Predictive power of SCTR, any hard data?

mastercharts

I am trying to see if anyone has done any studies of correlation between buying securities with high SCTR and the rate of success or return per trade?
I did a bunch of long only runs with simple trading strategies on SPY, QQQ, IWO, XHB, XBI and DXJ for 2012-2015. I tried to see if trade success rate and return correlates with a high SCTR at the time of entry. What I found is that, if anything high SCTR was actually a detrimental factor in the rate of return, and marginally helpful in improving success rate.
See attached for my results.

Any feedback would be appreciated!

Best,

Alex

markd

Well, to be very technical and picky, the SCTR Line on the weekly chart appears to be a - sampling - of the data used to draw the daily SCTR Line.

Smoothing implies all the data for the week is averaged in, or otherwise mathematically manipulated to come up with a new value. In this case, I think they are just plotting whatever raw daily value comes up on Fridays after the market and ignoring Monday through Thursday. I haven't checked what they are doing mid-week on a weekly chart - probably as you suggest using the closing value for the current day.

markd

Hi Alex, interesting question and results. Designing a good test is always a challenge, and it's always difficult to say whether the results are representative of anything beyond the sample that produced them.

That said, I share your skepticism of a very simple approach to using the SCTR rankings. It seems that often when a stock rises into a high or very high SCTR rank, that's the peak of the move for that stock. On the other hand, there are stocks that hit say, 90, and hug that level for weeks or months. The problem is, hitting 90 (or 70 or 80, etc.) is not by itself an indication the stock will continue to gain. So what is?

Something you might like to look at is adding the SCTR Line indicator to your charts and see how it varies with the stocks' price action. I have it on my weekly charts with horizontal lines at 20 and 80. At least on some stocks, it seems to behave like an oscillator - when it gets below 20, think about a buy set up, above 80 a sell set up. Another interesting use seems to be, if a falling stock's price goes lower for a few bars, but the SCTR doesn't follow, that might be a buy. Vice versa at highs for rising stocks (see XOM June 2014 vs. May).

One more thought - your test was on index based symbols. I think SCTRs was really intended to measure actual equity issues against other equity issues with similar characteristics. You might get a different result if you picked several stocks from the same industry and compared their SCTR behavior.

Kevo

Since SCTR is a daily calculation, would it not make more sense to put it on a daily chart instead of a weekly?

I thought if SCTR goes below 20 that's more like shorting territory? Should you not wait until at least it crosses above 50 before considering to buy?

I tend to use SCTR as an additional filter or alternative filter in scans. For example, if I am looking for oversold symbols where the long-term trend is positive, I might add SCTR > 55 as an additional/alternative filter.

Maybe on high SCTR symbols, when the public rushes in to buy, it becomes overbought and then falls. Watching price and volume action becomes important.

markd

Hi @Kevo , I'm assuming the weekly chart just uses Friday's daily SCTR value, like it uses Friday's daily close. Good enough to get a general idea, I think.

20 could be shorting territory, just like 80 could be buying territory, provided the trend in place is going to remain in place. As you rightly note, the price and volume action is important for making that judgement.

What I wanted to say was, from casual observation, it seems that relatively few stocks do trend for very long, or perhaps I should say, relatively few stocks maintain their SCTR level for very long. For many stocks, it seems that by the time they reach very high or low SCTR levels, they are ready to turn.

I think a way to use SCTRs to find emerging winners might be to test for an improving SCTR, maybe from below 30 some weeks ago, to above 40 or 50 today, or whatever levels you feel comfortable with. You would miss the beginning of the move, but might get the most profitable chunk out of the middle in the shortest time. In Wyckoff terms, you might catch the mark up phase.

Kevo

Thanks @markd

I should have known... Looking at historical SCTR from a weekly perspective should indicate if the SCTR has staying power, and maybe a 50 line cross on the weekly is stronger? maybe if it stays above 50... It's now added to a weekly ChartStyle. Thanks.

I'm sure I have tried this before and forgot... it looks like the end value of the SCTR line on the Weekly chart matches the end value on the daily chart during the week. So, for the current week, they match. For the previous week, are you saying the Weekly SCTR line uses last friday's close? I am fairly sure that the SCTR line has to be scaled up somehow for the weekly chart.

I have the horizontal line on the 50 in gray color.

markd

There doesn't seem to be documentation for SCTR Line as an indicator, except Chips blog post announcing it's available. I don't think it would need adjustment for a chart time frame, since the value itself is derived from multiple time frames, so it's independent of the chart settings you choose - unlike say, MACD or RSI which would use weekly data on a weekly chart but daily data on a daily chart.

Incidentally, Chip's post has a link to an excellent example of SCTR Line improving in the weeks before a huge jump, while price itself stays flat.

stockcharts.com/h-sc/ui?s=GOOG&p=D&b=3&g=0&id=p24012235419

Kevo

@markd ,

If I am reading you correctly, it looks like you are saying that the SCTR line appears identical on Weekly and Daily charts, regardless of the period. Since I was surprised to see this, I went to check, and I see that they are different. Thus, something is done differently to plot SCTR on the weekly like maybe use Fridays closing SCTR for the weeks previous to the current week like how you suggested to begin with.

The new ChartSchool does have an article on SCTR:
http://stockcharts.com/school/doku.php?id=chart_school:technical_indicators:sctr

markd

Yes, but the SCTR article does not explain SCTR Line.

I checked a chart (AAL) with SCTR Line, set the dates with the Select Start/End date option, and switched back and forth between daily and weekly. The SCTR values were the same for both time frames.

Kevo

Thanks for the clarification of both points.

I zoomed in using Select Start/End and I see that the weekly SCTR Line is like a smoothed version of the daily. It is clear to me that the weekly has less data points than the daily. I figure the weekly uses friday's close - for all the weeks prior to the current week.

It's good to occasionally dig into the details for better understanding.