Our 400,000 Trade Test Proves Using Stops Is Not Very Bright

[J_Smith]

man, your town was a zoo last night for st patty day. Insanity!

You felt right at home then

J_S

 

[marketsurfer]

This isn't proof of anything. It is not shocking that a system designed with terrible entries will produce terrible results.

If you don't know where to enter, how in the hell can you know where to exit?

100's of tests with multiple variations of entry rules were applied. It really doesn't matter to me if its "believed" or not. Just stating the results of the testing-

Could there be trading savants toiling away in some dungeon who obtain different results via intuition ?

Sure, but i prefer to stick to the tested facts.

 

[marketsurfer]

https://www.finsia.com/docs/default.../7_10-vanstone-jassa-n4-2008_web.pdf?sfvrsn=6

 

[destriero]

https://www.finsia.com/docs/default.../7_10-vanstone-jassa-n4-2008_web.pdf?sfvrsn=6

lol that's the paper?

 

[marketsurfer]

lol that's the paper?

Nope, just some more evidence that stops degrade performance of trading systems.

 

[I Know You]

Very very interesting, thank you for the recommendation. I will never use a stop again.

This is the results of 400,000 tested trades--- like it or not, it is the facts-- don't believe it? If you got the data, run the test yourself---so you either listen to quantitative facts or continue to go by "feelings" and old wives tales about the market.


Stops are an integral part of most stock trader’s arsenal. Many consider it foolhardy and naïve to trade without a fixed stop in place prior to entering a position.

As with most of the conventional market wisdom, these feelings do not hold up to vigorous testing. We have run 100’s of tests have that have proven fixed stops actually hurt the performance of any trading system.

One example took all shares trading above their 200 day Simple Moving Average closing at its 10 day low. The trade would be exited on a close above its 10 day Simple Moving Average or when a stop is hit at a variety of percentage points below the entry.

Out of nearly 400,000 tested trades, stops hurt the performance of the system. In fact, the tighter the stop, the worse the performance. Even stops as wide as 50% away from the entry dampened the systems returns. Of course there are psychological benefits to using fixed stops but at the expense of profit. Dropping the fixed stops is a sure way to improve the results of your trading system.

 

[botpro]

Here's a similar study about Stops:

"Stop-Loss Orders: Help or Hindrance?"
http://www.incrediblecharts.com/trading/stoploss-trading-1.php

One of the authors of the research paper from the year 2008 surf had posted here as a reference to back his own findings is the very same author of the above study.

That same author seems also to have conducted an other study about the usage of TrailingStops, but that study I have not searched/found yet.

 

[botpro]

What everyone seems to have missed ....

How does one get the results of over 400,000 individual trades, by only running 100's of tests?

A test can be long, looong, loooooong, veeery loooooooooog....
Some maths helps, really, try it!...

 

[destriero]

Nope, just some more evidence that stops degrade performance of trading systems.

From some dude from some uni that nobody has hear of, using ATR?

Surf, think of how much sim-monies you would have saved with your recent (last 10 trades) YM trades had you covered instead of adding.

 

[botpro]

Botpro is still trying to figure out how he must adjust his simulated data for the time change

Nope, you just demonstrate that you have absolutely no idea about system testing with simulated data.
I don't need to do curve-fitting. Curve-fitting is usually done by those who use past data,
but I prefer forwardtesting with realtime sources, and also simulated data because one can test quickly and cheaply 1000s of market situations.
Read my this recent reply in an other thread:
http://www.elitetrader.com/et/index.php?threads/100-wins-over-240-days.298459/page-4#post-4258495

Really, I just wonder about the IQ of those people who still question the usefulnes of simulated market data for system testing... Amoebe-IQ I would say...