Simplicity

Quote from learn&earn:

how were you cherry picking your trades? I have been waiting till about 10am (height of the bubble) and then getting in either short or long. Seems possible.
More...

Well I tested different strategies (the one you mention and other similar) on hundreds of stocks, future, indices and currencies over about 25 years.

I encourage you to test it for yourself. As I said, you'll surely learn something.
That's how I got to being profitable, by testing ideas.

Ninna
 
ive tested a bunch too, and the biggest eye opener is how many things don't work...almost nothing works, not even in backtesting.

I'm excluding, of course, basic strategies that work in backtesting but would never work in reality because of slippage and/or transaction costs....you can find quite a few of these that seem to work, but of course never would in real life.
 
Quote from nLepwa:

Let me give you an example.
We play a game several times. If you give me a correct answer, I give you $85. if you give me the wrong answer you give me $15.
You get the answer right about 1 time out of 100.
What's the expectancy of your wealth?

Ninna
More...

I agree that with this example the expectancy is not good. The problem is I'm not right on 1% of my trades. Lets say worst case its 20%. So being right 1 of 5 times and having a chance to catch a HUGE run is worth getting stopped out more often. We will see what happens. If i stop posting its because im on a private island and my idea WORKED lol.:D
 
If you're not going to back test it, then at least just simulate the trades in real time over say, a month, and see if it works, before you risk any money.

I personally think the thinking behind it is sound in that you should end up with a normal distribution of winning trades, but a fat tail distribution of profits/losses. Assuming you're cutting the losses short and running the profits, you would think it would work.

But if the tests say no, then too bad.
 
Consistent profits are made by having more winning trades than losing trades and/or having bigger winning trades than losing trades. Since you haven't defined any clear entry that gives you an advantage, off the bat you have to expect at least as many losers as winners. Since you are looking to have larger winning trades than losing trades, then your stop has to be much closer to your entry than your target. Assuming a normal probability distribution (it's not, but pretty close), then it follows that you will clearly have more losing trades than winners. So the question is will the fat tails make up for the low win rate? My bet would be not enough to overcome trading costs.

But sound trading strategies are based on very simple concepts. Take what you've got and identify an entry strategy that raises the odds of follow-thru in your direction and this could very well be profitable.

Like others said, it is best to test it. Much quicker and less costly than trial and error with real money.
 
Quote from learn&earn:

These stocks usually have HUGE volume and move 10% up or down after the open so you have a decent chance of catching a big move. Yes, you are going to get whipsawed often and maybe even stopped out 3 of 5 times but your 2 profitable trades will be huge winners. Is this a valid idea? Just a thought.
More...
Most of the time, 10% moves aren't linear. If you employ a reasonable stop, you're apt to be taken out long before 10% is achieved. The big moves are usually achieved in hindsight :)

For example, canned indicators can be backtested to find a periodicity that provides spectacular results (optimization). Use a different number and you would have gotten crap. IOW, what appears to work extremely well in one segment of data is no more than curve fitting. IRL, the ideal number changes and there's no way to know that, going forward.

And when you find something that works, generally, it doesn't work forever. You have to bang em out when it's working and adapt when it's not.
 
just an update this WORKS, probably not forever but right now it is really great. KV.A alone made multiple HUGE moves. I am going to start calling things real time. P.s I'm not really a big computer guy so how in the world would I go about backtesting a purely discretionary system?
 
You must log in or register to reply here.
Back
Top