I think there is a definite distinction in my view. The possible outcomes of a single trade is binary, namely win or lose, therefore 50/50. That does not mean the probability of that trade being a winner is 50/50. Probability is based on observed history and cannot be defined without historical sample, so if you have observed that specific trade winning 800 out of the last 1000 occurrences, the next time that trade shows up, it's possible outcomes are still 50/50, but it's probability of winning based on historical evidence is 80%.
The following 3 users say Thank You to monpere for this post:
This is more than highly arguable, it's downright false. The chance of the market moving in the direction of the trade is, in fact 50/50 - as the market can only go with or against the trade (assuming that staying still is not counted as an option - which it is sometimes).
But that is not what a real trade is. A trade is a bet on a set of parameters, each of which has its own varying probability.
So the true chances of a real trade being successful are significantly smaller than 50/50 if you account for the actual parameters necessary for a real-world trade.
If we could try to define it, we would have to say, at minimum, something like: "a trade is a bet that the market will go a certain direction, from a certain point, without going backwards beyond a certain amount, and go far enough in the right direction to be close-able for a reward."
By comparison, when stocks are chosen at random, then compared to fund managers' performance, there is generally only one parameter (performance v. the S&P).
When they created the "Monkeydex" (where a monkey picked some stocks & they outperformed the S&P), all the monkey had to do was pick some names - it didn't even have to pick a direction (long or short). Thus, the number of parameters was extremely limited, and the chances of success were automatically much higher. The monkey didn't have to set a risk-level it could not afford to go beyond, decide where to close the trade/leave it open, or anything else. Some of the monkey's picks could have lost 90%, then come back at the end of the year, and been counted as winners in the final totals - because risk-limitation was not a parameter.
So what is the actual probability of success for any given trade?
Virtually impossible to calculate, but that doesn't prevent us from making a basic rough-out of what the average probability might look like:
Let's assume, for the sake of simplicity, that all variables have 50% probability (which, of course, they don't).
What are the necessary variables? (as in cannot be eliminated for real-world trading).
Well, what are the things you're betting on?
1. Trade direction.
2. That it will go that direction without retracing back past a certain amount (your stop).
3. That not only will it go your direction, but go far enough to cause you to close it (reach some kind of satisfactory target). The trade can go your direction one tick, and you have satisfied variable #1 - but that does not make a successful trade - you are betting that it will go a certain minimum distance - or you wouldn't have taken the trade. This is significantly different from just looking back at the Monkeydex at the end of the year, & saying "yep, stocks are up."
So a rough-&-dirty calculation might look like this:
Probability #1 (direction): 50%
Probability #2 (stop): 50%
Probability #3 (distance): 50%
So that gives us, best-case scenario odds (based on random chance alone) of 12.5% (50% x 50% x 50%).
Of course, the probability of success could be significantly less than 50% for variables #2&3 - depending on how you handle them (too tight stops, unrealistic targets, etc.). And your chances of getting the direction right could, of course, be much better or worse than 50%.
But you get my point, right? If a trader is batting .500, in all reality, that's measurably better than just raw chance alone - if you account for all the necessary parameters of an actual trade.
Last edited by ddouglas; March 7th, 2012 at 08:25 PM.
The following user says Thank You to ddouglas for this post:
I don't want to rain on your party, but you understand what I mean, right? I hear "50%" all the time from all kind of different sources.
Truth is, the minute you add a qualifier to that 50%, it is no longer 50%.
If you open a bet at random on the open every morning in the S&P, & bet "long" or "short," you DO have a 50% chance of winning.
Will you hold it if it goes 50 points against you? What if it goes 50pts against you, then reverses and becomes a 20pt win? (stranger things have happened). That still counts as a win for probability purposes - and if you said "long, but no more than 45pts against me," you would have turned it into a loss.
Is an example this extreme totally useless? Of course not, everything else is just a smaller degree. I'm just exaggerating to illustrate a point. But if you chose "no further than 4.5 points against me" you are still potentially turning some winners into losers, and therefore reducing your chances below 50% - because some winners will retrace further than that before becoming winners.
50% chances on the direction of the market are only 50% if they are completely unqualified, and not limited by other parameters. Everything else is some degree smaller. (or a large degree smaller - depending on what it is).
Fair enough DD....what you have stated is a purely objective, empirical and mathematical point of view. Absolutely correct. And it subscribes to the random walk theory.
I doubt that view, markets are not random - what about huamn nature and discretionary ability. I do not take random trades, discretionary traders can't afford to. Purely objective traders HAVE to. As do systems. You point is correct for systematic trading.
When you buy weakness in strong market, the probability of that being a good trade is not the same as that of buying weakness in a weak market. And the other way around. These formulas have a place, the big picture context is everything to me. In the evolution of a trader, a newbie must treat every trade as if it will fail. An experienced trader will not.
Intuition, experience, information, risk tolerance, ability, tools....must skew this factor of chance....Trades taken with the structure of the market will do so too....key point is every trade cannot be viewed as such.....a machine acting on simple inputs is still influenced by the designers abilities and beliefs. If one believes that markets are an extension of human nature (and base one's trading on it)...you must re-examine the purely objective view that price action is truly random. I do not base my trading on random walk.
The following user says Thank You to Deucalion for this post:
I didn't say anything about random walk - nor do I subscribe to it.
I said that the idea of 50% probability is wildly flawed, and does not account for the other parameters that are imposed on real-world trading.
If you have an intelligent market bias, you can have a MUCH higher chance of getting market direction right. This has been proven over & over. But it is only one of the parameters that you need in order to be successful. You can still lose on the other two variables.
William Eckhart advised to try to not have more than three or four parameters to your system - because he believes that it makes your probability of succeeding too low.
Are you trying to suggest about trade execution probability without market context? I cannot even conceive of such a thing...I am not trying to be pedantic or an ass about it....but how does one reduce risk? The tools to reducing risk and making good decisions in anything increase the chance of a successful action...this would be true in anything. The probability of driving without ending in the ditch on ice for a driver with advance training is better than that of a noob....This, to me is very black and white...why does it not apply in trading? What am I missing here?
If price action is not random, then a trade taken in context with that view must have a greater chance of succeding than one without. Simply put.......
Please register on futures.io to view futures trading content such as post attachment(s), image(s), and screenshot(s).
Yes I know, I am being far too simplistic...but you understand the point I am trying to make....with discretion, tools, knowledge etc etc....you can skew that in your favour (not all the time but enough combined with positive expectancy). That is all. Lack of random Walk must be taken into account when taking a trade....I view it as integral to my trading plan and to my trades.
Yes, I understand that even with that intelligent bias, the probablity being 50% or 40% or 12.5% is a debatable point. And that was your point - the percent number. I agree, I have no clue on the actual percent number, I would much rather pay attention to bias and structure, trying to increase the chance and the gain from each successful chance (especially the 2nd point)...which is to lever up on what I call very high probablity trades. The last point is new to me over the past few months...but its net effect is crucial on the bottom line.
Last edited by Deucalion; March 7th, 2012 at 09:39 PM.
The following user says Thank You to Deucalion for this post:
Apples and Oranges. I love it when discretionary traders debate probability. The very fact that your enter, manage, and exit any trade with any discretion, makes it impossible for you to determine probability of anything. Why? because every discretionary decision introduces a different variable that may or may not have been represented in the historical sample, therefore negates the entire sample. You based your historical sampling on apples, and oranges, and in real life, you are trading apples on one trade, oranges the next, and apricots the following trade.
Unless you are mechanical trader, the only way you can legitimately rely on probability is if you record and analyze a large sample of your actual trades, and calculate the data based on the sample of actual trades. Even then, you will not be measuring the technical probability of a certain type of trade, you will be measuring how effectively you manage those trades when you took them overall. The next time you take one of these trades, the outcome will be less affected by the winning probability of that type of trade, but more by which discretionary variable(s) you introduce in the execution and management of that singular trade, and the same holds true for each trade you take after that.
The following user says Thank You to monpere for this post: