NPP builds a Emini system

N

no_pm_please

Fully disclosed of course. lol.

So, I was flipping through the Dec./Jan. issue of Trader and came across this quote "Trade with a trend, ride the winners, cut the losers. How you do it doesn't matter" from a past market guru.

What the hell I thought, I can do that in my sleep. Let's see if it can be made into a trend following system for the emini market.
Should also work on NQ and ER2 (at least in theory).

To start out, here's the idea.

1). Use RSI as the indicator for the trend. A 50.5 reading and we should be long. A 49.5 reading and we should go short. So the risk is the maximum of the length of time between bars and a range of 1.0 on the RSI.

2). If the market rises above 52.5 and then crosses below it, then exit any long trade. If the market drops below 47.0 and then crosses above it, then exit short trades. As long as the market runs it'll just stay in the trade. As time goes on the RSI will move in the direction of the move so it's just like having a trailing stop.
I made the short side a little lower due to short trades moving faster than long trades.

Now, the only 2 pieces left to figure out are:

- how many periods to use with the RSI ?
- what timeframe to select to base the periods on ?

I plan on testing 15 min., 30 min., 60min., 120 min., 180 min., and 240 min. bars for each year from 2003 - 2006. One year at a time and optimize the number of periods to figure out what was optimal for that year. This should take a couple of days. Then lets see if there's anything worth trading.
 
If you want to check your results against mine, here's a example:

Here's a test run on 15 min. ES data for 2003. I used 100 bars for the max lookback, 43 bars was the optimal per. I assume no slippage or comm. costs. You can see the basic idea follows the trendfollowing idea.
 

Attachments

Hi, NPP,

It's good to hear from you, and to have you back again.

I would like to say "Thank you", for your past contributions here on ET. They have helped me become a better trader.

Now back to your regularly scheduled programs :D

Schaefer
 
my results are a bit worse than yours.

Quote from no_pm_please:

If you want to check your results against mine, here's a example:

Here's a test run on 15 min. ES data for 2003. I used 100 bars for the max lookback, 43 bars was the optimal per. I assume no slippage or comm. costs. You can see the basic idea follows the trendfollowing idea.
More...
 

Attachments

So it looks like the optimal time in the trade would be 30-48 minutes? Help me out a little with the interpretation...


Quote from bidask:

here are my results for 2003 to 2006.

the numbers mean the optimal period.
More...
 
You must log in or register to reply here.
Back
Top