Page 1 of 1

Journey to turn the worst system into the best one!

Posted: Fri Apr 30, 2004 11:35 am
by K1
I picked this thread up on a different forum and would like to know if somebody has done any work on this topic, this was posted by a forum member:

" Basically, if x = -100 pips and then if we do like this x = x * (-1) then x = 100 pips. Believe me or not, it is just like turning -100 pips loss into 100 pips profit, if we reverse the signals! You may ask what does this mean? Ok, it means that if that worst system give a buy signal then reverse it into a sell signal and when the sell signal is given then reverse that into buy signal.

In short, buy if "sell signal" is given, and sell if "buy signal" is given.

I'm not sure which system is the worst one, so can someone tell me what's system the worst one you've met? I need a system that lose more than 100 pips a day. The bigger the loss (exclude spreads), the better it is.
"

What would happen if you have a system where you are 30% correct with your signals and flip these?

Posted: Fri Apr 30, 2004 12:14 pm
by Bondtrader
One easy thing you can do is take the BEST system you've ever seen, and just reverse its signals (longs become shorts and vice versa). Presto, you now have a completely terrible system.

In the trading world there's a name for this: "fade" the signals. XYZ is usually wrong, so if you do the complete opposite of XYZ, you'll make money. Whether XYZ is a person or a system. I seem to recall that one of the Market Wizards had a personalized license plate "FADE ME" but I could be wrong....

Posted: Fri Apr 30, 2004 3:12 pm
by Jake Carriker
K1,

There is a discussion of this type of fading strategy in The Ultimate Trading Guide by Pruitt, Hill, and Hill. They use Bollinger Bands as an example of a "good idea applied incorrectly" and use Bollinger Band breakouts (the modus operandi of some good trend following systems) as signals, as opposed to the conventional logic that says BBands should contain price.

There are a few qaulities that separate a good candidate for logic reversal from a simply crappy system. That book suggests that a good candidate will be: based on sound market principles, have a low trading frequency, and lose consistently over different parameter settings and markets.

They suggest most bad systems are simply that, and they will not overcome transaction costs and slippage when the logic is reversed.

A good example of faulty signal reversal logic follows (my example, not the author's):

A phase of the moon system that buys market X each time the moon is full during the second week of the month performs terribly. Therefore, I reverse the system logic and fade all signals. Am I likely to do better this way? Not really. There is no rhyme or reason why this system should work or fail. It is practically random, so reversing the signals won't help anything.

Now, if you could find a way to reverse slippage and commissions, that would be something! (actually you can do this. That is what brokerage firms and floor traders do. The overhead, capital requirements, and other barriers to entry are high though) :)

Jake

Posted: Fri Apr 30, 2004 6:13 pm
by Kiwi
Jake,

Do you really believe that brokers and day traders spend their time figuring out system signals and fading them? I doubt that any system other than aberration and the X day breakout (just looking for a double top really) attracts this type of attention and even in these cases the entry is day trading based on reading the tape to see if the order should be faded.

To me this is not fading in the automated way a system trader would have to fade them.

Happy to be educated otherwise though.

John

Posted: Fri Apr 30, 2004 7:26 pm
by Jake Carriker
John, not at all.

My comment was meant to imply that if you want to be on the profitable side of commission or slippage, you had better be a floor trader or broker.

I don't think there is much profit potential in fading popular systems, ala "turtle soup".

No worries,

Jake

P.S. - I have been both a broker and a day trader, and I did not spend my time trying to fade system traders in either capacity.

Posted: Fri Apr 30, 2004 8:29 pm
by Roscoe
Hi K1,

I do a lot of testing and one thing that tends to stand out after a while is that there are far more system ideas that don't work compared to the ones that do.

Sometimes when I have a real pig in testing I will try reversing the signals on the basis that if is really that bad then it must work better the other way 'round, but the result is usually that it is still a pig, just a different kind of a pig. As Jake says, most bad systems are just that, whichever way you play them.

No free lunch with this one IMHO.

Posted: Sat May 01, 2004 12:09 pm
by batuco9
I can see Roscoe's point.
I would say that the mapping between 'good' signals and 'bad' signals is not isomorphic (1 to 1). (although the mapping of 'all' signals to 'all' signals can be considered isomorphic or 1to1 on applying the '-1' factor).

If the mapping of 'bad' signals to 'good' signals were '1 to 1', then x=x(-1) would indeed convert a 'bad' signal into a 'good' one as indicated.

But there are vastly greater number of 'bad' signals than 'good' signals so most 'bad' signals will just map unto other 'bad' signals. Otherwise, we would have to have as many 'good' signals as 'bad' signals, which I don't consider anyone believes to be the case.
I would think that the probability of chancing on a 'very good' signal by reversing a 'very bad' signal would therefore be very low.

Posted: Wed Jun 30, 2004 11:21 pm
by batuco9
Looking at the threads I see this one is quiet Shishhhhhh - sorry if I finished killing it.

I just would like to point out the obvious - though it probably will not make too much sense to try to invert terribly bad systems, time may be well spent toying with their inverted components ......., which is probably where K1 was driving at anyway.

Just surprised nobody took me up on this.

Posted: Thu Jul 01, 2004 1:38 am
by Forum Mgmnt
I've often found good ideas in bad trading systems and bad results. In fact, some of my favorite ideas started out by tests that faired far worse than I expected because this taught me something about the markets.

Though I've never found a case where reversing a system worked inverse to the original system. I've often found good signals when looking at the opposite of what ended up testing poorly.

- Forum Mgmnt

Posted: Thu Jul 01, 2004 11:13 am
by p8
Let's look at it using signal vs noise concept. If the system gives the opposite direction of a signal that actually goes the other way, then, yes, perhaps you can just reverse what you do and improve the result. But if the system gives a signal when it is actually just a noise, then no matter what you do it's still just a noise. The false signal only let your trade get caught in a random disturbance. Thus, reversing the direction won't improve it in any way.

Posted: Fri Jul 02, 2004 12:31 am
by leonardo
The worst system you could trade:

The doubling-up system, the martingale. Doubling up after you have a $500 dollar loss expecting the market to just have to retrace just a little bit to make up the loss and your specified profit.

Starting the buying process only when there is a downsloping 50 day moving average on your favorite market, (or selling only when the moving average points up) trying to get just $500 dollars profit from every sequence.

Trading this system could lose you an infinite amount of money, just trying to make your $500.

Or if you want to risk $500 to possibly get an infinite amount of money, trade the opposite.

Leonardo---

PS: If you ever have extra time on your hands you might want to program this concept. It is extremely eye-opening how much drawdown can be encountered before the $500 is captured. And if you even have an uncle point of $10,000,000 how many times that can be hit in a 20 year period in various ordinary markets.

Posted: Wed Jul 07, 2004 1:41 pm
by efficiency
OK gentlemen (and ladies): Other than not being early, WHAT IS WRONG with a simple system of buying a 120 day HHV, filtered by a rising ADX above 30, setting an intital stop at 2 x the 20 day average true range, then once in the money, adjusting the stop to a 21 day LLV and riding the trend until stopped out?????

Posted: Wed Jul 07, 2004 2:36 pm
by Ted Annemann
There's absolutely nothing wrong with trading that way. And there's absolutely nothing wrong with getting your trading signals from a Coke bottle that allows you to communicate with super-intelligent beings on Mars. If you haven't heard the (true) story, check it out:
http://www.traderclub.com/discus/messag ... 20000924pm

The big question is, how well does your proposed trading method fit YOU? Will you be able to faithfully carry out its signals? Or will you find it irresistable to "jump" the system? Does it trade too often for you? Not often enough? Does it require you to constantly place new stop orders at new stop prices, every single day, for every single market in your portfolio? If so will that be a problem for you?

Are you interested to know how your system performed in past history, using a computer backtest? Are you interested to know how a bunch of other systems performed in past history, so you can compare your system's backtest results against them? Are you interested enough to spend the time and effort necessary to obtain the software and historical price data, then put these systems into the software and run the tests? Are you interested enough to spend the money that this kind of testing will cost?

It all gets down to you and who you are (as a trader) and what you want. For that reason, we the other readers of your post, can't offer too much specific advice. We're not you.

However, I will offer this one opinion: your system is very well described. There's nothing "wrong" with its definition. I used Google to translate your words into French, then translated the French output back into English. The final result is not nearly as clear as your original:
Messrs WELL (and ladies): Other that not to be early, WHICH IS ERRONEOUS with a system simple to buy 120 one day HHV, filtered by a ADX rising above 30, placing a stop intital to 2 X the average range true 20 days, then silver once, adjusting the stop over 21 one day LLV and assembling the tendency until decree out of the?????