Malvolio, take the example of a fair, six sided die. On the next roll, we have:
1/6 chance of being a 1
1/6 chance of being a 2
1/6 chance of being a 3
1/6 chance of being a 4
1/6 chance of being a 5
1/6 chance of being a 6
Now we all agree, we don't know what the next roll of the die will be (because we can't predict), but I hope everyone on this forum can work out what the chance of the next roll being less than 3 (aka a 1 or a 2).
All these examples of roulette wheels, dice etc are missing the fact that those analogies don't apply to the markets, because the markets constantly evolve. Yesterday's dice had numbers 1-6, but today's may have 2-7, tomorrow's 3-8. Maybe it'll go back to 1-6 the day after, but maybe not. And two years from now, who knows what dice will be used?
That's why even purely mechanical ATS's have to be tweaked constantly, and eventually (often abruptly) replaced.
To get an accurate idea of a method's probability, you'd have to know both its probability based on the sample you have, and the probability of that sample accurately representing the next 1,000 or 10,000 or 100,000 trades. Your estimate of that probability is in turn based on a more limited sample of how abruptly and frequently the market changes - and so on. And all of these probabilities will overestimate stability.
The following 2 users say Thank You to futuretrader for this post:
It's obvious that the market isn't fixed odds like a roulette wheel, but it doesn't mean the analogies aren't useful for explaining things, nor does it mean that statistics aren't useful to traders. There'd be a lot less people employed in statistics if we only used it for fixed odds propositions!
Yes, thatīs clear, but when you are trading you cannot know what other traders will do. So for me trading has to deal more with emotions of the market participants than with mathematical probability. Your are not playing lottery. Trading is people making trading decisions. You cannot calcluate the probability of the next trade. How you could? How you would calculate the probability that a support will hold or a trend will continue?
Of course you know what the expectancy of a series of trades is or will be. But every time a setup you have can fail.
You can't predict the outcome of the die roll will be a 3-6, but you can say the probability of the roll being a 3-6 is 2/3. That's the point.
Trading is people making decisions. People make the same decision when presented with the same question time and time again, that is how you predict what they will do. If you're saying that the market doesn't repeat, then fine, but you ought to throw away 90% of your T.A. books, and I'm really intrigued as to how you make money trading knowing that the market doesn't repeat.
Calculating the probability of a trade being profitable and knowing the future are two different things, I don't see how that's a difficult concept to comprehend.
There IS a chance that any trade will fail, hell, there IS s a chance that all future trades will fail, that doesn't make the chance 50/50.
Please explain how the expectancy of a series of trades (with wins the same size of losses), can be anything but 0, with the chance of every one of those trades being a winner being 50%, after you've posted that here, go collect your noble prize and retire.
You can assume it is the same as the histoical expecancy. That does not say that you know the outcome.
if you have a 2% edge, i.e. I expect 52% of the trades to win, then I also expect the next trade to win with 52% probability OR the edge to deteriorate (or become better).
That still is not knowing whether I win. The argument is mathematically flawed BUT it serves a good point (i.e. make the trade "insecure" about the next outcome). There are strats (Scalping) with a very high win ratio, and your monkey so to say tries to make that a certainity mentally - which is BAD. From a psychological point it is likely advantegous to ASSUME the next trade has a 50% probability.
I really have to try out those coin flip strats. And try to run them through a backtest & optimizer
You're claiming that individually, all your trades had a 50% win rate, but over time you expect to have more than 50% of them win, do you not see how absurd that is?
2 outcomes != 50% chance
Assume that the win rate stays the same, independance etc. The last 100 trades have a 60% win rate, you expect the next 100 trades to have a 60% win rate. Correct?
Then each of those 100 trades individually must have a 60% chance of winning. If they each had a 50% chance of winning then half of them would of won, half of them would of lost, and your win rate over those 100 would be 50%.
A well designed mechanical AT that takes the right factors into consideration does not have to be tweaked constantly nor replaced abruptly. Market conditions do change but a good AT will stay out of the market when conditions are not favorable to the setups of the AT. Meaning equity curve will stay flat until conditions come back that are favorable. Nothing wrong with flat for awhile just don't want to lose big while waiting for right conditions to return. This can be designed into a system.
As mentioned above while market conditions do change in most cases they return and have cyclical nature about them to a large degree. Key is not to design a system that only works well when market is trading in some outlying fringe condition. But design for the way a market typically trades. That way can have reasonable assurance of system performance over time. The market typically will come back to it's normal ways when it does trade fringe. Of course it is possible a market will never come back to the "way it traded" but is not usually the norm in my experience although I have seen it happen.
I also think the presenter used his examples more to demonstrate concepts than present perfect math. We all know the market has a dynamic element to it.
I enjoyed this webinar can't talk enough about managing risk. I may manage different but key is manage it and always be open to how others do it as can make us all better at what we do.
"The day I became a winning trader was the day it became boring. Daily losses no longer bother me and daily wins no longer excited me. Took years of pain and busting a few accounts before finally got my mind right. I survived the darkness within and now just chillax and let my black box do the work."
Last edited by liquidcci; March 23rd, 2012 at 11:13 AM.
That is the crux here. I think the issue is mostly that 95% of the peopel can not deal with statistical math, and for the other 5% the exampe is disturbingly wrong bvecause they focus on teh math.
Let's get into the other side - psyhcology. If I ahva starat wih a 80% winning chance per trade historical (some funny scalping) then assuming per math the next trade has a 80% win expectancy... after a number of wins (LIKELY) the 2 losses in a row or 3 (unlikely, but will come over a larger sample) will get my monkey playing - because basically I cam not used to this. 80% is good enough ppeople never get comfortable with the loss.
Taking the "I expect the net trade tio loose with 50% chance" attitude helps to overcome this and points out the psychological issue.
Still for the math inclined this is arguing along "1+1=2.2" and as such it is alienating our ingrained thinking.