Some signs of the right TCs - page 8
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
There is an obvious and trivial approach that any TS is profitable only when the market is in some suitable condition for it (on the smartlab, Gorchakov seems to hold such an approach). There is no universal system-greats and there cannot be.
A state can be described as a certain set of time series and we need to check our TS on the whole set. We need to make sure that the TS is tuned to the state as a whole and not to a particular series. But we only have a single series. We assume (hope) that a slight change (of a certain kind) in the series we have will leave it belonging to the same state as the original one.
I totally agree with that. There is a problem detecting this "definition area", but that's why the trader is attached to the TS :-)
I will argue about the possibility of transformations of quotes, though.
1. The price series cannot be shifted left or right because there is a calendar. And the resulting price series is rigidly tied to real time. At least, there are weekends, holidays and other periodic events, the reaction to which is different.
2. A series cannot be moved up and down by a constant - percentages will change. A change from 1 to 1.001 is one thing, and from 2 to 2.001 is something else - the market participants will react differently.
3. The series cannot be multiplied by a constant - there are regulatory mechanisms. There are regulatory mechanisms up to the fact that in some cases they simply close the trades, in others other interventions are possible.
Unlike abstract series, the time and price moments are very important for quotes. They (price, time) are not just coordinate points.
You and I have such different perceptions on many algorithmic trading topics that any attempt to get a grasp of your opponent's point of view is doomed to fail. And that's putting it very mildly.
Imho, a proper TS should show plus or minus the same results in different market periods. Let's say the TS is averaging 5%. If there is a trend in the market, there is no trend in the market, but it gives 5%...
My colleague writes about the state:
Forum on trading, automated trading systems and testing of trading strategies
Some signs of a good TS
Aleksey Nikolayev, 2020.02.29 12:01
There is an obvious and banal approach that any TS is profitable only when the market is in some suitable condition for it ... There are no universal systems-greats and there cannot be.
The state can be described as a certain set of time series and we should check our TS on the whole set. We need to make sure that the TS is tuned to the state as a whole and not to a particular series. But we only have a single series. We assume (hope) that a slight change (of a certain kind) to the row we have will leave it belonging to the same state as the original one.
It seems to me that we should not speak of a state, but of a model(say an econometric model) that describes the behaviour of the traded asset. Of course the model can be broken. But the probability of such an event should be small...
If we multiply the price by a constant, nothing changes. It is simply moving the graph axis up or down.
Another thing is to move the chart left or right. At first, it will affect the results, but after a while it will stabilize.
Once upon a time, 5-6 years ago, I collected real ticks, added random numbers to them in a reasonable range (using rand) and tested.
Results were terrible :)
I came to a conclusion: no one will be able to develop a fully automatic trading robot that will steadily show more than 3-5% profit per month.
If we multiply the price by a constant, nothing changes. It is simply moving the graph axis up or down.
Another thing is to move the chart left or right. At first, it will affect the results, but after a while it will stabilize.
Some time ago, 5-6 years ago, I collected real ticks, added random numbers to them in a reasonable range (using rand) and tested.
Results were terrible :)
I came to a conclusion: no one will be able to develop a fully automatic trading robot that will steadily show more than 3-5% profit per month.
If we multiply the price by a constant, nothing changes. It's just the axis of the graph moving up or down.
Multiplication is a vertical stretch.
Multiplication is a vertical stretch.
Yes, I wrote it wrong, the graph axis does not move during multiplication. And there is no point in doing so.
It's enough to test the robot with the same parameters on different symbols and that's enough to check its stability.
It's enough to test a robot with the same parameters on different characters and that's enough to check its stability.
Off-topic.
Off-topic.
Why off-topic?
When we increase the price or multiply it by a constant, the nature of the price movement does not change.