Wednesday, July 25, 2007

Trend Days

Some of my own observations of trend days.

Read Full Post.....
I have an archive of the eMini S&P500 Futures intraday charts from the past three and a half months. I went back to review the charts to see how many of them were trend days.

Before I share the results of that research, I want to first qualify what makes a trend day. The Market Wizard Linda B. Raschke wrote an article about it. She characterizes trend days as follows:

  • open and close are near opposite extremes
  • the opening half hour makes up a small portion of the day's range
  • intraday price retracements are minimal and shallow
  • daily trading range has expanded
To that, I would add the following additional characteristics:

- typically, either the 10 or the 20MA has not been breached for more than 2 bars during the life of the trend. At a minimum, it must follow a manually drawn trendline. This is paraphrasing LBR's point about minimal and shallow retracements.
- trends typically take 3 to 6 hours to run its full course, although I have seen nice trends start and finish within 90 minutes. From this we can infer that as the trading session progresses without any signs of a trend, then it becomes more and more unlikely that a trend will develop. The absolute last chance for a trend to develop would be around 2pm EST.

Anyways, my own review of the daily ES charts reveal 26 trend days out of the 86 trading sessions that I have archived. That means a trend day will occur 29% of the time. Just to be real conservative and unbiased, let's throw out the worst 10 of those trend days. Then that will reduce the frequency of trend days down to about 18.6% of the time. Well, guess what, that means on average, a trend day will occur (almost) once per week !!! Yes, believe it or not, a trend day occurs more often than you think. So, the implication of this is that each trading session that occurs with out a significant trend taking place actually increases the odds of trend day happening in the subsequent trading sessions.

So, starting from the acknowledgment that we will never truly be able to predict with 100% accuracy whether today will be a trend day or not, we can still gather pieces of information which will help give us an edge in identifying trend days.
In that same article, LBR outlines some tips to help identify which days will be a trend day. I want to increase the odds even more. Here are some additional things which I think will help me identify a trend day.

- economic reports. Nothing moves the markets better than a better or worse than expected economic report. Sometimes, these economic reports can trigger a trend day, especially reports that are released at the 530am or 7am(EST) times. Some of the more significant ones that are known to move the markets include the CPI, PPI, retail sales, and of course, the Federal interest rate announcements.

- breach of a significant support/resistance level. Breakouts fail as often as they succeed, but the ones that do actually succeed tend to set the tone for the day, and thus more likely to produce a trend day.

Ok, that's what's in my head at the moment, hopefully I can use this knowledge going forward to take advantage of upcoming trend days.


Richard said...

I'm glad to see your description of what you call a "trend day" because we were not on the same page on wallstreak. I would not have considered a day with a 90 minute trend a "trend day."

I think you are off in your thinking about the odds, though. In my archive of 100 coin tosses, I've found that 50 of them came up tails. So, the implication of this is that each coin toss that occurs without coming up tails actually increases the odds of tails in subsequent tosses.

I don't think so! :-) What you're saying may be true, but it doesn't follow from the data you've presented.

Phileo said...

I understand what you are saying, ie. the probability of a trend day occurring on any given day is only 18%, even if 20 days have passed without a trend day.

However, comparing the markets to a random scenario like a toin coss is not accurate. I believe the market does not behave in random fashion. Chaotic, erratic, and volatile fashion, yes; random no.
The markets follow patterns because the people behind them are creatures of habit.
Therefore, I still stand by my point that each day that passes without a trend day means a trend day is more likely. Yes, I don't have data to back that up. But that does not change my belief