It may not be strict enough but to be tested it has to be classified which is exactly as I have done. Granted I could be picking up on late day reversals but again a market, which opens on it lows and closes on its highs or vice versa is a Trend Day. The filters are exactly in my post above I am classifying a market as having an Up Trend Day being one that opens within 10% or 25% of its lows and closes within 10% or 25% of its lows - vice versa for Down Trend Days. The histograms are showing the distribution of Opening Range as a percentage of the 10 day average range on what I have classified as Trend days or Non Trend days. So if the above were true you would see a bulge in distribution occurring with opening ranges below 30% which we are not. This makes absolute sense to me as a Trend Day by definition should move of the open quickly, hence the opening range is in all likelihood going to be larger than a Non Trend Day. My last paragraph should clear this up as the Histograms show all OR%'s but of a 10 day as oppose to what you suggest which is a 5 day. There is no point in filtering for opening ranges <30% of the 10 day average range. By being selective you cannot see the the whole distribution and therefore are unable to prove if its statistically significant or not. Perhaps you can share your results as I have done so we can see where we are not understanding each other. Thanks.
Are you kidding me buddy, a market that closes on it's highs or lows is automatically a Trend Day in your book? If we can't agree on the definition of a Trend Day then not much point in talking.
You already have mine from my previous post and I am now waiting to hear yours. If you read my previous posts you will notice that I posted from an objective standpoint by providing evidence to support/negate the discussed subjects. I would highly recommend that you start responding in a similar manner. Can you also please stop responding tu quoque, since its very annoying and does not aid in debate? Thanks.
This is a hell of a Trend Day in Wheat back on March 1st. The Low was on the Open and High was on the Close. Just ignore how it chopped sideways for 4 hours near the Lows.
But what you have just stated completely disagrees with what you told me about trend Days in this post. Please clarify, Thanks.
If all you are testing against is if the Low is at the Open and if the High was at the Close then you are completely missing everything that is happening throughout the trading day. I can't believe that I have to sell you on the idea that the Opening Range is significant. Have you read the book "The Logical Trader" by Mark B. Fisher? If you can't write code to identify what a Trend Day is then I can share some logic with you to help identify what a Trend is. But I should not have to argue with you over what a Trend Day visually looks like. For an uptrend think stair steps going up from left to right, here is a visual aid:
As I have said previously that is how I classified a Trend Day. If yours is a better classification then please, as I have asked you to before, post your description here. You don't have to sell me on anything and yes I have read "The Logical Trader " by Mark B. Fisher. I can write any code I want but, unfortunately as I said in my first paragraph if you are not going to give me your full classification for a trend day then where do I start? By sharing the logic with me hopefully that should clear our ideas around what a trend day is and then my test results should be the same as your results - which you have still not provided. Once again another different version of a Trend Day. Are you sure you have classified this in code, as you have stated? To Summarise; I have given you the classification that I used for a Trend Day along with providing the distribution plots. You on the other hand have done nothing else aside from posting tu quoque, a few hand picked charts and questioning my own knowledge of ACD. If you can't produce evidence to substantiate your claims I suggest you put up or shut up.
It's a great discussion and we could actually get somewhere. There is no indisputable definition of a trend day and we could debate it back and forth and get nowhere. Steidlmayer suggests a range extension that is CONSIDERABLY more than the Initial Balance area. It's a few years since I read the TPO stuff but I think he used 30 minute brackets and the IB would be the 1st hour. I think it's fine that we have whatever OR we use, but if our OR represents 20% to 30% of the ATR, then the range extension should be >2 times to 4 times the OR, say a minimum of 2 to 3 times the OR, ending pretty much at the extreme. Personally I would accept within say 10% of the extreme, given the very common profit taking into the close. I'm not looking at any of this from the perspective of what are best definitions, simply if we can get a working definition that would allow analysis yielding a useful conclusion that can be acted upon. http://www.investorguide.com/definition/trend-day.html http://www.mypivots.com/dictionary/definition/216/trend-day http://www.mypivots.com/dictionary/definition/96/initial-balance-ib
I guess pages 10, 11 and 12 of The Logical Trader flew over your head. You are still defending your lazy definition of a Trend Day. Sure here is some basic logic for detecting a Trend Day UP. Adapt it to whatever language you write code in: Iterate through your main loop of intraday data Uses 5 min bars for (i=Start_Of_Regular_Trading_Hours+1,....,...) { if 5min_bar is within regular trading hours+1 then observations++ if Close_Of_Current_Bar>Opening_Range_High then var1++ } if ((var1*100)/observations)>=75 AND High_Of_Day is within 9 Bars from Close_Of_Day { printf ("Congrats you have a Trend Day UP!"); Trend_Day_Up=True Trend_Day_Up_Count++ } Let's try this for a start.