This post was contributed by a guest author, and does not necessarily reflect the views of Richard or MovetheMarkets.com
I’ve been working on a swing trade system that buys breakouts of NR7 candlesticks on a daily chart. The entry would be a break of the high plus a tick or two, and the stop a bit below the low of the NR7 candle. Jamie has been providing his trading expertise, and I have provided some mathe-magic. I use the fabulous stock scanner Stockfetcher to find the candidates. It’s easy to use and very easy to program it to search for what you want. The 2.0 version has a great interface.
This is what I’ve got so far in my Stockfetcher code:
Show stocks where close is above 1.00
and average volume(30) is greater than 200000
and volume is greater than 50000
and average day point range(1) reached a new 7 day low
and average volume(3) has been decreasing for the last 5 days
and volume is more than 30% below volume 1 day ago
and volume is more than 30% below average volume(3)
and draw ema(5) and draw ema(30)
and date offset is 0
Most of those things are self explanatory. Average day point range(1) gives the range of the day in a clean way, and finding a 7 day low gives you an NR7. Stockfetcher can also do percentage range, but I’m not sure whether points or percentages are better to use. The volume statements are there to look for a volume contraction along with the range contraction. I used an average volume to allow for a bit of wiggling, though the volume trend will be right. I picked the different volume numbers and the 30% volume reduction to narrow the number of candidates. Date offset lets you see scan results from the past–an offset of 2 gives results from 2 days ago and so forth.
Looking back, the following symbols were results of a scan on Monday, 4/23:
SAI, ATHR, PIR, UTIW, WLT, ICFI, DLIA
Out of those, SAI, ATHR were triggered and were up today. The others didn’t trigger an entry, so no losers on that day. In my examinations on other days, some candidates have triggered an entry and then dropped to stop you out. The system and the scan still need work in that regard.
Running the scan today, I get the following stocks as NR7’s at the close:
JDSA JSDA, JASO, IMAX, ILE, AIXD
We’ll see how they do tomorrow. Please offer any suggestions to improve the system or the scan!
EDIT: Per the comments, here’s the new code that yields tons of results:
Show stocks where close is above 1.00
and average volume(30) is greater than 200000
and volume is greater than 50000
and average day point range(1) reached a new 7 day low
and volume is more than 30% below volume 1 day ago
and date offset is 0
/*Lines below are just for chart formatting*/
and draw ema(5) and draw ema(30)
and draw average day point range(1) 7 day high
and do not draw average volume(30)
This post was contributed by a guest author, and does not necessarily reflect the views of Richard or MovetheMarkets.com
April 24th, 2007 at 9:20 pm
Nice work Prospectus!
I will put these stocks on my watch list for tomorrow.
Just curious about the 7th line of code re: volume less than average by 30%. Let’s say we have a stock that gaps down followed by three days of selling on declining volume. IE. Big volume spike on the gap down and above average volume on day 2 and 3 but declining from the spike. The fourth day it prints a NRB on volume that is 30% less than the previous day. Will it show up on the scan?
Check out the chart for AVCI for a visual of what I tried to describe above.
Look forward to your feedback. And thanks again for the mathe-magic.
April 24th, 2007 at 9:41 pm
Isn’t it hard for a stock to be 30% below an average that it’s taking part in? or did you mean 30% below the 3-day average from 1 day ago?
I personally always use % for price movement, but for this scan in doesn’t matter much because it’s only comparing recent amounts, and not talking in absolutes.
April 25th, 2007 at 5:52 am
@Jamie: Thanks for your help! Math is useless without trading knowledge.
In the case of AVCI, it would not make the scan as written above. The average volume(3) is not decreasing for the last 5 days–it’s only decreasing for the last 2. It looks like that filter is too restrictive, so I’m tossing it.
If I take out the references to average volume(3), then the scan returns 88(!) results for yesterday, including AVCI. I need something else to cull the candidates to look for the best ones. My first thoughts are to crank up the volume requirements and lower the volume contraction requirements. Any suggestions?
Keep giving me examples like AVCI, and I can keep tuning the scan to look for the right kinds of things!
@Richard: The volume can fairly easily be 30% below it’s moving average both taken from the same day. Volume can expand and contract with much more volatility on a percentage basis than price typically does.
I only noticed a slight difference in using % vs. points for defining an NR7, and that was when you had a slight rise in a stock over a week. You could have a range that’s slightly smaller at the end of the week and would be an NR bar on a points basis, and the same bar could be slightly larger on a % basis and therefore not an NR bar by that definition.
My scan returns too many results, but I don’t know the kinds of things to look for. I can math the sucker to death, but without trading insight I’m just shooting in the dark. :)
April 25th, 2007 at 7:01 am
JASO and IMAX are up. The only problem is you could drive a truck through the spread!
April 25th, 2007 at 10:46 pm
If you believe that decreasing volume for 5 days is better than decreasing for 2 days, and the 5-day filter still finds plenty of stocks, then I don’t see how that’s too restrictive.
If you want to identify decreasing volume in a fuzzier sense than “strictly decreasing for 5 days”, maybe look for decreasing momentum of the moving average of the volume? Or, maybe, compare avg(3) to avg(5) to avg(7) or something. So that, recent volumes tend to be less than slightly less recent volumes.
April 25th, 2007 at 10:48 pm
Oops, I see that you were doing momentum of the avg volume in the first place… but maybe comparing the different average spans will work better for you.