Entries only – thank you.

Greenhorn backtester here.

I’m following the recommendation of Chuck Le Beau from his book "Technical Traders Guide to Computer Analysis of the Futures Markets" and testing entries that have an exit x days from entry. He suggests that the winners be greater than 55% before employing exit strategies. I like this basic starting point so that is what I’m doing.

I work for a living and can only study the market in the evening and weekends, and therefore I’m totally end of day data based.

I’ve been testing entries and a hold for 2, 5, 10, 15 and 20 day periods.

My background is point and figure charts, and I cut my teeth always purchasing on pullbacks. So, pullbacks are what I have been primarily testing to date.

Here is my hoped for discussion. Entries, entries and more entries with holds from 1 day to whenever we get taken out. Can they be discussed without disclosing your edge? If they can’t, then this will be just another quickly dead thread.

I’m wondering if your entries are primarily based on pullbacks, breakouts or are they about even? Do you concentrate on patterns, indicators, or a combination of the two? (The vast majority of my entries so far have been based on patterns. I am doing poorly.)

Okay, I just typed the above, and just before hitting the go button, the little voice said, “What the hell are you doing? If you knew backtesting jack crap would you help this internet land neebie?” I said (I talk to myself all the time) “Yes, if I could do so without disclosing any edges, but perhaps leading him to do his own research and find success, I would”.

Well there you go …

Gary Fox

 

I think a very big hint is available at no charge from the nice people who run Wealth-Lab's website. Take a look at their page of Top-25 performing systems (which they call Chart Scripts): http://www.wealth-lab.com/cgi-bin/WealthLab.DLL/getpage?page=Top25.htm

Notice that the ones which use Market Orders Only ("can be easier and less stressful to trade") are consistently worse than the systems which use Stop and/or Limit orders. By a lot. A whole lot.

Conclusion? If you want maximum profitability, stick with entries (and exits) that use Stop orders and/or Limit orders.

Here is one example entry that, in my testing anyway, performs nicely on commodities. "P1" and "P2" and "P3" are user adjustable parameters that you, the trader, select. P1 and P2 are integers. P3 is a decimal number like "2.5" or "0.7"or "3.333".

a. Calculate ATR = AverageTrueRange(P1 bars)
b. Calculate M = MovingAverage(Close, (P2 bars))
c. Calculate U = M + (P3 * ATR)
d. Calculate L = M - (P3 * ATR)
e. Enter long next bar at price = U on a Stop order
f. Enter short next bar at price = L on a Stop order

This is a Volatility Band Breakout entry, where we are using AverageTrueRange as our way of measuring Volatility. The Bands are at plus-or-minus P3 ATR's away from the moving average. Notice that we enter on a Stop order (not a market order), following the hint given by Wealth-Lab.

 

Thank you Sir!

I will look over that site, and have a go at your suggestion. Much appreciated.

I recently dropped you a note at another thread, as I enjoyed your postings.

http://www.elitetrader.com/vb/showthread.php?s=&threadid=100591&perpage=6&pagenumber=8

I received the idea book last night. I'll get after it tonight with a drink and cigar.

Thanks again.

Gary

 

i would strongly suggest a small little book by perry kaufman called smarter trading.

every market has a consistent definable cycle. nobody can find it not even the cycle gurus for ages now. you need an edge to make money in the markets so i will not say more on that.

once that cycle has been found you then are left with the scraps of dat which can be dealt with in the following manor.

you can see if there is enough room to make a profit or you can simply stop out rather than stop and reverse. either way the cycles are where the money is at.

mb

 

Thank you Sir!

I've seen the book. I'll get it.

I got a feeling there's some real gold right here. Hope I can find it.

Thanks,

Gary

 

i traded that exact signal for several weeks and had a really fantastic run with it initially, then the market i was trading sniffed me out and blew it out of existence lol. ...resulting in a death by 1000 cuts type drawdown before i turned it off

that was maybe 6 months ago and it's just barely creeping out of the dd now across a wide parameter space.

i would beware with bands based on ATR times a coefficient. i know it's just being presented as an example, and it's a very good one, but it's also very easy to curve fit that code. maybe there are other means, for example continuous optimization to keep something like that working, but i've never implemented anything like that

 

Hello AC,

I'm at work, so I don't have my backtest results -- but I do recall running the suggestion with various criteria over the russell 3000 with look backs of 100/300/500/1000/2000/ days.

Performed rather poorly as I recall. Caught the tops of many swings. Now, if one was to revisit this with a counter trend in mind .....

Gary

 

It is thery good that you send us your simply sistem. and I for my part undertake to show all the Perfomance of this sistem.

 

Hello Mark,

You have the part about "Nobody can fnd it not even the cycle gurus for ages now" dead right. Prior to your post, my ignorance relative to cycles would fill the old Kingdome. Now, it would fill the Kingdome less a dustpan. Even at ET, they get no love. This thread is a good representative:

http://www.elitetrader.com/vb/showthread.php?s=&threadid=54465&highlight=cycles

In a nutshell: The majority here, and on the net (other than vendors) say cycles add very little to trading success.

I guess you have found something in your cycle analysis that in fact does work. That sir is cool! I guess that would be a monster edge.

Take care,

Gary

 

Hi Gary, I guess my post about position trading wasn't very clear, I trade stocks, not futures.

I look for stocks breaking out with heavier than usual volume, (you can use your tc2000 to scan for them, they have a pre-canned scan called 5 day volume surge, or you can write your own pcf), and I buy breakouts inside the pullbacks.

My favorite book is "How I Made $2M" by Nicolas Darvas. He too bought stocks that were breaking out on unusual volume, but his entries were at the new high, not the pullback.

Some examples of my better trades are buying NTRI in March of 05, but I sold too soon in June of 05. I also bought PTC in in March of 05, sold it about four weeks later. I shorted MTH in Jun of 06, but that was based more on sector analysis of the housing market than by technical analysis.

I cut my losses short so I don't have a lot of worst case trades, although I took a huge loss in BKHM in April of 06 soon after I bought it when it gapped through my stop. There's no way to protect against a gap move like that except by not having too much of your account in one position, which I unfortuanately did at the time.

As of right now the only position I have is ZIXI which I bought on Nov 19 at 3.65.

I mostly use my own scan in TC2000 that looks for a 20% increase in 3 day volume twice in a row. I often catch stocks before they show up on everyone elses radar by seeing them with subtle increases before they really get going. The disadvantage of it is that you see a lot of illiquid stocks because a they often have subtle increases in volume.