Failed forex strategies, opinions and reasoning. - page 8

 
Yuriy Asaulenko:

Well, it's akin to secret knowledge. )) As it goes: two analysts - three opinions. Oil futures price forecast based on macroeconomic indicators. Are you kidding? I have a 15 minute trade from open to close. Several times a day going long and short. According to your theory we should have all been bums long ago.))

But here's the thing I agree with, TA doesn't help))

Well you don't know how your system will work further, maybe you just adjust the model to the current conditions and you will be lucky for a while. I had systems that showed great profits for more than a year on history, and 2 months on forward. It has to do with optimization. And then they stopped working and that's it. I am not talking about systems that give 1-5% monthly, with such small risks luck may last a very long time, and if we want to earn tangible money and quickly wrap the capital, the risks multiply with this optimization approach.

The coolest thing I managed to do through chart analysis is 50000% for a year on backtest and 1800% profit for 2 months, after that that's it... :) But there was an obvious one year collapse of the euro against the dollar, with almost no significant corrections, and now from standing in a flat. And the same situation can be repeated, say, once every 10 years on average, but that's not certain :)

 
Maxim Dmitrievsky:

so parameter optimisation works for a very short time and there is no way of knowing when a new optimisation should be carried out and when the market has changed, and it changes instantly under the influence of external factors

I already said in the next thread that you need to analyze what the market changes and when these changes occur. This data is much more constant than what it affects.

It's like looking at a child's parents and assessing their behaviour in order to understand the child. These children are very similar to their parents, for example, if there is fighting at home, the child is fighting too.

 
Lilita Bogachkova:

In a neighbouring thread I have already said that you have to analyse what the market changes and when these changes take place. This data is much more constant than what it affects.

It is like looking at a child's parents and assessing their behaviour in order to understand the child. These children are very similar to their parents, for example, if there is fighting at home, the child is fighting too.

Yes, or go down to the dna level of the parents and do high-frequency trading, with lots of trades, trading small inefficiencies, with little risk per trade and excellent execution. These are pretty objective things, and that's where the main money in trading is spinning right now
 
Maxim Dmitrievsky:
Well, you do not know how your system will work further, maybe you have just successfully adjusted the model to the current conditions and will be lucky for some time. I've had systems that have been showing excellent profit for more than a year on history, and 2 months on forex. It has to do with optimization. And then they stopped working and that's it. I am not talking about systems that give 1-5% monthly, with such small risks luck may last a very long time, and if we want to earn tangible money and quickly wrap capital, the risks multiply with this optimization approach.

As O.Bender used to say: I don't want an eternal primus needle, I don't want to live forever.

The lifetime of a system with a rigid algorithm without significant upgrading is 1.5 to 2 years. You can somewhat extend it by making the algorithm adaptive within some limits, it will be 2 - 2.5 years.

So, we all know). As for fitting. Models don't need to be fitted, they need to be built. A freshly cut system does not need to be optimised. Yes and the big question is; what is optimisation? What are the criteria? Finding the maximum profit by selecting parameters? - absolutely not serious.

 
Yuriy Asaulenko:

As O.Bender used to say: I don't want an eternal primus needle, I don't want to live forever.

The lifetime of a system with a rigid algorithm without significant upgrading is 1.5 to 2 years. You can somewhat extend it by making the algorithm adaptive within some limits, it will be 2 - 2.5 years.

So, we all know). As for fitting, I don't fit the model, I build it. A freshly built system doesn't need optimisation. Yes and the big question; what is optimization? What are the criteria? Finding maximum profit by fitting parameters? - Absolutely not serious.

Well, by optimisation I meant the selection of various variable parameters of the system, which can also be picked up through a neural network, yes, or through an optimiser. If we're talking about a more or less complex model, it should have a decent number of optimizable parameters/weights :) At least at the stage of model building
 
Maxim Dmitrievsky:
Since we are talking about a more or less sophisticated model, it must have a decent number of optimizable parameters / weights :)

To build a model, we use some kind of mathematical apparatus - say, statistics, etc.. Once it is built, it is ready to be used - the problem has already been solved. Then we give it to uncle Vasya the tester-optimizer (neural networks, etc.) and say - adjust parameters and increase the maximum profit! Profit may be maximal, but there may be nothing left of the system. And the profit will be only in the tester.

This is exactly how we would imagine a normal algorithm to be developed: 1. Modelling and development in external environment - MathLab, R, SciLab, Excel, etc. 2. Program writing in final language - by taste (MQL, C++, etc.). 3. Testing of the ready program on the interval of not more than 1 year, or better less, in order to find out if there are any errors in the program. 4. Work on virtual trades - 1 month. 5. Working with small lots - 1 month. 6. standard operation.

 
Alexander Laur:

2. the basis of trading is the choice of the direction of a position to be opened. It is mistakes in choosing the direction that reduce your deposit. Everything else: position volume, stop size, take size, etc., although important, is secondary. These secondary parameters are qualitative (direction error) and quantitative (% of deposit). Once again, the choice of direction is the basis of trading.

I wrote in another thread, but I'll repeat: it's not the direction of the movement that's most important, but the PURPOSE of the movement! Knowing the goal, the question of direction resolves itself. Knowing the Goal, the risks are calculated. And in case of an error, it is always necessary to have an alternative scenario. In the case of an error, it is also necessary to know the point the price should reach, i.e., the target.
In principle, knowing the both objectives (base and the opposite alternative), you can weigh the risks/benefits of both scenarios and understand, in which direction the price will go first, and in which direction it will go later. But there is still a risk of error, because not everything is in our hands. Or rather, we cannot influence anything.
 
Alexander Laur:

2. the basis of trading is the choice of the direction of a position to be opened. It is mistakes in choosing the direction that reduce your deposit. Everything else: position volume, stop size, take size, etc., although important, is secondary. These secondary parameters are qualitative (direction error) and quantitative (% of deposit). Once again, the choice of direction is the basis of trading.

The quality of directional choice depends on the goal achieved in the chosen direction.
 
Alexander Laur:

You know best. :)

Example: BUY with TP of 10, 20, 30, 50, 60 ... 100 pips and so on.

As long as the TP falls within the daily average volatility of the instrument, the choice of direction is less important than if the TP is outside the average volatility.

 
Alexander Laur:
The answer is simple enough: stick with this strategy for at least a few months. :)
Explain how you determine the TP distance if you have the same entry criteria for 10 pips and 100 pips.