setting your stoploss

[michal.kreslik]

Hello, friends,

I would like to open a brief discussion on how does changing the stop loss in your trading system influences the trading system itself.

Recently I had a discussion on this theme with my friend. I expressed the simple and clear fact that all other factors being equal, then changing your stoploss means that you are changing the number of winning trades for your system. My friend didn't agree with me.

Anyway, an evidence that this is actually so is that based on the premise that you never know which way the trade is going to go between your entry and exit (if you know, then you're not having a trading system, but a coin mint ), you will have no winning trades if your stoploss was zero and you will have all winning closed trades if your stoploss was not set at all (infinity). Anything in between zero and not set (infinity) changes the number of winning trades less dramatically, but the statement is still true.

I'm open minded and I'd like to hear if someone thinks the opposite - that changing your stoploss doesn't change the number of your winning trades for your system. I thought that this fact is so straightforward that I was not even questioning it in my mind. But I'll be happy if someone disproves me.

Michal

[TheRumpledOne]

Michal:

I told you that in LOGIC, there is ALL, SOME and NONE.

Sometimes changing the stop loss effects the outcome but sometimes it doesn't, it is THAT SIMPLE!

It is a matter of LOGICAL FORM.

The statement:

"If you change the stoploss then the outcome changes" is FALSE because it can be shown that there exist at least one instance where this is false.

The statement:

"Sometimes when you change the stoploss then the outcomes changes" is TRUE because it can be shown that there exist as least one instance where this is true.

QED.

[jhtumblin]

I think this is just a clash of the literal use of the definition of changing your stoploss changes the outcome. You are both right depending on how you state it. Michal is examing the system's outcome over a lengthy period of time, I would guess at least a month or longer for daily trades, shorter if intraday. Avery is looking at each instance as an individual event rather than the macro picture.

So, that said, my take on it would be:

(Michal): If you change your stoploss for a system with a moderate number of trades over a moderate length of time or longer, then you are changing the total number of winning trades.

(Avery): If you change your stoploss for a system with a minute number of trades over a smaller length of time or shorter, then it is possible that the stoploss does not affect the total number of winning trades. It is also possible that changing the stoploss may never affect the number of winning trades if number of trades is < 2.

But now, both of you must remember that in trading, infinity technically does not exist because any trader is limited by his own capital (I know of no one with infinite money.) So when you are trying to prove your point more valid than the other by showing example trading systems with different stoplosses, the stoplosses must be realistic to a trader with say, $1 million equity ( I think that is a fair number, but I don't really believe this discussion is going to get this far.)

I think I just solved world peace.

[TheRumpledOne]

jhtumblin:

What you AND Michal are failing to realize that LOGIC is a matter of FORM or STRUCTURE.

X implies Y ( X -> Y) is a form. The implication is TRUE when, X and Y are TRUE or when X is FALSE. The implication is FALSE when X is TRUE and Y is FALSE.

To show X -> Y is FALSE, all you have to do is show ONE instance of Y being false.

I stand by my proof above.

[peter_77_peter]

Well let´s have a look at this from the probalities perspective.

Lets make assumptions:
(i) your trading signal is equally distributed correct/wrong, that is the probability distribution of price would fade off from the point of taking the signal equally distributed in both directions.

(ii) You would apply the same stop amount used for cutting your losses as a profit target to realise your profit.

(iii) You would get another signal after you´ve just finished your trade.


I could be proven, applying the law of large numbers that

A) applying any stop loss would result in higher income in any case if the number of your trades is going to infinity, therefore infinite capital is needed. It is simply following from assumptions (i) and (ii): In the long run all trades within the interval: [decision point minus stop loss, decision point plus stop loss] would add up zero, but the winners outside this range would add to your outcome in the long run.

B) Any Change of the stop loss would change the time needed to reach a given amount of x money, because the distribution of price is fading away from the decisive point, the expectancy of reaching outcome x would get any bigger the closer the stop, and any smaller moving the stop any further.

Therefore, Michal is right in theory. Practically TRO is right, there is no infinite money, but this is needed for calculating with the law of large numbers. Without infinite money, even if the probability distribution is as stated, still anything pathetic can happen with some probability.

Lets have a look at keeping assumption (II) and (III) and changing assumption (I) to:

(ia) The expectancy of a good trading signal is better than the expectancy of a bad trading signal. That is the probability distribution of price would still fade off from the point of taking the signal, but prefering the direction of the signal by some factor.

(A) and (B) can still be proven, but you´ve got the probability distribution on your side, that is still anything can happen, but because the good case has a higher expectancy, the probability of ending up winning is higher, therefore your risk profile is better. The better the signal the less money needed to make profits. Still, this is all deduced from the laws of distribution, requiring large numbers, still, Michal is right in theroy, and practically it is TRO, because, still anything pathetic can happen that is your money can be consumed before you make the profit wanted. But, as stated the better your signal, the more pathetic TROs Case is getting in terms of probability (still his point of view is stated completely right in terms of logic), and the more realistic Michals Case.

Lets keep (ia) and (iii) and change (ii), therefore, lets assume, you would be able to let the winners run further than the amount x of your stop loss. Likewise the change from (i) to (ia), the change in assumptions from (ii) to (iia) will result in an even more kindhearted distribution.

If you would keep (ia) and (iia) and change (iii) in a way that makes it possible to get signals while you´re trading, adding up, or cutting down your positions, would help you as well.

Same for any reasonable change the risk of your trading style for the better.

So I would agree to jhtumblin, everybody is right, from some point of view. And I guess we all agree that trading is still a betting game but stop losses are a good idea, and should be put according to your risk profile, your money account, the quality of your signals, and the distribution of the market you trade guessed in terms of volatility.

I hope I have contributed to a clarification of this thread and not confusing it any further. Any pirate having a cannonball thought on board, bringing my ship of thought sinking to the ocean bed?

[TheRumpledOne]

LOL...

You are cracking me up!

Michal's error was in the FORM of his statement:

IF YOU CHANGE THE STOP LOSS THEN THE OUTCOME CHANGES.

The tautology would be:

ALL CHANGES TO THE STOP LOSS CHANGE THE OUTCOME
THE STOP CHANGES
=====================================
THE OUTCOME CHANGES.

But since there is ONE INSTANCE where a change to the stop loss does NOT change the outcome, it's FALSE!

Had Michal said:

SOME CHANGES TO THE STOP LOST CHANGE THE OUTCOME
THE STOP LOST CHANGE
======================================
THE OUTCOME CHANGED

That would be TRUE because there exist at least ONE INSTANCE where that is TRUE.

The error in logic lies in the universal declaration he made.

For the record, I studied logic in college and made "A"s.

[jhtumblin]

TRO, I understand where you're coming from, and I realize the form of your statement. But just as Michal's statement is not always right, it is not always wrong either, that is why I say this is a clash of the literals.

There are examples where changing the stoploss affects ALL outcomes and doesn't prove Y false in any instance. It's just dependent on the timeframe which one examines. The reason for my statement regarding longer and shorter timeframes is because in "general" Michal's theory is correct only for a given length of time while your theory of focusing on only the incorrect instances, is only correct when those instances exist.

Now, the incorrect instances may occur frequently or be absent altogether depending on the timeframe you are looking at. If you were to take the market in its entirety, from beginning to end, then your statement is correct because you can find at least one instance that is false. However, as system builders we rarely look at all data across all timeframes when developing, therefore, Michal's statement can be just as valid in correlation with the data on screen.

[peter_77_peter]

well for the record, I´ve studied mathematics at the university, having courses in logic as well as in probability. Logically you´re completely right TRO, but, as jthumbling set up a connection to probability which was probably the way Michal was thinking to, your statement gets some probability, therefore, of course you´re right, Michals Statement can be proven to be no logical tautology, but, assuming that Michal simply lacked saying that his statement has been made within a framework containing the perspective of probability, as I`ve tried to explain, jthumblins statement saying that the both of you are right, can be clarified explaining the integration of your logical point of view into a framework based on probability....of course your point of view gets some extra points for trading relevance, because probability deals with assumptions and large numbers, that are kind of impractical for real life, but still, this framework can help clearing up the mind if you are dealing with risk more generally.

maybe I should change my forum name to mad scientist.....lol.....

[michal.kreslik]

OK, so let me be more precise.

I made a statement based on some implicit assumptions that I thought were commonsense among traders. I said "Sun is a star" and I implicitly assumed that we can read English, we agree on what means the word "sun" and on what constitutes a star.

But this doesn't seem to be the case . Correct, I made a mistake by not explaining what is the sun and what is a star. In this respect, as I did not specify the "sun" and "star", I was not right for someone who doesn't agree what "sun" and "star" means with me. But that was because I thought this information was implicit.

Ok, so I need to be more precise to make my point. By the way, a tautology is a statement that is always true regardless of what are the truth values of its substatements. So if I say "I'm Pamela Anderson and Pamela Anderson has blonde hair, so I have blonde hair", I have just said a tautology which is true regardless of whether I really am Pamela or not. (I'm not ) So this logical domain is not what we are dealing with here. I could easily construct an always true tautology by stating "When you change your stoploss, you change the number of winning trades, so if you changed your stoploss, you changed the number of winning trades", but that wouldn't make us any pips

Also, I maintain that preferably, we want to asses the performance of a trading system based on its performance over as many trades as possible, so in an ideal case, over an infinite number of trades. We don't want to assess the trading system based on just a couple of trades, because that can seriously distort the results. We should not be interested in what happens with any single individual trade. We should be interested in what happens over many, preferably "infinite" number of trades instead.

Anyway, the entire problem boils down to this true statement (now I hope I'm precise enough) :

• You can't eliminate the possibility of the trade going any arbitrary way between your entry and exit, so all other factors being equal, you can't eliminate the possibility of changing the number of winning trades for your trading system if you change the stoploss.

Also, if we consider the following assumption:

• "Over infinite number of trades, the trades will go all possible ways between entry and exit"

then this statement is true, too:

• "Over inifinite number of trades, you change the number of winning trades by chaning your stoploss."

Michal

[peter_77_peter]

so, first, STUPID ME, LOL, I have to cannonball my own pirateship because I have´nt been consistent in my argumenation as well. In the following of my basic assumptions, i´ve constantly used the later introduced (iia) instead of (ii), but with that formal correction it should be reasonable.

Therefore, I Agree with Michals Update, and I think that finally all pirates have cleared about formality in assumptions about the stars and the sun and everybody (more or less) agreeing with everybody else, at least about the point that we all are not Pamela Anderson but pirates, and therefore we consider the stars and the sun for navigation..........