Lessons taught by a Coin
"Your idea has to be original only in its adaptation to the problem you are working on"-Edison.
In the last chapter of Security Analysis (1934), Ben Graham mentioned about an experiment done by Frederick Macaulay to prove the invalidity of technical analysis.
"...Frederick Macaulay plotted the results of tossing the coins several thousand times (heads = one point up; tail = one point down) & had thereby obtained a graph resembling in all respects the typical stock chart--with resistance points, trend lines, double tops, areas of accumulation, etc. Since the chart could not possibily hold any clue as to the future sequence of heads or tails, there was a rather strong inference that stock charts are valueless." (Pg no: 608)
I got fascinated & wanted to try it out myself. The plan was to plot a chart with 100 coin tosses & see how it looks like. But then I looked at the chart & wanted to check if regression to mean holds. In the process I ended up with 400 coin tosses & few important lessons.
Invalidity of Technical Analysis:
Everybody will agree that it would be fallacious to look at the above chart and conclude that there will be a more head or tails in the next 50 tosses.
One argument that, I think, people who believe in technical analysis can use against the logic here is 'Each coin toss is an independent event, whereas daily stock price moments may not be independent events. How do you rationalise it?'
I don't know the answer but it reminds me of wonderful quote by Ben Franklin,
"There's none deceived but that he trusts."
There were many instances where I had sequence of 5 heads or tails in a row. I could clearly hear in the back of my mind, while the coin went up in the air, that this time it should be head or tail based on the past vivid sequence.
All this even after being aware of the fact that each coin toss is an independent event & it has got nothing to do with the past events. This is also one of the reason that I continued till 400 (ie) I did it until I stopped predicting, subconsciously. This reminds me of what Tversky & Kahneman have said,
"Under some conditions, when the conditions are knowable, we should not trust our intuition because we are liable to predictable errors & biases. Also the fact that we can predict in advance, when intuition might falter, gives us some hope of implementing procedures to avoid the errors."