Allow me to state what I should have stated before: all coin-flip trends are coincidental. How could they be otherwise, since they are all randomly generated? No reason for one head to follow another other than chance. Your first mistake is the following. You have asked many times the question: If you flip a coin and get 10 heads in a row, are you in a heads trend? You expected the answer to be NO, either because of the coincidences or because you thought the trend was unexploitable. As I have shown elsewhere and will repeat below, the trend, properly defined, is very exploitable: Your first bet: $0 Toss # 1: heads ; Total win: $ 0 ; Your next bet: $0 Toss # 2: heads ; Total win: $ 0 ; Your next bet: $0 Toss # 3: heads ; Total win: $ 0 ; Your next bet: $2 on heads Toss # 4: heads ; Total win: $ 2 ; Your next bet: $3 on heads Toss # 5: heads ; Total win: $ 5 ; Your next bet: $4 on heads Toss # 6: heads ; Total win: $ 9 ; Your next bet: $5 on heads Toss # 7: heads ; Total win: $14 ; Your next bet: $6 on heads Toss # 8: heads ; Total win: $20 ; Your next bet: $7 on heads Toss # 9: heads ; Total win: $27 ; Your next bet: $8 on heads Toss #10: heads ; Total win: $35 ; Your next bet: $9 on heads Toss #11: tails ; Total win: $26 ; Your next bet: $0 So putting just $2 of your principal at risk in a timely fashion could earn you $26 (or more if the trend continues). That's the power of trading with the trend (in this case, the trend is 3 or more heads in a row). So if the random trend is sufficiently long, it very obviously can be exploited. But most randomly generated trends are too short to be exploited profitably. And since you never know in advance which trends will be long, trying to exploit them all will make you a net loser. Your second mistake is assuming stock market trends are no less coincidental than coin-flip trends.
Your first mistake is the following. You are not defining trend correctly, trend in the markets is perception by those who trade and in flipping a coin it is "chance". If anyone was to buy the stock market past thirty years, many more times it went up than down. If you flipped a coin past thirty years heads in theory it be 50%. Las Vegas will show that chance is minuscule exploitable.
Your mistake is not properly reading my OP. As I have painstakingly stated, the definition of trend is context-specific. It should be obvious that the context of a stock market is very different from the context of coin flips, notwithstanding the assumption by some (not me!) that stock prices move randomly in some fashion. But even the random walkers don't think stock prices move like accumulated coin flips. They assume another random mechanism is at work.
"stock prices don't move at all." Yeah you and I live on different planets. Edit: if what you meant was price doesn't move itself, well Duh! I never said price moved itself. A trend is a pattern. The question is whether or not the pattern is exploitable. Sometimes it is and sometimes it isn't. It boils down to how trend is defined in a given context and to the frequency of exploitable trends versus the frequency of those trends that will cost you if you attempt to exploit them.
if we talking about the trend that formed in changes of prices over time, then any definition of the trend should involve time period which is defined in abscissa (5min, 1hour, 1day) of the chart So the definition should start like that: The trend of prices in the current time period....
Every price series, volume series or other financial time series is based on a specified timeframe. I am thinking about definitions of trend that go beyond the fact that there are specific intervals of time between the data. For example, it doesn't matter whether you're flipping a coin once a second or flipping once a day, my definition of "EITHER 3 or more heads in a row OR 3 or more tails in a row" applies universally to all timeframes. As Visaria pointed out, whether or not that is an acceptable definition is a separate consideration. Maybe the best context for considering specific timeframes is if trend is being defined over multiple timeframes. Now that would be both challenging and highly informative, if multiple-timeframe considerations lead to a breakthrough in trend analysis.
Even Ed Seykota admits trends are meaningless. It super funny to me that when guys who combined have made billions from the market ( verified) lay out strong arguments that trends don't exist or are meaningless like niederhoffer ( who proves it statistically) and seykota who proves it metaphorically--- AND no name, no credentials, no nothing ass clowns on the internet think they understand the market better----- seriously, think about what you believe and who you listen to. surf http://www.seykota.com/tribe/tsp/trends/index.htm There is no way to determine the current trend, or even define whatcurrent trend might mean; we can only determine historical trends. The only way to measure a now-trend (one entirely in the moment of now) would be to take two points, both in the now and compute their difference. Motion, velocity and trend do not exist in the now. They do not appear in snapshots. Trend does not exist in the now and the phrase, "the trend" has no inherent meaning. When we speak of trends, we are speaking, necessarily, from some or another view of history. ED Seykota
Seriously, surf, this should be embarrassing for you. Nothing is meaningless unless you define it as such. Nobody appointed Seykota as final arbiter of what trend is. He has his definition, I have mine. Moreover, he is confused about his. He gives a general ambiguous definition, then goes on to give an analysis that can only be correct if he is defining trend as the derivative of price. Huge clue for both you and Seykota: Price Is Not Differentiable. Too funny. Please, stop embarrassing yourself.