Can predict indicator - need trading idea

i didnot understand you dark

the data you used are from past and the results is the future for the backtested data but that future is the present ( when trades are taken)

4 (edited by Irmantas 2016-07-13 09:01:53)

Re: Can predict indicator - need trading idea

Hi dark,
Will try to add some guesses smile Can be very wrong. I think it should depend on specific indicator. For example:
If SMA is higher BUY,
If SMA is lower SELL smile

If RSI is oversold and your ANN shows change UP, and market is in range BUY
...

Maybe adding some threshold filter would be even better. If indicator moved by so much, you enter, if your ANN shows just a little move to your indicators, no entry. However most indicators are lagging, trying to catch up with price move by some period, so interestingly is it enough to out weight that by your 75%+.

Re: Can predict indicator - need trading idea

ahmedalhoseny wrote:

i didnot understand you dark

the data you used are from past and the results is the future for the backtested data but that future is the present ( when trades are taken)

Hi ahmedalhoseny
I'll try to shed more light smile
For example, right now is the beginning of the day (let's say 1 AM). I can take SMA data for previous 5 days (for examle % change for each day), feed it to my ANN and it will say "next day (meaning today) he SMA will go UP for X %" -

http://www.darkmindfx.com - download free COT report and US economic indicators for your platform

Re: Can predict indicator - need trading idea

Irmantas wrote:

Hi dark,
Will try to add some guesses smile Can be very wrong. I think it should depend on specific indicator. For example:
If SMA is higher BUY,
If SMA is lower SELL smile

If RSI is oversold and your ANN shows change UP, and market is in range BUY
...

Maybe adding some threshold filter would be even better. If indicator moved by so much, you enter, if your ANN shows just a little move to your indicators, no entry. However most indicators are lagging, trying to catch up with price move by some period, so interestingly is it enough to out weight that by your 75%+.

Hi Irmantas
Yes, I've tried sometging similar to what you've said - but only with SMA. So I've made an assuption that SMA will go the same way as the price (or visa verse). In general I cannot say that it was really successful strategy - for example for SMA(20) I've received 56% of correct predictions - e.g. in 56% of cases the direction of SMA predicted was the same as for the price. For shorter SMAs (like 5, 10) it was almost 50% - so no difference from just tossing the coin.
But I like your idea to combine several indicators and the threshold (I presume especially threshold) to check whether the signal is strong enough. For example it can be linked to avrage day range of the last month/quarter

http://www.darkmindfx.com - download free COT report and US economic indicators for your platform

Re: Can predict indicator - need trading idea

Hi Dark Mind,

Which ANN software are you trying or testing now? Neuroshell or Trading Solutions or some other software? What's your general feedback or comments about the software performance?

Re: Can predict indicator - need trading idea

It is very easy to predict the direction of SMA, but I don't think it is easy to predict the price.

It took me only 15 minutes to write an EA that predicts the SMA with 90% success.


http://s32.postimg.org/9r9x23ts5/screenshot_1918.png

The EA is attached below. (v2 draws arrows smile )

Post's attachments

MA Prediction.mq4 3.8 kb, 7 downloads since 2016-07-18 

You don't have the permssions to download the attachments of this post.

Re: Can predict indicator - need trading idea

Based on the above EA I tried to predict the Close price. The success was only 47%;

http://s32.postimg.org/qwwkkge4l/screenshot_1919.png

Re: Can predict indicator - need trading idea

Popov wrote:

Based on the above EA I tried to predict the Close price. The success was only 47%;

That's what I'm saying - I've got pretty the same case. But the thing is not to predict price itself but - how to use this predictions in trading? I.e.: if we are using tech indicators in trading, and if they help us (do they) - why can't we make a quick look into the future of these indicators and try to trade based on this future? Indicators are just a mirrors reflecting price at different angles. So we can have an ability to know what this mirrors will show us let's say tomorrow.

P.S. Regarding predicting SMA vs predicting price. I think we shoukdb't rely only on one indicator - we should have bunch of predictions and process them respectively to assess the price. It could be for example average of the predictions based on these indicators or something like that.

http://www.darkmindfx.com - download free COT report and US economic indicators for your platform

11 (edited by hannahis 2016-07-31 11:59:17)

Re: Can predict indicator - need trading idea

DarkMindFX wrote:

That's what I'm saying - I've got pretty the same case. But the thing is not to predict price itself but - how to use this predictions in trading? I.e.: if we are using tech indicators in trading, and if they help us (do they) - why can't we make a quick look into the future of these indicators and try to trade based on this future? Indicators are just a mirrors reflecting price at different angles. So we can have an ability to know what this mirrors will show us let's say tomorrow.

Predict Prices

If I can predict prices, I can then know what trade/positions to take.  If I know the prices of EURUSD is going up, of cos it will influence what types of trades/positions to take.

Can tech indicators help us?
I can't imagine anyone trade without using any indicators.  Use intuition?  Use guessing game?
Of cos tech indicator can help us...it's depends on which appropriate indicators/parameters we are using to measure.

Indicators are just mirrors reflecting price at different angles...ability to show us the future?
Let's get it right first of all.  It is price that influence the curves and movement of the indicators, not indicators influencing/directing the prices.  When price move, the indicators curve/graphs on your MT4 moves or cross over.  When price don't move, indicator graph don't change.


Can we use indicators to predict the future?
Yes of cos.  Or else how do we trade or make our trading decisions?

Forecasting/prediction vs Actual Trading
Anyone fool can forecast “correctly”, it’s a matter of timing. The price/trend will go up…it’s a matter of when (after how many downs)?

Forecast and place actual live trades with stop losses and managing that trade to profitability have NOTHING in common. Timing is everything. It works, but you must understand that Timing needs confirmation. Price and volume. If you can understand Timing works. Even poor technique works if timing is good… without timing, nothing works.”

Forecasting and placing actual live/demo trades with good high entry/exit accuracy requires totally different level of skills.

Anyone can forecast…it’s a matter of when to place the trade and when to exit. I can forecast that EURUSD is going up…it’s a matter of when.

What makes using tech indicator hard to predict future prices or make future trading decision is because...we don't know which direction the prices will go in the future and how much is the rate of changes etc.

When I look at my indicators and see a "cross over" is about to happen.  I can predict that the price is going up or down "soon" but how "soon"? I don't know, it can be the next hour, the next day or the next week....cos it all depends on how "fast" or "slow" the market/price move.  And since price don't move in consistent speed or direction, I can't tell when will the "cross over" happened.  But I can tell that when the "cross over" does happen, a breakout is very likely to happen and I  can make good pips at this juncture (prediction vs actual).

Essentially, every trade we place, we are "guessing/predicting" via mathematical calculation/probability/indicators etc the direction of the price will go.  We can either predict and enter before the "cross over", enter at the "cross over" or enter after the "cross over"....hence Timing and Prediction are utmost importance to be profitable.  Lastly, exiting or predicting when trend will change is another part of the prediction/trading strategy.  Without good entry and exit plan, there is no profitable EA.

Hence we can see, Forecasting (in terms of giving a general prediction of price direction) isn't enough to be a successful trader.



a) If a system can only forecast but can’t automatically trade for me, that system is rather “useless” to me. What’s the point of paying so much for a system when it can only forecast but can’t automatically execute for me the trade with high entry/exit accuracy. If I pay for a system, it is to trade on my behalf so that I don’t have to sit in front of my computer, staring at the market/software to wait for the “right” timing of entry/exit.

b) When a system that can only forecast, it is easy to blame the customers for poor results because the blame is shifted to the customers for entering wrongly or not reading the indicators correctly.

c) You can be 70% accurate but the other 30% when you are wrong, can wipe out all your 70% profit and make huge losses. Predictive power is not good enough to be profitable.

d) It take more than good predictive skills (of the trader or software) to be profitable. Besides, good predictive power, a system/trader need to have high entry/exit accuracy and good management. Predictive power is only a small part of the profitable equation.  Entering/exiting a trade with good timing makes a whole lot of difference between profitable and losing.

IF a software company claimed to have a set of indicators and algorithm that can calculate when is the window of opportunity, whereby I can predict the highest probability and learn trading setup so that I can start using right away to “be there” when banks “open up their vaults” to trade…

I wonder what is stopping this software company from turning it's trading theory into trading rules for automated trading?


Is it the lack of technical expertise or resources to transform these trading theory into specific rules to scan the market at all times, looking for the “correct” or high probability market entry/exit and execute these trading rules via an automated system?

Is it the lack of faith in it's own system?

Is it the fear of facing the truth that the system isn’t profitable after all?

What happened if a predictive software company  really turned it's trading theory into applicable concepts and make it available for automated trading?

The answer is very simple.  The customers will know EXACTLY how profitable the system is. No more guessing, no more blame shifting. All trading performance can be easily measured statistically and the customers will know for sure whether the system truly work or not.

But with a forecasting system…it’s hard to quantify the software’s success since it’s is all very SUBJECTIVE and DEPENDABLE on customers’ level of understanding of how to apply the theory or concept, dependable on the timing of entering and exiting from the trade, money management, and a whole lot of many other factors which a software company can claim why the software or system fail…it’s all the customers’ fault, lack of skill, lack of trading, lack of discipline, lack of whatever you can think of…bottom line is, such software companies can always hide behind this excuse of SUBJECTIVITY, to explain why customers aren’t getting the kind of success rate they hope to get.

However, in a fully automated trading system. There is no grey areas…no more pointing of fingers. The automated system is fully quantifiable without anyone to blame. It is either profitable or not. Straight forward and objective.

Honestly, if I have such powerful predictive indicators as what some companies claimed to have and given the team’s ability to “program” unique algorithm etc…nothing would have stopped me from going another step further to transform these highly predictive indicators or trading theory into automated system. It definitely require a higher level of skill to create highly profitable automated system. It would definitely put one’s trading theory to the ultimate test…with 1 final verdict, profitable or unprofitable (if profitable, what’s the win/loss ratio, Profit factor, sharpe ratio etc), all these quantifiable measurement of success, nothing subjective, no grey areas, straight forward results/performance.

Thus I'm highly skeptical of those companies that sell "forecasting" or predictive or trade signals etc and invite their customers to their trading rooms to see how they use those indicators to trade etc.

1. Why waste so much time attending those trading room sessions to learn how to trade when you paid so much for a system that can't even trade for you?

2. Why some of these customers can attend these trading rooms and still didn't know how to trade on their own?

3. Essentially these companies is telling their customers that without their indicators to help them predict and with their mentorship, these customers can't trade confidently on their own.

4. If I pay so much for a predictive system and yet still have to watch the market to enter the trade and at the end of the day still clueless of how to trade, I might as well save up the money paid to the software company and use it as my initial capital for Zulu trading, at least I can choose which trade signal provider to follow with proven past performance/results.  This forecasting companies hardly produce any of their own live/real account to show how profitable it is to use their systems.