The Flip-A-Coin trading strategy and lessons learned from it

[abogdan]

Quote from NET:


My argument is that it is measurable and predictable--predictable in the sense that the trend continues until a measuring device (trend line, for example) shows the masses have changed behavior. Then that new behavior persists until yet again another measurable change manifests itself.

Abogdan, Do you disagree with this thinking? [/B]

I don't want to sound as I have all the answers, and I respect your point of you after reading yours and boundle's discussions, but, if I may, there is a problem with your statement.
I'm sure you are familiar with the Gödel's Incompleteness Theorem. That is why the question of what comes first: masses changed their behavior and influenced the price or price have changed and influenced masses behavior is not a correct one (Chicken and Egg). You can not rely any measuring tool that only uses information that is used to construct the tool. I'm not sure I was clear, but I hope you'll understand.
My point is quite simple: Price moves are the result of human behavior (Underlying Sentiment), but you can not use price moves to analyze the behavior. You need something outside of the closed loop system to construct meaningful analisys.

 

[NET]

Quote from abogdan:

NET wrote:

“Isn't this what happens when stock prices are advancing? Human behavior, driven by the greed emotion (this is not a negative by the way) will have many jumping on board up until the value does not seem so good. Resistance is established, and profit taking sets in.”

Bingo! This I support 100%. It was my point! Price chart patterns do not exist!

Hmmm.... We see this point differently. However, I think I understand what your trying to say. Perhaps there's confusion about what we mean by a pattern.

 

[abogdan]

Quote from OldTrader:

"Sentiment Analysis" as you put it is exactly what the chart reader is attempting to figure out from the chart. As the actions of the various traders filters into the chart itself, the chart reader should be able to detect the shifts in sentiment.

OldTrader

With all do respect, I appreciate your point of view, but isn't it too late to react on something that is already happened? Chart patterns are only derivatives of people's actions not other way around.

 

[bundlemaker]

Quote from abogdan:

With all do respect, I appreciate your point of view, but isn't it too late to react on something that is already happened? Chart patterns are only derivatives of people's actions not other way around.

My take on this is it is not too late. Markets don't turn on a dime. I have discovered two very simple chart patterns that do a superb job of divining market particapants expectations (ie. their pychology).

Abogdan, I am most sincere when I express my interest in your ideas. But, so far you haven't shared any, you're just doing mental masturbation. With each passing post, it's easy for onlookers to suspect you have ulterior (commercial) motives.

You said previously you were willing to share. I invite you to do that right now. Lay it on the line, tell us your hypothosis and how you apply it to trading.

Best Regards,

 

[ig0r]

Quote from abogdan:

I don't want to sound as I have all the answers, and I respect your point of you after reading yours and boundle's discussions, but, if I may, there is a problem with your statement.
I'm sure you are familiar with the Gödel's Incompleteness Theorem. That is why the question of what comes first: masses changed their behavior and influenced the price or price have changed and influenced masses behavior is not a correct one (Chicken and Egg). You can not rely any measuring tool that only uses information that is used to construct the tool. I'm not sure I was clear, but I hope you'll understand.
My point is quite simple: Price moves are the result of human behavior (Underlying Sentiment), but you can not use price moves to analyze the behavior. You need something outside of the closed loop system to construct meaningful analisys.

Makes sense. So you're saying that large shifts in psychology (which occur at market tops and bottoms) don't always reveal themselves on the chart (at least not at that moment), but can still be measured? Only way I can see this working is maybe taking a large pool of people trading the market (1000 to 10000) and constantly polling their sentiment. This would work as long as they are all honest Oh, in addition, you can't let them see the results of the poll until everyone has voted, so their sentiment is not influenced by those of others.

 

[NET]

Quote from OldTrader:

"Sentiment Analysis" as you put it is exactly what the chart reader is attempting to figure out from the chart. As the actions of the various traders filters into the chart itself, the chart reader should be able to detect the shifts in sentiment.

Quote from abogdan:

With all do respect, I appreciate your point of view, but isn't it too late to react on something that is already happened? Chart patterns are only derivatives of people's actions not other way around.

I think OldTrader has it exactly right, and so do you in that chart patterns are derivatives. But I disagree that it is too late to react, because human beings are creatures of habit, that habit repeats, predictably.

If that is not the case, then entire industries will have to change, from evaluating one's credit score to assessing insurance premiums, even the civil and criminal justice systems. All of this is based on a set of rules that were derived from past human behavior in order to deal with future human behavior.

Understanding past human behavior through the visual representation of a chart--to gain some level of insight to future human behavior (i.e., an edge to increase the probability of an outcome to that of better than random) is not an exercise in futility.

Don't you agree?

 

[OldTrader]

Quote from abogdan:

With all do respect, I appreciate your point of view, but isn't it too late to react on something that is already happened? Chart patterns are only derivatives of people's actions not other way around.

It's always too late to act on what has already happened....but that is NOT to say that what has already happened may not have some type of predictive value to it.

Let's put it this way: let's assume you have made a study of various bottoms, and have arrived at a few different patterns that these bottoms typically form. Now, you see this pattern forming one more time. You take a long position based on the historical precedent for this type of pattern. If precedent works as in the past, then clearly it's NOT too late, as a large rally will result from a bottom.

Now we all know that no analysis is perfect...and therefore, we trade with stops. If historical precedent is not to work this particular time, then we will be stopped out.

By the way, I have made a fairly extensive analysis of various sentiment measures. These don't work all the time either...and from a trading perspective when used alone have serious problems in my opiniion from a risk control standpoint. This year for instance has seen many of the sentiment types of measure fail completely in the rally.

But I digress...still waiting for your "holy grail".

OldTrader

 

[abogdan]

Qoute from bundlemaker


"You said previously you were willing to share. I invite you to do that right now. Lay it on the line, tell us your hypothosis and how you apply it to trading."

Have you noticed that you cannot obtain any historical data (historical bar charts) on four most crucial parameters: Iside Ask Price, Ask Size, Bid Price, Bid Size? Never mind trying to find historical Market Depth (Level II) data sorted by market makers and ECNs! I have done it! I now have 3 years of complete data for all the NASDAQ stocks quote-by-quote, tick-by-tick, market maker-by market maker. Why am I telling you this? Because of the following:

1. By grouping different market participants by their reaction time to any price change I have constructed the Market Sentiment Spectrums. What are they? They are correlation functions of New sizes on either Bid Side or the Ask Side of the book followed by the price moves.
2. By measuring relative strength of different market groups activities (in terms of their sizes) I was able to much the short price moves ($0.10 – $0.20) with the correlation patterns that exist between the two consecutive spectrums.
3. Integration functions of AskSize/BidSize in real time preceded the price moves by at least 2 –3 seconds.
4. Calculation of the rate of new Asks vs. new Bids on the book sharply indicated the reversal points when the book stopped responding to one side or another.
5. Calculating amount of canceled orders vs. orders that stayed on the book indicated the possible upcoming move strength.
6. By measuring the first derivative of the size changes sorted by market makers I was able to find the “vacuum” points on the book that stopped further price development.
7. By calculating relative amount of new orders vs. different price moves it was possible to determine the “head room” of upcoming sentiment waves.
8. Buy analyzing Price move vs. Size move ratio I was able to estimate the strength of the Sentiment therefore approximate the upcoming price move.

Here it is as I promised! I hope you did not suspect the lag of my integrity, did you?

 

[bundlemaker]

Abogdan,

Thank you. Frankly, I only barely understand what you are describing; so the following is stated in the context of a sincere student:

1) I suspect that combining certain chart patterns with watching the bid/ask volume (as I do using esignal formula for this) produces at least a similar effect to what you describe.

2) A major difficulty with looking at such short term data (cover 2 to 3 seconds as you mention) is that a HUGE portion of that data is noise. I have seen strong evidence that suggests anything less than 30 min bars has significant noise and anything under 15 min is largely noise. My own experience has born this out also. The only trader I know for a fact making significant $$$ trades completely automated, and also trades 30 min bars.

Please comment.

 

[NET]

Quote from abogdan:

Here it is as I promised! I hope you did not suspect the lag of my integrity, did you?

Abogdan, I've enjoyed the debate!

The information you've provided is indeed quite interesting. However, what you are doing is no different than other forms of technical analysis.

Here's what I mean:

You've taken years of data and analyzed it to derive patterns and formulas for the purpose of developing some predictive value on future behavior. No matter how you cut it, or what you call it, you're using past human behavior to gain insight to future human behavior.

Your method is different, in that you're using patterns in bids, asks, sizes, etc., but you're still predicting future behavior from past behavior. This is the essence of technical analysis, which comes in many forms.

The way I see it, you're confirming that technical analysis has validity. Your really arguing about the mechanics or method one uses...