Trading

[pig4bill]

For EOD, Metastock is pretty good. Not programmable to the degree that Tradestation is, but not bad. And if you really want reliable results, you'll have to buy data. The free data I've seen doesn't compensate for splits or symbol changes.

[Mr. Now]

I look for entries where I can define the risk with a tight stop. Usually the setup features an in-place trend, narrowing daily range and well-defined price zones (pivotal points).

I trade primarily long stock no margin in qualified plans. My typical holding period is up to 90-120 days and losing trades are usually about 3 to 5 days in duration. I use a 'time stop' that defines specific price action (direction, magnitude) I must see for the trade to remain valid. All of this is defined in advance of placing the actual order.

So for example if I am expecting range expansion within 5 days from entry and it does not materialize, I may take 1/2 of the position off or totally exit the trade.

I assume you are shorting in your method since you mention finance charges.

In developing your systems, starting with long equity no margin is a prudent place to begin. Once you have good results here you can branch out.

Margin can make your approach and testing more complex and often harder to define. Margin can later be added to long-equity systems once you know you have an approach that is working.

Reducing holding period often leads to prohibitive transaction costs that invalidate otherwise marginally profitable systems.

[kimchi]

[ QUOTE ]
I look for entries where I can define the risk with a tight stop. Usually the setup features an in-place trend, narrowing daily range and well-defined price zones (pivotal points).


[/ QUOTE ]

Not exactly sure what you mean here. It sounds like you're looking at the long-term trend and finding shorter-term pullbacks to the shorter-term trend.

If the ATR is narrowing, then I assume you can place larger positions with the same risk and you will be able to place closer stops.

It sounds like you're using the same basic principal that I'm currently working. I previously wroked with trend-following using a volatility/break-out model which adapted it's lookback period with changing volatility. I also used various filters to avoid taking positions counter to the long-term trend.

I'm currently testing a system the buys value instead of strength (like trend-following) finding the long-term trend and placing buy stops just above (below when shorting) the market when it reaches an extreme according to some oscillators.

This has a slightly higher success rate, shorter holding periods, and lower financing charges. It is also psychologically easier to execute these types of entries (very important) as watching a profit of 1 or 1.5 times your initial risk melt away and hit your stop (as in trend following) was a real killer.

[eastbay]

[ QUOTE ]
Do not confuse the % chance that you have direction right with your reward:risk ratio. These are two differenty things to consider. Most entries are 50-50 and when you are initially testing out your premises, this is a prudent assumption to make.

Those making comments on ENTRIES, please place them under this sub-thread.

[/ QUOTE ]

Mr. Now, how can an entry (only) be 50/50, 40/60, or anything else without an exit? Doesn't make any sense. Only an entry/exit pair can have a win% and an EV.

eastbay

[Mr. Now]

Your post covers alot of ground right away.

[ QUOTE ]

If the ATR is narrowing, then I assume you can place larger positions with the same risk and you will be able to place closer stops.


[/ QUOTE ]

This is true where risk is defined by stops and represents % of account, say 1%. I prefer entries when bars are narrow; larger size is one of the reasons. Another is that bars tend to oscillate from expansion to contraction and back. I'm wary about entering when bars are wide; I prefer to exit on wide bars.

[ QUOTE ]

I'm currently testing a system the buys value instead of strength (like trend-following) finding the long-term trend and placing buy stops just above (below when shorting) the market when it reaches an extreme according to some oscillators.


[/ QUOTE ]

This is a trend-following breakout strategy. You enter on violation of trading range in direction of longer-term trend. I like this approach. One issue is, when the trade works the bars exhibit range and volume expansion at entry. Slippage ensues since the buy stop is actually a market order.

It is sometimes better to enter narrow bars on a tight stop in advance of the potential range expansion.

A good example is SLW at 12. Notice how it hits the upper bound of the recent trading range as volume dries up. Then it prints a very narrow bar on the 2nd trading day in July. This often signals pending range expansion:

http://stockcharts.com/h-sc/ui?s=slw...p;g=0&id=0

Another variation of the breakout approach buys the retest. Often on a legit breakout, price comes back to touch the old resistence to make sure it is now support. That's a low-risk place to enter and often includes narrow bars.

Here TASR breaks above 9 and comes back to retest that level before moving higher on increasing volume. Note the small bars. (Thanks to David Sklansky for bringing TASR to attention at 7.85)

TASR:
http://stockcharts.com/h-sc/ui?s=tas...p;g=0&id=0

With the failed breakouts you mentioned before. You may be able to still use that entry method by adjusting. Keep everything the same, but buy 1/2 the total position on the initial breakout and the other 1/2 on the retest (ifwhen it comes).

Now you start with smaller size and get larger on the retest with conviction and confidence assuming the approach backtests well on the instruments you are considering.

See also:
http://www.patterntrapper.com/subscriber/NarwRng.shtml

Note that SLW above prints an NR7 the 2nd trading day in July in the above example.

Backtest and patiently test forward in small size. I am guessing you are risking 1% or more per trade which is way too large for initial forward testing.

[Mr. Now]

Look at the backtesting process as an end in itself. Decide to take total responsibility. To do so, consider mastering a real programming tool and take total responsibility for the system testing and results.

I am a programmer. I use C#. You can use VB also. Both are free and you can get them here:

http://msdn.microsoft.com/vstudio/ex...s/default.aspx

CAUTION: Steep learning curve. Very few people actually prove willing to bite this bullet.

[pig4bill]

Personally, I think that's a false economy. You can spend hundreds or thousands of hours chasing after systems that may already be in the public domain on Tradestation or Ensign, etc. Easylanguage (Tradestation) is probably going to be faster to program in as well.

Get involved in the Tradestation coomunity and you'll find hundreds of free systems available, with source code. You may get an idea you would have never thought of by using those systems, and it would be a great starting point for your modifications. I think Metastock comes with 80 systems when you buy the package, and they have lots more that they sell. You can run a list of hundreds of stocks through a system to see if it works well for any of them.

[kimchi]

[ QUOTE ]
See also:
http://www.patterntrapper.com/subscriber/NarwRng.shtml

Note that SLW above prints an NR7 the 2nd trading day in July in the above example.

Backtest and patiently test forward in small size. I am guessing you are risking 1% or more per trade which is way too large for initial forward testing.


[/ QUOTE ]

Interesting charts & analysis - thanks.

I tested using 1% risk excluding charges, and forward tested using the smallest positions I could take with my small account size. This was generally around 1% as I selected less volotile markets for testing so I could take less risk.

I used spreadbetting (mug's game) and CFDs (taxes and expensive).