a toin coss will give a binomial distribution , yes ?
but the market isnt like that - it can stand still as well as go up or down -
just trade options and you will now what I mean as time value erodes .
unless you can get the coin to land on its edge ?
I have found a random number generator very useful for testing different systems.
but the market isnt random either.
I think that the fact that TA works on randomly generated charts means that it doesn't matter that much if the markets are random or not. Having looked at a lot of randomly generated charts, there is some difference between them and charts of the major markets. This makes me believe that there is an element of randomness in the markets but there are also some non-random elements that make the charts look different.
It would be good to see a randomly generated candlestick chart. My spreadsheet skills aren't up to that task.
Is this a case where the Kelly formula applies? Because you have no edge in tossing a fair coin, at any finite time you could be plus or minus a fortune, even though the expectation - for infinite series of tosses - is zero (and commissions likewise). In case anyone is interested, I attach a pdf of the original Kelly paper on information and games of chance, which I found somewhere on the web.
Why are HLC not independent of the open? Are you suggesting that during the day the price is constantly referring to the openening price. I appreciate that in terms of up or down it is but it is not controlled by it.
To take extreme cases what happened on 9/11, black monday etc did the price refer to the opening before dropping - I think not.. Surely OHLC are just part of a series related to one another but independent of one another
A more realistic way of creating a random stock price pattern might be to imagine an x sided die, say one with 21 sides, and then throw it 10000 times. The facets of the die would be numbered +1 to +10 and -1 to -10 and zero (to represent unchanged).
Then start your chart with an arbitrary stock price, say 100, and start throwing, adding each result to the running total. So you might get 100, 101, 100, 103, 102, 99, 99, 101, 102,104, 107, 105, 115, 111 etc. Then plot this on your chart as just closes. After a while it may start to look like a typical daily stock chart. I don't know how to incorporate the OHL bit though!
However stock prices do not generally move up or down by 10 as often as they move up or down by a lower number, so you might want to load the die so that the higher the number, the progressively less chance there is of it coming up. I don't know whether, for realism, this progression of chance should be logarithmic or linear. Try both perhaps. Either way the outcome then won't be truly random because it is weighted towards the lower numbers, more like a stock price movement.
We seem to have got locked into the thought that the markets are random surely they are not.
They react according to stimuli - not always in the manner expected but they do react. If they were truly random they would be pinging around all over the place (try a random number generator), instead they follow some sort of pattern which is what we try to latch on to. If they were truly random Fibonacci and all the rest of them would be useless as they use patterns to predict.