Wednesday, October 20, 2010

Dow Industrial Forecast as of 10/20/2010

I rarely want to do post hoc analysis on these forecasts, but to demonstrate how close they have the potential to be, you can see how close the actuals were to one of the forecasts yesterday. Now the question is: Will it continue on that path? Let's see what the updated forecast looks like first.


Here are the close predictions for today. You can see that the orange path shifted a little, but the picture remains fairly consistent -- a push in the Dow to 11,200 or so and then a pullback. It feels like the Dow wants to hit that area, one way or the other. That's the consensus of the projections, but as I saw yesterday, the consensus can be wrong, so does that give the purple forecast a chance? Possibly. The red forecast is highly unlikely, in my opinion.


The high/low channel also seems to favor a push to 11,200, but I wouldn't be too bullish, since the gains are contained to that level, which also happens to be key resistance. A break above 11,258 is a big event because it gives a bullish signal. I don't believe it will happen though.

12 comments:

  1. tempe .on cpce chart you are noting open postions. Dose that mean close previous postion?

    ReplyDelete
  2. I'm working on how to better perfect that signal. I see a clear wedge on 5 period EMA of CPCE, so that's bearish overall, but I'd like to see a break down of the 20 EMA before opening short positions. To be honest, I'm already pretty fully short, but I will add more on that signal.

    Here's a new chart I'm trying to work on.

    http://stockcharts.com/h-sc/ui?s=$CPCE&p=D&yr=0&mn=8&dy=0&id=p68373158031&a=202131234

    ReplyDelete
  3. tempe - i checked your cpce chart just now and you have updated it with new rules. Thanks. Although i am still trying to understand your new rules one thing came to my mind is you are looking for divergance first (and thats why you have mentioned about no divergance still today) on ema 5 but what about looking for divergance in cpce itself (instade of ema). As I see all previous cases qualify for divergances in cpce itself (in addtion to ema 5 too) and there is currently divergance in cpce althought its not in ema 5 yet.

    ReplyDelete
  4. Very true Yash. You could be right. I am sort of expecting that 5 EMA to come back down in the next few days to form a divergence, but you are right, could be just a divergence in CPCE only is needed. Let's see what happens. I do think that wedge breaks up and the market falls though.

    ReplyDelete
  5. ema 20 break means spx ema 20 break ..correct? that is good extra condition i think..means extra safeness I think. but then if we are looking for ema break on spx itself then may be ema 13 is enough too. rest all rules using cpce divergance , its slope (ema of slope) may be good enough. conditions of slope > 0 and < 0 still exist ..correct? Will you try ema 13 instade of ema 20 on spx and see what happens.

    ReplyDelete
  6. One more thing is reliable cycle man has important cycle low this week and he has given time till wekeend. After that it should go up at least sometime in november to get real down. Not sure how that will spell out on this cpce chart I don't know. May be it won't give short signal all the way upto november high.

    ReplyDelete
  7. Yes, a 20 EMA break on SPX. I haven't toyed with other times but that could work better. Someone has a good simple system that uses a 3/10 EMA crossover. I'm following this as well since it works pretty well. Here:

    http://stockcharts.com/h-sc/ui?s=$SPX&p=D&yr=0&mn=10&dy=0&id=p58773466613&a=211623130

    The cycle guy you refer to could be right, since the elections and Fed meeting don't happen till early Nov. We'll see. Could be a surprise coming at any time.

    ReplyDelete
  8. I think these ema systems are good and simple. ema 13/34 is very common swing cross overs of course 3/10 of this chart are more closer to trend change. In todays world I think its best to stick to such simple system as otherwise we get infleneced by outside factors.

    ReplyDelete
  9. Great Job! Your Deja Vu charts are wonderful. I have a question. When you show the 4 different alternative paths, is that because you found one instance of each of these paths for the best matches in the past? It would be interesting to try to group outcomes of the top (say 25) best matches into perhaps 3 or 4 trajectories that are most similar to one another. Then you could indicate the probability (percentage of past occurrences) each of these groups happened (i.e., trajectory 1 happened 10 times out of the 25; trajectory 2 happened 6 times out of the 25; etc). Of course, this could perpetually be tweaked by trying to account for "all things being equal" in the markets.

    ReplyDelete
  10. Peter, thanks for the suggestion -- it's a good one. I will think of ways to incorporate what you're thinking by somehow grouping past history. My methodology is a work in progress.

    As for what I present, they are simply the best historical fits for a 2 period, 3 period, 4 period and 5 period cluster of daily returns. So for example, on 3/24/2009 and 3/25/2009, the market posted returns of -1.49% followed by 1.17%. Yesterday and today posted similar returns. So what I do is take the next two days of returns and plot those out as a potential forecast for the 2 period cluster. I do the same for a 3-5 period cluster of returns (that's why there are 4 lines). I take the best fit and plot them out as a possible forecast.

    My current thinking is that the market tends to move in "thrusts", and if you see a certain thrust occur now, it's possible it will repeat the same pattern as the past. My algorithm can look back as many days as I want, so I could find the best year that fits the last year of data in terms of returns. The catch is that the difference is probably significant, so I'm not sure how useful it is for going back more than 5 days or so.

    Anyway, thanks again for the suggestion. There will be more to come as I think of some other ways of presenting a forecast.

    ReplyDelete
  11. I think these deja vu charts speak to me a bit, becuase I live in Florida, and this is alot like the Hurricane path models we have to watch every year HAHAHA

    as the storm gets closer, one path starts to pop out a bit more, and then other models will usually start to come in line and you can see where it will go

    example:

    http://icons-pe.wunderground.com/data/images/at201019_model.gif

    ReplyDelete
  12. haha Brian -- the markets may resemble a hurricane soon enough!

    ReplyDelete