Exits-- targets vs. trailing stops

Quote from acrary:


I look at the morning range and take trades after 2pm et in the direction of a breakout if the previous days range hadn't been extended. Often this leads to a strong directional close for the day. Since this happens often, you get lots of trades and most of them don't win, but overall it's a winning strategy.

Here's the results of a model using range extension risking 1/2% per-trade since 1983 starting with 150k in the account:
More...

There is something strange here. The system results that you posted average over 100 trades per year. However, I just ran a test to find all the days where the daily range is below previous day's high or above previous day's low until 2pm EST, and then a range expansion breakout takes place in the last two hours. From January 1996 through January 2003 there have only been 180 such days.

Am I missing something here?

-bbc

PS All the testing was done on tick resolution data from TickData.com
 
i use breakeven and targets. after 1R stop is moved to BE. then its either 3R gain or BE. Ive gotten 22% wins, 44% BEs, and 34% losses. Overall that gives profit factor of 2. but i got another problem with commissions killing the whole thing since im daytrading but thats another topic.
 
Traderkay...

I would sell 1/2 to a 1/3 of my position once 1R is reached and then place the stop at breakeven.

You need to reguarlarly pay yourself $ even if it is little... because those little gains add up.. and you will maintain more consistency.

You are pretty much trying to swing for the fences... on every trade... either.. you strike out, walk, or hit a homer....

Personally.. I'd rather hit singles, doubles, triples too...


--MIKE
 
Quote from blueberrycake:



There is something strange here. The system results that you posted average over 100 trades per year. However, I just ran a test to find all the days where the daily range is below previous day's high or above previous day's low until 2pm EST, and then a range expansion breakout takes place in the last two hours. From January 1996 through January 2003 there have only been 180 such days.

Am I missing something here?

-bbc

PS All the testing was done on tick resolution data from TickData.com
More...

Yes, you're missing something. Range extension means the market moves a certain amount above the high or a certain amount below the low. When it doesn't happen in the morning and the trading range is less than expected, then it's a possble setup for a afternoon trade.

Today is a possible range extension day. The high this morning was above Friday's high, but not high enough to invalidate the trade. The range as of now has been 10 pts. which is about 40% less than the average daily range. If the range remains contained, then this model will buy/sell the breakout above or below today's high/low after 2 pm.
 
Quote from blueberrycake:


However, I just ran a test to find all the days where the daily range is below previous day's high or above previous day's low until 2pm EST, and then a range expansion breakout takes place in the last two hours. From January 1996 through January 2003 there have only been 180 such days.
More...

I think what you're describing is a continuation day. I started work on a system for this as a followup to extension days. I was working on the longside trades and stopped work...lost interest.

Here's the code I was working on at the time if you're interested:
 

Attachments

And here's the annual results since 1990 for the previously posted code. I think there's an edge here, but since I didn't finish, it'll wait for someone else to exploit.
 

Attachments

If you have a viable way of projecting a price target with accuracy, then go for it. Otherwise, a fixed amount target is prone to curve fitting to past data.
 
Quote from acrary:


Here's the code I was working on at the time if you're interested:
More...

For those of us who are TradeStation challenged, can anyone explain what the 'doji(X)' function does?

-bbc
 
It's a true or false statement. If doji(20) means the open and close must be between 20% of the high/low range for it to be true.

Ex.
O = 1000
H = 1010
L = 1000
C = 1008

h - l = 1010 - 1000 = 10 pt. range
abs(o -c) = abs(1000 - 1008) = 8

1 - (8/10) = .2 or 20% so the doji(20) would equal true
 
You must log in or register to reply here.
Back
Top