The appropriateness of using Take Profit and Stop Loss in automated TS. - page 2

 
Figar0 писал(а) >>

Perhaps I have confused everyone), technical stops and takeovers are of course necessary. Perhaps, I wanted to know something else: its ability to work profitably without take and stop may not be a measure of system's performance. The use of stops and takes as part of a system, although it allows improving the system performance in some cases, actually leads us astray regarding the robustness of a system. My point is this.

Some time ago, in a similar argument about the need for stops,

I gave you this example:

- Open up your entire account history and put it in excel...

and recalculate all the losing trades as if you had

stop-loss for 20-25 pips, and look how much better your deposit looks... ;)))

*

It is clear that the direct recalculation with lots will be somewhat distorted,

but it's not a problem to correct it for accuracy...

The main thing is that there is a mathematical benefit from using stops!

That's all we need...

*

It's a classic, and of course some tactics and strategies are possible,

but I think it follows from the rule about the need for stops...

 
kombat писал (а) >>

Some time ago, in a similar argument about the need for stops,

I gave you this example:

- Open up your entire account history and transfer it to excel...

and recalculate all the losing trades as if you had

stop-loss for 20-25 pips, and look how much better your deposit looks... ;)))

what about profitable trades? where profits were taken after a drawdown of more than 20-25 pips?

- i would not have made them, instead i would have made a loss :-)

As for me, the real results are similar to the above-mentioned ones, but in any case they should be taken into account when evaluating an order.

I decided not to (at least for the majority of strategies).

Figuratively speaking - the use of TP and SL is an attempt to bury my TS (trading system) in a "pipe" (everything outside the pipe is not ours).

The system itself must monitor the market and know when to close. And it is known to be much more difficult than to enter correctly.)

IMHO of course.

 
xeon писал(а) >>

what about profitable trades? where profits were recorded after a drawdown of more than 20-25 pips?

- I would not have had them, instead I would have had a loss :-)

As for the other Sociometry, I've already traded with a lot of people.

I decided not to (at least for the majority of strategies).

Figuratively speaking - the use of TP and SL is an attempt to bury my TS (trading system) in a "pipe" (everything outside the pipe is not ours).

The system itself must monitor the market and know when to close. And it is known to be much more difficult than to enter correctly.)

IMHO of course.

Takes.

One of the tenets of trading states:

- Increase profits and limit losses.

Of course I use Takes, but very rarely, more often I place

A pending order with opposite direction (lock).

*

In automation, for a more efficient exit and minimization of troubles with a lot of positions

the most acceptable are trailing stops, classic trailing stop and the

The variant of "trailing profit" (or equity) for portfolio trading...

(as well as for classics too...)

*

Your opinion that stops should be adaptive, or otherwise, is correct:

- watch the market

There are no particular problems in its implementation, the matter is to develop the algorithm,

and it's no problem to put it into code ... ;)))

 

In my opinion, the effectiveness of stops and takes depends on the ideas behind the TS. If for example the system works inside a channel in fast markets, then stops and take's provide obvious advantages.


From another purely theoretical point of view, if the system has constant stop and take levels for each order, then knowing the estimate of the ratio of profitable trades to losing ones, you can relatively accurately calculate the probability of getting any drawdown level. However, the practical value of this possibility is rather doubtful, unless you have no guarantees that for some time the market will allow your TS to work with the same percentage of profitable trades.


In the case where stops/stops are traded or positions are closed on the market, we are dealing with a much greater inaccuracy, because we have to use estimates of the average level of realized stops and takes.

 

Generally, by its nature, take profit is applicable to systems that operate according to the following principle:

1) pips (exactly pips, not scalpers)

2) hand trading:

2.1 for those who know the bible, enter the trade and then pray to reach the take

2.2 purely intuitively determine the direction and set a take profit on a small number of points. Then we go out for tea.

2.3 Trading based on noise and news - take a close take profit, and a trader closes or rolls down the terminal, to avoid nervousness.

The list can go on. In any case take is used when the trading strategy does not have any clear exits. And detecting take value in the optimizer (with a range of 5-100 points) IMHO is a crisis in the trading system. At the very least it is applicable for normalization or for picking statistics on profit. Not more.

The use of stops seems to me necessary. Although everything depends on the system. It is difficult to operate with hypothetical notions, therefore I will give an example of what I am currently working on and what I am planning to do. As such, the stop is placed far enough - accumulation of levels on Phoebe+last unbroken level (i.e. Sapport - resistance on large TF). The role of stop - avoiding force majeure.

In fact, of course there is a loss. But unequivocally there are no such unpleasant moments when the close stop is triggered, and the price turns around and goes in the right direction. The system is not reversible, i.e. there is a hard exit from the transaction (either with profit or with loss). Large drawdowns are excluded because of the work on a small TF. The only key (and at the same time a classic illustration) - work in a narrow range. This is where all efforts are directed.

And again, identification of a stop in the optimizer is, IMHO, a questionable activity. In one case it is avoiding unnecessary losses, and in the other - just bigger than usual moose.

 
Figar0 писал(а) >>

Lately I try to develop systems that do not need stops and takeovers, only a control function. The system lives with the market and tries to have its "opinion" on any situation and at any time it gives one of 4 signals (two to close positions, two to open). I think a "correct" system should act like that, and take and stop is a relic of "manual" trading. However, something is eating me... Am I rejecting something useful and meaningless? I am interested in your opinions, colleagues, maybe some links or ideas.

Stops and TPs, as the deal targets, are not required for this system. The system "knows" what and where, so why should we shackle it? The anti-catastrophic limiters we were talking about are needed, of course.

 

The Expert Advisor I sent to the Championship uses both Tics and Stops, but that's for the Championship (distances are re-calculated every time)... What am I getting at?

My point is this: to determine whether we need stops (maybe lots, if someone prefers them, but I don't like them), we should first decide whether the system is fully automatic or semi-automatic (technical stops don't count). From personal experience, when working with a semi-automatic, such greed or opposite generosity limiters are very useful. When we have an automaton, it is much more logical to use the opposite or argumentative signal to exit. IMHO, of course...

 
it depends on the input data and automation, i.e. time delay to exit:
consider simple MTS.
- If MTS is based on muwings and their derivatives, then the only way to get a profit is to use a stop to close a slow TS,
(output signals are integrated)
- if MTS is based on low-latency data, then neither a take nor a stop is needed, and no trail will do any good, i.e. they worsen TS, they are present only as a safety feature.
- If MTS is scalping, then it needs a TP by definition)), while a stop will be set at 1000((, - only as insurance, and you can't go closer than that, it will kill all profits.
 
xeon писал(а) >>

What about profitable trades? where profits were recorded after a drawdown of more than 20-25 pips?

- I don't think we would have had them, instead we would have had a loss :-)

You have to calculate the ratio. I.e. when the floating loss reaches 20 points, with what probability it will decrease and turn in the profit, and with what probability it will grow to some "unacceptable level".

xeon wrote >>

As for me, I also spent a lot of time trying to decide for myself whether or not to use takeoffs and stops (other than those technically necessary).

I decided not to (at least for the majority of strategies).

Figuratively speaking - the use of TP and SL is an attempt to bury my TS (trading system) in a "pipe" (everything outside the pipe is not ours).

The system itself must monitor the market and know when to close. And it is known to be much more difficult than to enter correctly.)

IMHO of course.

An idea. Probably, if the presence of a "pipe" improves the performance of the system, it indicates that the system is too simple and does not take into account all important situations. In fact, TP and SL are stubs that allow a simpler TS algorithm to be used, albeit at the cost of a potential loss of profitability.

...And since no TS is able to account for all market situations, then... ;)))

 
Figar0 >> :

It is suggested that we briefly discuss the subject. I have been puzzling over this question for a long time and have not been able to draw the "right" motivated conclusions for myself.

----------


I made a simple experiment, considered three cases

1) without stop-loss, following the signal

2) with a stop-loss but after the stop-loss triggered a pending stop should be placed in the same place as the first one and this should be done until I get a signal to stop the order

3) with a stop-loss order and after it triggers, the order finishes its existence

The signals for entry and exit are good, but in one place there is a big drawdown, parameters of the stop loss have not been optimized.

Can any conclusions be drawn from these drawings about the need for a stoploop?


1 variant without a stop

Net profit32453.46Total profit52687.80Total loss-20234.34
Profitability2.60Expected payoff66.23
Absolute drawdown1656.90Maximum drawdown14604.66 (44.43%)Relative drawdown44.43% (14604.66)
Total trades490Short positions (% win)227 (60.79%)Long positions (% win)263 (79.85%)
Profitable trades (% of all)348 (71.02%)Loss trades (% of all)142 (28.98%)
Largestprofitable trade995.96losing deal-572.93
Averageprofitable deal151.40losing deal-142.50
Maximumcontinuous wins (profit)57 (6194.59)Continuous losses (loss)18 (-2229.50)
MaximumContinuous Profit (number of wins)17306.84 (19)Continuous loss (number of losses)-6653.33 (17)
Averagecontinuous winnings11continuous loss5


without a stop


2 variant with stop and continue

Net profit26829.02Total profit53752.63Total loss-26923.61
Profitability2.00Expected payoff37.89
Absolute drawdown2508.64Maximum drawdown12291.59 (42.67%)Relative drawdown42.67% (12291.59)
Total trades708Short positions (% win)274 (49.64%)Long positions (% win)434 (48.39%)
Profitable trades (% of all)346 (48.87%)Loss trades (% of all)362 (51.13%)
Largestprofitable trade1073.40losing transaction-106.68
Averageprofitable deal155.35losing trade-74.37
Maximum numbercontinuous wins (profit)57 (6194.59)Continuous losses (loss)122 (-10279.91)
Maximumcontinuous profits (number of wins)18805.48 (19)Continuous loss (number of losses)-10279.91 (122)
Averagecontinuous winnings9continuous loss10


with stop and continuation

3 variant with stop and no continuation

Initial deposit10000.00
Net profit17706.70Total profit32485.31Total loss-14778.61
Profitability2.20Expected payoff36.14
Absolute drawdown1642.64Maximum drawdown4125.05 (13.81%)Relative drawdown22.80% (2708.72)
Total trades490Short positions (% win)227 (51.10%)Long positions (% win)263 (62.36%)
Profitable trades (% of all)280 (57.14%)Loss trades (% of all)210 (42.86%)
Largestprofitable trade574.46losing deal-88.12
Averageprofitable deal116.02losing trade-70.37
Maximumcontinuous wins (profit)57 (6194.59)Continuous losses (loss)28 (-1899.27)
Maximumcontinuous profits (number of wins)6194.59 (57)Continuous loss (number of losses)-1899.27 (28)
Averagecontinuous winnings9continuous loss7


with stop without continuation