Coin flipping has no memory.
Trading Lessons/Insights From Coin Flipping
- Thread starter tradingjournals
- Start date
Coin flipping has no memory.
That's his argument. He says the market has memory and he can model the next price move based on the prior price move minus any 'random news'...
I cannot wait to see the trolls and flaming I get when I start talking about a synthetic market price simulator--ha!!! Should have it complete in the next few weeks.
Masterjaz
You cannot predict the stochastic component. The answer is to trade when that component has minimum influence so that your system can maintain his edge.
Just make sure you are out of the market during important schedulled news events. There are also random news events you do not know when they come. Some will move prices to your favor, other will act like black swans. There is where position size and money management play an important role. But as you said, from the law of large numbers, the long term expected gain from all such events is zero. Just do not get cought in a black swan with a large position size beyond what you can manage.
Thanks for adjusting my understanding of his post. I thought he was discussing it as coin flip + news, as opposed to what you explained.
Try not worry about the trolls, just keep focused on the topic, and know that there are people enjoying what you write, including Yours Truly!
Your point about isolating the news is a good point.
I have a different view on "after the news is released" in the case of the overall market: The market takes time to incorporate the news, and it can be a way street for the price direction.
Quote from nLepwa:
Or you can trade the noise.
Ninna
You mentioned a number of times the Arcsine law. Could you discuss it a bit as it relates to coin flipping? I read a piece about it in relation to coin flipping, and the author of that piece wrote some correct, but also some incorrect statements.
All: in a next post, I am going to "introduce/point out" some elements of the coin flipping that may lead to different ways of how to play it.
TradingJournals,
Here is my thoughts on the arcsine law. I'm sure Ninna will provide a better explaination, but me putting into words helps improve my own understanding.
The arcsine law describes the likely distribution of results especially related to coin-flipping like situations where, intuitively, we would assume fluctuation about the unchaged, zero, line. So, you flip a coin x numbers of times and you'll get approximately the same number of heads and tails with a few runs of each. Net-net, it should be close to zero. Like this simulation:
But the fact remains that many simulations produce curves that are dramatically above or below the zero line such as these:
The reason is that the long runs of heads or tails have a lower probablity of happening, that probabilty is non-zero, and when they do occur they move the curve signficantly. The result is an equally unlikely move in the opposite direction must occur to get back to zero. So once a run occurs, whether it be good or bad, the curve is likely stuck on that side of the zero-line, and as time goes on, it will continue to remain on that side of the line. That is not to say that it cannot return to zero, just the probablity is diminished.
The main reason I enjoy the coin-flip discussion is not for forcasting market movements, but as a mechanism for evalutating my equity curve. That is to say, how can I determine if my system is truely profitable based on the equity curve, when, in reality, it may be just breakeven and I had a good run, ala the curves above???
Something to think about...
Right, wrong, close, Ninna can fill in the gaps.
Masterjaz
The arcsine law describes the fraction of time spent in positive (or negative) territory. Even just two deviations on the win side can lead to the curve getting more likely stuck to the win side than to the lose side.
It would be worth thinking about the paths that lead to a given win amount W, during N flips. More paths would come from past winnings, than from past losings of equal amount, even if the latter two have the same probability of taking place! This may not be intuitive.
Right.
Let X be the number of times the RW crosses the origin. (i.e. the number of times the RW switches from positive to negative).
The Arcsine law states that the most probable value for X is X=0. The second most probable value is X=1, then X=2 and so on.
The bigger the imbalance (bias) between head and tail, the more probable is X=0.
For instance when the probabilities of head and tails are 100% and 0% respectively then X=0 has probability 1.
Now how can it be exploited to trade?
Ninna
Let X be the number of times the RW crosses the origin. (i.e. the number of times the RW switches from positive to negative).
The Arcsine law states that the most probable value for X is X=0. The second most probable value is X=1, then X=2 and so on.
The bigger the imbalance (bias) between head and tail, the more probable is X=0.
For instance when the probabilities of head and tails are 100% and 0% respectively then X=0 has probability 1.
Now how can it be exploited to trade?
Ninna
You first have to understand what the arc sine law says. I do no think you understand it, I am sorry to say that. I hope you are flexible enough to admit you do not understand it. If you do not admit that, you will only hurt yourself, not me.
What is the arc sine law? The arc sine law states that the probability distribution of the proportion of the time that a Wiener process is positive is a random variable with probability distribution the arcsine distribution.
What does it mean for trading? It means that given the assumptions of the arc sine law, over time a trading system will have long periods of gains and long periods of losses. Gains and losses will not be each assigned half of the time.
What is your mistake? Your mistake is that the naive arc sine law application perpetuated in the internet by various bozos and other clowns assumes that a system trades either long or short. If that is true, then the arc sine law applies and precicts that the system will have long periods of gains and long period of losses.
Hey stupid clonws! I do not need an arc sine law to tell me that. Screw you all idiots. (Sorry, I get upset when I realize how many stupid people are around). Of course markets go up longer periods and go down long periods.
But if I have a system that trades both long and short, the arc sine law does not apply in general. More importantly, if I change and upgrade my systems often to account for new market conditions, the assumptions that apply the arc sine law go down the drain. Proof
Suppose I have two systems A and B and one plays the market long with timing and the other plays the market short with timing. According to arc sine law, A will have long periods of gains and B long periods of losses.
Next I combine A and B to a new system C. Now, my system C has the net profits that depend on system profit factor.
Thus, the arc sine law does not applly. QED.
In essence, the key is the timing of the long/short trades which is not taken into account by the arc sine law. Since this is the purpose of trading systems, the arc sine law as far as long/short position/intraday systems does not apply. It may apply to longer term trend following systems but who cares about that anyway.
What is the arc sine law? The arc sine law states that the probability distribution of the proportion of the time that a Wiener process is positive is a random variable with probability distribution the arcsine distribution.
What does it mean for trading? It means that given the assumptions of the arc sine law, over time a trading system will have long periods of gains and long periods of losses. Gains and losses will not be each assigned half of the time.
What is your mistake? Your mistake is that the naive arc sine law application perpetuated in the internet by various bozos and other clowns assumes that a system trades either long or short. If that is true, then the arc sine law applies and precicts that the system will have long periods of gains and long period of losses.
Hey stupid clonws! I do not need an arc sine law to tell me that. Screw you all idiots. (Sorry, I get upset when I realize how many stupid people are around). Of course markets go up longer periods and go down long periods.
But if I have a system that trades both long and short, the arc sine law does not apply in general. More importantly, if I change and upgrade my systems often to account for new market conditions, the assumptions that apply the arc sine law go down the drain. Proof
Suppose I have two systems A and B and one plays the market long with timing and the other plays the market short with timing. According to arc sine law, A will have long periods of gains and B long periods of losses.
Next I combine A and B to a new system C. Now, my system C has the net profits that depend on system profit factor.
Thus, the arc sine law does not applly. QED.
In essence, the key is the timing of the long/short trades which is not taken into account by the arc sine law. Since this is the purpose of trading systems, the arc sine law as far as long/short position/intraday systems does not apply. It may apply to longer term trend following systems but who cares about that anyway.