turn around of unprofitable strategy

[TZoell]

Hi

My question is, when I have a strategy which has a very poor performance, so it makes a great amount of looses, and I turn everything around (for example old strategy would enter a long order at a certain point ... new would enter a short order and so on). Then the new strategy should have a very good performance or not?
So for example the old strategy does -10000 Dollar over a certain time, the new should make profit of 10000 Dollar (of couse without commisions).
Is this correct or do I have a error in reasoning?

[NinjaTrader_Brett]

Hello,

Thanks for the note.

I will leave this open for other traders to comment on. However this falls more under trading education which unfortunately I'm not qualified to assist with.

However our education partners are:

http://www.ninjatrader.com/partners#Education

[koganam]

Hi

My question is, when I have a strategy which has a very poor performance, so it makes a great amount of looses, and I turn everything around (for example old strategy would enter a long order at a certain point ... new would enter a short order and so on). Then the new strategy should have a very good performance or not?
So for example the old strategy does -10000 Dollar over a certain time, the new should make profit of 10000 Dollar (of couse without commisions).
Is this correct or do I have a error in reasoning?

Not necessarily. It depends on the parameters of the trades. As an example try using the SampleMACrossover, strategy that ships with NT. Given a long enough period, it must lose, as it can be mathematically demonstrated that it is a poor strategy. Once you get to that point, reverse the orders, and test again, over the exact same time length. You will still lose.

As an aside, a moving average crossover strategy CAN be made profitable, with suitable targets and stops, but in a stop-and-reverse mode, we can mathematically prove, assuming the markets trend only 15% of the time, as seems the case (the last 3 years of POMO-induced, Bernanke printing press fed, rocketship one-way up market can reasonably be excluded as a free market), that the strategy will more probably lose than win.

Remember the question that you asked. It was about a losing strategy. My answer is to that end, so the fact that MACrossovers are immensely profitable in a trending market is irrelevant: I am talking about a non-trending market, where MACrossovers will generally perform poorly, as per the conditions that you asked about.

IOW, I am just giving you an example to demonstrate that which you asked, so other responders can please leave aside the philosophical discussion as to whether trend trading is better than swing trading, etc, etc, etc.

[drolles]

As an example try using the SampleMACrossover, strategy that ships with NT. Given a long enough period, it must lose, as it can be mathematically demonstrated that it is a poor strategy. Once you get to that point, reverse the orders, and test again, over the exact same time length. You will still lose.

As an aside, a moving average crossover strategy CAN be made profitable, with suitable targets and stops, but in a stop-and-reverse mode, we can mathematically prove, assuming the markets trend only 15% of the time ...

Hi Koganam,

Thanks for the thought provoking post.

That’s a big and very definitive statement. I’m very interested in what evidence you have for that? Can you quote any papers or statistical robust work?

Kind regards,

drolles

[koganam]

Hi Koganam,

Thanks for the thought provoking post.

That’s a big and very definitive statement. I’m very interested in what evidence you have for that? Can you quote any papers or statistical robust work?

Kind regards,

drolles

There really is no substitute for you sitting down yourself and doing the experiment that I outlined. Yes, I have done it multiple times, over different time periods. In fact, over most time periods I could not even get it to show a profit either way when I used an exhaustive optimization!

However, as you seem to want to read what others have done, CXO advisory is a market research group that does a lot of statistical testing of trading methods. Here is the page I got from a search on their site: http://www.cxoadvisory.com/?s=moving+average+crossover

Here is another person's view: http://marketsci.wordpress.com/2008/...vers-debunked/ As to whether he is an authority, sorry, but I really do not put much mileage in so-called authority. To give just but two examples of authoritative madness, do you remember what the Church did to Galileo when he said that it was the earth that moved, not the sun? Or how Aristotle's authority had the world believing for ages that there were only 4 elements: Fire, Earth, Air and Water?

Question everything, including whatever I say, especially because I am no authority, but question authority too, not because we are rebels, but because we can think for ourselves, and should never let anyone tell us what we are allowed to think. OK, I am off the soapbox now.

[TZoell]

thx for the long answer

But it is not exact what i meant. I will give you a example.
Lets say we have a strategy which enters a long or a short trade at 12 AM. When the underlying is in the plus it enters a long order, when it is in the minus it enters a short order. Both orders have a stop of 10 ticks and a profit order of also 10 ticks.
When this strategy generates a great amount of looses over lets say 3 month and then I turn it around and backtest it over the same time interval ... it should then generate the same amount of winnings ... is this correct? Or do I miss anything?

[koganam]

thx for the long answer

But it is not exact what i meant. I will give you a example.
Lets say we have a strategy which enters a long or a short trade at 12 AM. When the underlying is in the plus it enters a long order, when it is in the minus it enters a short order. Both orders have a stop of 10 ticks and a profit order of also 10 ticks.
When this strategy generates a great amount of looses over lets say 3 month and then I turn it around and backtest it over the same time interval ... it should then generate the same amount of winnings ... is this correct? Or do I miss anything?

As I said in the earlier post, it depends on the parameters of the trade. Is this a random entry? What conditions are required for entry? IOW, there are defintiely situations in which what you propose might be workable. You will just have to test it for yourself.

What I am saying is that just reversing entries will not necesarily turn a losing strategy into a winning one, and then I proceeded to give you an example that you can experiement with to demonstrate that point.

[TZoell]

maybe I do not understand you correctly but lets look at the example I posted.
So long entry at 12 AM when underlying is in the plus, short entry when it is in the minus. Stop 10 ticks and profit order also 10 ticks. Lets say this strategy produces
-10.000 Dollar in 1 month.
So when we change the parameters to short entry when underlying is in the plus at 12 AM and long entry when it is in the minus. Stop and profit order again 10 ticks from the entry. So this would turn the strategy 180° around. Now the result for the last month should be +10.000 Dollar.
Is this correct?

[koganam]

maybe I do not understand you correctly but lets look at the example I posted.
So long entry at 12 AM when underlying is in the plus, short entry when it is in the minus. Stop 10 ticks and profit order also 10 ticks. Lets say this strategy produces
-10.000 Dollar in 1 month.
So when we change the parameters to short entry when underlying is in the plus at 12 AM and long entry when it is in the minus. Stop and profit order again 10 ticks from the entry. So this would turn the strategy 180° around. Now the result for the last month should be +10.000 Dollar.
Is this correct?

Please look at your original question and my response. I said: "... not necessarily". That implies that there are situations where it can be true. I also know that often, it is not, and I gave you an example. Maybe you have found one of those situations where your assertion is true? But how do you know that your assertion is true for this particular scenario?

First of all, I do not understand "in the plus" and "in the minus", so I cannot directly address that matter. IOW, you are right; I do not understand.

Secondly, if "in the plus" and "in the minus" are delineated, definable, terms, then you can code what you just described.

Thirdly. So code it, and find out for yourself.

That answers the particular setup you are talking about that I do not understand because I am unfamiliar with "in the plus" and "in the minus".

My only other suggestion is that rather than get into a philosophical argument with me, as I have given you the exact data and scenario to prove (or disprove) to yourself that a reversal of orders will not necessarily result in a reversal of trade outcomes, please go ahead and do the experiment. Once you have empirical evidence, then go about finding those exceptions that you seek.

I have no interest in a philosphical argument with anyone. I had many of those when I was trying to become a profitable trader instead of a losing one. Remembering the pain of those losing times, when someone asks a question, I try to explain what I know from the benefit of my experience, painful as it was. If instead of, at the very least, doing the experiment to see things for yourself, you want to get into defending a position about which you asked for opinions, instead of examining the position in the light of information elicited by your question, that is your prerogative.

Do you want to be profitable, or do you want to be right? If you believe so fervently that you are right, and so do not need to test your theory, then I guess you can test it live in the market instead. Reverse your entries, with your real money on the line, and if you become profitable, then this whole discussion is moot, because you will be both right and profitable. OTOH,...

As for me, I NEVER trade a method based on a theory UNLESS I HAVE PERSONALLY tested the theory/method, whether in Backtest, Simulation or Market Replay; usually in all three of them. Then again, that is just me. I guess losing enough money taught me something after all.

[TZoell]

first I want to say, that I am from germany and I find it sometimes a bit difficult to translate everything correctly ... so sometimes I translate things word for word ... that results in things like 'in the plus'
with in the plus I meant, that for exampel the DAX opens at 5000 points and after 2 hours the price is higher then the opening price. We say for this in german 'Der Dax ist im Plus'. The best translation I found was 'in the plus'. I hope you understand now what I meant ... maybe you can tell me the correct word in english ^^

I do not have a interest in a philosphical argument either. I payed someone to code me a strategy related to the strategy I descriped here and it was very unprofitabel. So before paying again to turn things 180° around I wanted to ask if I had maybe a error in reasoing. If I did not have to pay for it, I would test it on my own of course.

[koganam]

first I want to say, that I am from germany and I find it sometimes a bit difficult to translate everything correctly ... so sometimes I translate things word for word ... that results in things like 'in the plus'
with in the plus I meant, that for exampel the DAX opens at 5000 points and after 2 hours the price is higher then the opening price. We say for this in german 'Der Dax ist im Plus'. The best translation I found was 'in the plus'. I hope you understand now what I meant ... maybe you can tell me the correct word in english ^^

OK. Yes, translation can sometimes produce strange expressions, but now that I seem to understand what you mean, I would say that you can yourself create that strategy using the Wizard in NT. Then you can test it for yourself.

Essentially you need two variables:

Mar****penTime
TimeSinceOpen

Then you just want your conditions (pseudocode) to be:

 
if Price at TradeTime (which is Mar****penTime + TimeSinceOpen) is above Price at Mar****penTime, buy with StopLoss of 10 ticks and ProfitTarget of 10 ticks;
 
if Price at TradeTime is below Price at Mar****penTime, sell short with StopLoss of 10 ticks and ProfitTarget of 10 ticks;

Then just run it through the strategy analyzer and see for yourself.

Then just reverse your buys and sells and see what happens.

... I payed someone to code me a strategy related to the strategy I descriped here and it was very unprofitabel. So before paying again to turn things 180° around I wanted to ask if I had maybe a error in reasoing. If I did not have to pay for it, I would test it on my own of course.

Unfortunately the problem with all trade method development is that you cannot know the results until one tests the method, so paying for coding services can get mighty expensive. There really is no substitute for learning to do at least the simplest things for oneself, and NT gives us a remarkable platform that makes that not too difficult.

As far as whether there is an error in the reasoning, that one means we probably have to wax philosophical. Just think about it. There have been millions of traders before you and I, and the brokers' statistics tell us that it seems over 95% fail. That tells us that most traders must have been, or are trading systems that lose. Surely, among all those who came before us, many traders must have thought: "As I am so good at making losing trades, I should just reverse my entries, and I will become profitable". But the attrition rate is still 95%. I would try to draw at least a preliminary conclusion from that. (Incidentally, after my many initial losses when I started trading, I did try that one for a while: my reversed trades made me lose money too).

[TZoell]

thanks for help ... I will check weather it works or not

As far as whether there is an error in the reasoning, that one means we probably have to wax philosophical. Just think about it. There have been millions of traders before you and I, and the brokers' statistics tell us the it seems over 95% fail. That tells us that most traders must have been, or are trading systems that lose. Surely, among all those who came before us, many traders must have thought: "As I am so good at making losing trades, I should just reverse my entries, and I will become profitable". But the attrition rate is still 95%. I would try to draw at least a preliminary conclusion from that. (Incidentally, after my many initial losses when I started trading, I did try that one for a while: my reversed trades made me lose money too).

I thought this would be the same like playing poker most people lose because of rake/commisions. Until last week I thought, that creating a winning strategy would be as hard as greating a strategy that produces great amount of looses and that most strategies are about break even and commisions are the only reason, that most people lose