The riddle: the distribution bell rattles - the broker says the price, whoever it hits sheds tears and loses their deposit. - page 8

 
"Checking the equal probability of reaching symmetric levels and the futility of tails.
To test on historical data, we will write a simple robot that will open trades when the "channel" is broken: the price range. The robot will close trades when one of the symmetrical levels is reached - the opposite border of the channel or the line equidistant from it from above.

If the probability of reaching levels is not 50/50, we will get some definite financial result on a palpable set of deals - either positive or negative. In particular, if the "tails" mentioned above do not really fit the normal distribution, then opening on a breakdown of the channel outwards will produce a steady profit.

For tests with 1 point commission we have an average profit factor (Profit factor) of 1.008. Tests in conditions of an ideal broker give us a profit factor of 0.992. That is, the influence of the probability distribution deviation from the normal distribution is so small that there is no profit to be made only on the knowledge of its existence".
This is an excerpt from that article.
 
Martin Cheguevara:
This is a good idea I was happy too when I did this...... except for one thing... It will work with commission of less than 0.01%... it also depends on timeframe. If you do not know what kind of deal you want to make, you may try to do it on different timeframes.

In general, any lengthy event with many participants is almost guaranteed to produce many different statistics. Some of them can actually be used.

This particular statistic (see chart) is designed for intraday trading, mainly uses TF 1m, and provides several trades per day. To calculate losses from spread, slippage etc. 3 points are subtracted from each profit or loss trade in the model.

I do not know if this statistic is available for other timeframes. I guess it is unlikely.

 
Novaja:
"Checking for equal probability of reaching symmetric levels and futility of tails.

.....

This is an extract from that article.

It just says that you can only work with asymmetric or biased (which is basically the same thing) relative to the start of the trade statistics.

 
Yuriy Asaulenko:

This only suggests that you can only work with asymmetric or biased (which is essentially the same thing) relative to the start of the trade.

This is not enough, this is the principle of using "tails", while the profit in this case depends on the spread, rigidly, there is no possibility to earn by this system on forex. On the stock market the "tails" will be heavier. The problem lies at the top.
 
Actually, the idea of increment is not clear at all - why shouldn't there be chaos? Wouldn't it be easier to imagine that there are goals the price moves to and new goals are set after these goals are reached? Then we have a time interval with permissible ranges of exit from it, and a start point and an end point; naturally, distribution of price delta from time interval to another interval will be different, and quite probably random. And now draw a line from the movement start point to the target point and imagine that the target is not the movement itself but its direction to the target point; then we will have non-uniformly distributed deviations from the central line and this will prove that the movement is not random. Another question is that we can only observe such points in history.