Random Trading

Originally posted by MUChris
Hey Vlad,

I've got some UAL, WCOM, and Pets.com I'd love to sell you bro.

Chris
More...

I was talking about EXPECTATION! The OUTCOME/REALIZATION does not have to be anywhere near what is expected in any given case. When you bought those, you had to be expecting a positive return. Otherwise... you need serious help.:(
 
Originally posted by MUChris
Oh, by the way, we weren't talking about if YOU had a positive expectation on a particular stock, but if EVERY stock has a positive expectation
More...

It <b>DOES</b> dude, do you have a BA in English Literature or what???? Stop wasting everyone's time.
 
Originally posted by profitseer
Streaks are an almost certain statistical certainty. You don't have to believe in them. They are not just anecdotal. I gotta get back to work. After the close I will try to find the proof. It is all formulas, I don't understand them, but maybre somebody else will.
More...
OK, you are right. I guess I was not very clear on what I meant by believing in streaks.

The issue is that there are two definitions or perspectives on what one might mean by a "streak". By one definition, when one studies long sequences of coin-flips, one will indeed see long steaks of heads that seem improbable. In any random sequence of 1000 coin flips, a streak of say 9 or 10 heads in a row is quite likely. But this is the "middle of the chart" perspective where one can look across the sequence of coin-flips and pick out strange patterns anywhere they occur.

In fact, if one looks for strange patterns, one is bound to find them (e.g., long strings of heads, long strings of tails, long strings of perfectly alternating heads and tails, etc.) People are really good at finding this anomalous sequences in random heads-tails data. Studies of human cognition show that the human definition of randomness tends to flip from heads to tails too often and that truly random heads-tails sequences "look" nonrandom to people (i.e., they see "patterns" in random data). By the first definition, streaks do occur.

The second definition of a streak is predictive -- do I expect the very next N flips to all be heads or do I expect the very next flip to be tails after a long "streak" of heads. This "right side of the chart" perspective is the important one for a trader. For a trader, the question is NOT do streaks ever occur? Instead, the trader is concerned with betting on whether the next flip is a continuation of a streak or not. For a fair game of coin flipping, predictions of the onset or end of a streak is a pointless exercise, the coin does not know or care and stubbornly sticks to its 50-50 odds. By the second definition, streaks do not occur.

Sorry for the confusion, but finding patterns in middle of the chart and predicting/betting on patterns on the right side of the chart are two very different things. Streaks occur in the middle of the chart, but are not predictable on the right side of the chart. (Now, whether the market or a trading system follows the same statistical patterns as a coin-toss is an another, and more important, question).

Happy trading,
-Traden4Alpha
 
Originally posted by permabull
that was one of the most informative posts I have read on this board.

but the question remains unanswered, are the markets random?
More...

Of course not, are you kidding??? Regardless of whether you take a long term perspective or a short term one, it's easy to come up with a number of reasons for nonrandomness. Some of the moves are certainly noise, but many aren't. Predicting them is a different thing though.

Now, if (a big IF) markets are semi-strong form efficient, and all public info is indeed instantaneously and correctly incorporated into prices, then the future moves will depend on future (random) information...
To the extent that it's not true (plus b/c of all market frictions) the moves will not be completely random.
 
Well, If the market moves up one tick, every time heads comes up, and down one tick every time tails comes up, at about noon today, the guy in Philly who flips the coins, sure was flipping a lot of tails very fast.
 
We all want to know whether the markets are random. I started this thread trying to think of the markets in terms of coin flips. BKuerbs pointed out that in a 100 tosses of an unweighted coin, one should expect 50 heads with a standard deviation of +-5. Two std. dev. would then be 50 +-10. And he stated that in any 100 toss sequence we should see 40 to 60 heads with a 75% probability.

I indicated in a previous note that if one just looks at updays versus down days(that is price direction only) for any stock, the number of updays is about 50 for the past 100 days. To put some real meat on this statement the table below shows the data for 102 NASDAQ stocks for the past 100 days. The first column is the number of updays grouped in to bins of 5. The second column is the number of stocks having updays in this range, and the third column is the percentage that this number of updays represents(by dividing column 2 data by 102):



Code: 
 Upday Range              No. of Stocks              Percentage
     _____________________________________________

      61-65 updays                 0                      0%
      56-60                       1                      0.98
      51-55                       6                      5.88
      46-50                       31                    30.39
      41-45                       42                    41.18
      36-40                       21                    20.59
      31-35                         1                   0.98
      26-30                         0                       0
    ______________________________________________
                  Total          102                       100%

If you add up the percentages from 41 to 60 you get 78.43%.

My interpretation of this is simple. Stock price DIRECTION is random and follows a binomial distribution with a probability of an up move or down move of 0.5 for either direction independent of what the price ACTION(that is the amount of the move) might be.

Those of you with Metastock can carry out the above test on your own portfolio of stocks using the following formula in EXPLORER

UpDay:=If(C>Ref(C,-1),1,0);
Up100:=Sum(UpDay,100);
Up100;

The first line simply assigns a 1 to everyday for which the day's Close is > than the previous day's close, otherwise a 0. The second line simply sums the values for the past 100 days and the third line displays the result.
 
On trending days, stock prices generally follow the overall market or sector sentiment; therefore, this is predictable and not a random occurence. On choppy days, stock movements can be seen as random. Stocks with very bad news will tend to go down and vice versa; again, this is not random.

So is the market random? The answer is sometimes yes and sometimes no. The goal is to find the high probability scenarios and trade them and avoid the low probability ones. Simple as pie!
 
Markets consist of three phases of human behavior: fear (selling), greed (buying), and confusion (consolidation). These phases are orderly and quantifiable, hence price patterns.

Trading is not about PE's and balance sheets. It is about - to paraphrase profitseer - quantifying investor behavior and responding consistently to it.

That the three phases occur in irregularly irregular fashion is the randomness.

But we don't care because we are quantifying sentiment and reacting to it. Again and again and again.

One thing is for sure and definitely not random...
 
You must log in or register to reply here.
Back
Top