Combining trend and flat strategies into one TS = grail? - 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
it all comes down to a future trend/flop filter. If the filter indicates a trend, the targets are big, the stops are small, the flop is the opposite. I.e. in simpler terms it's tp/sl control. If this ratio changes only when a new position is opened, then this is simply combining trend and flat systems in the portfolio (even if they are in the same code, or changing the active system gradually via MM). If the ratio changes for already opened positions, then it is a combination of trend and flat system. I.e. we entered with a trend system, and exited with a flat system, for example.
What if you combine systems on two timeframes...
Suppose we have a low-profit trend strategy, which gives some kind of long-term profit, but there may be large drawdowns on the flops.
Suppose we have a channel strategy that is also not very profitable. Suppose it also eventually gives profit due to numerous small fluctuations but it shows greater drawdown on the trend.
We do not switch between them, but use both of them simultaneously and without pauses, each following its own algorithm, but on one currency pair.
We set the trend one on tf=1d, while the channel one on tf=4n. When there is a trend on 1d, 4n is enough of losses, on the trend of 1d we take large profits. When we have a narrow flat on 1d, for 4n this flat looks like a relatively wide channel and in this case the channel strategy will take many relatively small profits which compensate for the drawdown of the trend strategy. Right? All that remains is to find two marginally profitable opposing strategies :-?
Oh man, I do not know why we should use trend strategy and flat strategy! I would use the same trend strategy on 1d and on 1n at the same time. When we go flat on 1d, we make loss, and on 1n we look - there is a trend! Profit! Beautiful! And when 1N is flat, the 1d strategy simply will not open a single position. As the result we will get profit from trends with two strategies, while losses on flats are incurred only with one - with the smallest one :-)). Yes? And we don't need to look for two strategies, we just need to find one...
Who has thoughts on this? Without being specific, has anyone implemented "combining the incompatible" in a single TS? Not an alternation of one with the other, but a complete and organic combination of a flat strategy and a trend strategy in a single TS ... What are the principles and characteristics of such a "hybrid"?
Depending on the strength of the trend, set the function determining the severity of entry/exit conditions
The stronger/clearer the trend, the milder the conditions for making a deal
How to quantize the trend - it is up to you to decide on the specifics of your technical analysis
That's the thing about trend/flat systems, you don't need to know the current state of the market. All a trend system has to do is make money on the trend and not lose money on the flat. A flat system is the opposite: it should make money on the flat and not lose money on the trend. Since both states change each other on the markets, the growth phases of one will be counted with the consolidation of the other, which gives a significant advantage in the end.
You should not try to combine both systems into one. It is better to use them simultaneously. In general, I'm against the philosophy of filtering. It's better to use several opposite TS instead of filters. The risks of one will be hedged by profit of the other, the total result will be summed up, and the number of deals will remain at the same level. Nothing of the kind can be achieved with filters.
Don't try to combine both systems into one. It is better to use them simultaneously.
Why not a hundred systems? What's the point of trifling?
It's better to use several opposing TSs at once instead of filters. The risks of one will be hedged by the profits of the other,
Who has thoughts on this? Without being specific, has anyone implemented "combining the incompatible" in a single TS?
Why incompatible? The negatively correlated ones, if they have these same non-linear correlations, are very complementary.
I don't think anything, but specifically combined.
What are the principles and signs of such a "hybrid"?
The signs are elementary: a synthetic instrument without drawdowns on history, i.e. it shows only profits.
But in order to make a profit not only on the history, but also on the forwards, one should follow the rules. Namely, any negatively correlated financial instruments can be combined into a no-loss portfolio - to fit it. However, not any instruments, but only those that have real and stable negative correlations, for example, securities of competitors, will yield profits in forwards. The more competitors' securities in the portfolio, the higher the result.
Such tricks will not work with currencies - tested, forwards do not hold. Competitor's shares or futures on interchangeable commodities or raw materials are suitable.
As for TS with negative correlations, i.e. when some traders are profitable in a sideways trend and others are losing on the contrary, as it was said above - anything is possible to fit into the story. Another thing is the stability of negative non-linear correlations.
The other compote is the stability of negative non-linear correlations.
Explain the non-linearity.
But on forward, not any, but only those with real and stable negative correlations will give profits
The correlation may also be positive. What tests do you use to analyse correlation stability?
You combine two hemispheres - left and right - open your eyes and look at the chart!
You find patterns and trade them.
Explain the non-linearity.
A simple example with two negatively non-linearly correlated TS:
When the first TS earns, for example, 10 pips, the second one loses 6 pips
When the first TS loses 10 pips, the second one earns 7 pips
I.e. +/-10 pips of the first system have a different number of pips of the second one depending on whether the first system is losing or not.
In this case we set both systems to trade with fixed lot, the first 1.3 lots, the second 2 lots
Here we have it:
- The first one is winning, the second one is losing: 10 * 1.3 - 6 * 2 = +0.1
- the first one loses and the second one gains -10 * 1.3 + 7 * 2 = +0.1
We obtain a no-breakage synthetic TS that, on average, always trades in profit - a constant positive mathematical expectation. Gamblers call such situations forks.
If x pips in any direction of the first system results in -n * x pips of the second one, then this is a linear negative correlation. It is impossible to earn anything stably. The simplest example of a linear negative correlation is two blocked positions at one and the same symbol. No matter how one sets the lots, it will not be able to equalize them to drawdown-free, because at different lots the system is risky, while at equal ones it is blocked.
A simple example with two negatively non-linearly correlated TS:
When the first TS earns, for example, 10 pips, the second one loses 6 pips
When the first TS loses 10 pips, the second one earns 7 pips
Such TS have Spearman rank correlation coefficient equal to -1. Pearson's rank correlation coefficient is about -1.
This is a very high degree of correlation characterized by Pearson's CC. Such trading with highly correlated TS, as you know, is a kind of pair trading.
If x pips in any direction of the first system is equal to -n * x pips of the second one, it is a linear negative correlation.
And here Pearson's AC is already equal to -1. It is true that one cannot earn anything on it.
That is, it is the stable(MO = const) and high (but |KK| < 1) linear correlations that can be earned. However, even the stability condition can be missed.