Traders' Roundtable

Hello

Let us say that you have found a great system that gives you wonderful results. The only problem is that the system is based on something that has nothing to do with the market, for example the planet movements, or how many times it rains on a certain spot etc.

Would you trade that system? It might have worked historically but nothing tells you it will work in the future...

mathi wrote:It might have worked historically but nothing tells you it will work in the future...

Nothing (that I have found anyway) will tell you that anything will work in the future...

I wouldn't trade it if was based on planet movements or raining... technical trader would try to find situations when people are being caught in the market and run to liquidate by doing this creating clear picture what to do

BARLI wrote: technical trader would try to find situations when people are being caught in the market and run to liquidate by doing this creating clear picture what to do

Sounds like Barli was short cocoa recently...

I wanted to go short cocoa on 7/12/06 after it broke down a little, but thought it would begin "hanging" in the sideways range now again (as cocoa often does), that kinda lost opportunity sometimes takes me out of psychological balance, I am working on it. but you're right, lotsa Longs were "caught" on 7/17/06 and looks like lotsa protective sell stops were hit. I hate that fills come back only 30 min after in Cocoa and Sugar (I use Man Financial as a clearing firm). Actually folks who went short on 7/12-7/14/06 had guts to go against the crowd which payed out handsomely for them, this kinda situation I meant when I talked about "going against the crowd".

Hi mathi :

Trading based on planetary movements sound so risky!! BTW, how do you determine which planetary movements are bullish, and which are bearish?

Even when you are very confident of your next trade based on fundamentals and/or technical analysis, there could still be a possibility that your trade would "go the other way" due to un-predictable market reaction.

Solution : Practise strict money management and stop-loss for every trade no matter how confident you are.

i dont know about you, but when i hear of people researching stuff like planetary movements, and magic retracements and astrological signals etc etc, i get the vibe that they havent quite stopped searching for the holy grail. That one big discovery that only they will know about and that no one else will ever find out about, thats gonna push them into the pot of gold at the end of the rainbow. Even if you did find this holy grail in astrology or planetary movements or moon states, you`ve still gotta apply a stop loss methology and position sizing algorthym, which basically leads you right back to basic meat and bones trading.

Just my opinion, but not only does the system or idea have to have worked historically based on some kind of statistical examination, but the system or idea has to have some kind of basis in causality. This is one of the reasons I discount 4-year cycle, 20-year cycle, Kondratieff waves, etc., as indicative of anything but gullibility.

There really are only two kinds of trades, mean reversion and trend following. The "why" of the system should address why price will either continue to trend or revert to the mean. If it doesn't, perhaps it is merely a statistical oddity.

by the way Gann was trading using astrology, so does Michael Jenkins..they seem to be good market forcasters, so maybe theres something about it..

Even if correlation is proven, it does not prove causation.

On the other hand, one doesn't need to prove causation to make money.

On the third hand, it is valuable to have a cause as assurance that the coincidence will not simply cease to make money.

TrendsCatcher wrote:On the other hand, one doesn't need to prove causation to make money.

Very true indeed.

true.. I like the quote by Livermore you put. No one ever knows (except the company's CEO and an accountant) what a company is really worth..

TrendsCatcher wrote:On the other hand, one doesn't need to prove causation to make money.

Quite a few people think that the recent abysmal performance of Dunn and JWH is because they trusted their trading systems too much. Each of these money management firms has had huge success for a lot of years, each uses a set of very simple mechanical trading systems, and each boasts that they have never changed these systems in the past and never will change these systems in the future. Which might be their downfall.

I grabbed this image off another bboard, the investment performance ("equity curve") is the solid black line, the S&P 500 is the blue line, and the customer dollars under management is the olive-green histogram. Dunn and JWH describe themselves as trend followers. So do the other three managers shown: EMC (Elizabeth M. Cheval, an Original Turtle), Transtrend, and Winton. Clearly some trendfollowers have done better than others in the last few years. Who knows, perhaps the ones who mistrusted their systems, and therefore made changes, have enjoyed greater success recently.

"The ballistic descent of trendfollowing's reputation into the sewer" viewtopic.php?p=23566&highlight=dunn++jwh#23566

Any ideas as to what the latter three are doing differently?

levijean wrote:Any ideas as to what the latter three are doing differently?

i share your curiosity...
my guess, based on dreams of having millions of dollars to trade:

the more obvious
- 10s of systems per market, diversifed by timeframe and method
- 100s of markets

the more esoteric
- risk models that manage +ve and -ve volatility spikes
- models that estimate and manage the probability of a trend ending
- dynamic sector allocation models
- seasonality models exploitling the cyclicality of trend following returns

can anyone think of anything else?

yeah sluggo, could give us some insight if you have one? Winton seems to have a very smooth performance

levijean wrote:Any ideas as to what the latter three are doing differently?

Maybe another question would be: Do any of the posters have an equity curve that resembles the latter three?

Could it be as simple as BET SIZE?

Smaller size will smooth out a curve ...