Territory of probability - page 7

[St.Vitaliy]

Stasikusssss:

I'm not a professor, I don't know how to say it scientifically.

But to calculate the probability you need to know the number of possible outcomes of the event.

For example, probability of tails = 1/2 = 50%, probability of six on a die = 1/6 = 16.67%.

But the price series, the results of TC - this is a completely different area, how to calculate the probability there scientifically I do not know, and even if it is possible?

Statistics can be applied to the results of TC, but there will not be any probability, variance, expectations.

Suppose I performed 100 test trades.

I have made 60 losing trades of which the average loss is 70 pips.

40 of them are profitable, the average profit is 150 pips.

Why can't I predict whether it makes financial sense to continue trading?

Clearly, each new trade will make a correction to the indicators. Then I can count the deviation of the performance of the last 25 trades from the underlying history of 100 trades. And as long as these deviations are within sigma, I can leave the parameters of the TS untouched.

[Dmitriy Skub]

notused:

Now, dimensionality changes quite often and they try to apply the same methods to all states, which leads to wrong conclusions and losses.

+100500

[St.Vitaliy]

By the way, I'll open up a piece of the grail.

In all the books, articles and forums I write about testing with a large number of trades, then forward testing and also hundreds, thousands of trades. There is a deep sense in this.

But I noticed that mathematical expectation (I consider it in relative units (stops) but not in points) of stable systems do not change considerably after 20-25 trades. That is, if during a test period we have conducted 500 trades and on the average everything seems to be OK, but breaking the series into 5 parts and comparing them with each other you see that they are very different, then TS in general is not stable and some price artifact pulls it. In this case you need to add a filter and cut off unnecessary deals trying to determine this artifact, but if it is not periodic, then completely abandon the TS...

[qarabala Nagi]

teoriya veroyatnosti eto absurd. ya kak matematik ne rekomenduyu

[Dmitriy Skub]

qarabala:

teoriya veroyatnosti eto absurd. ya kak matematik ne rekomenduyu

What do you recommend?

[St.Vitaliy]

They once wrote a super-duper chess program. Put him up against a man. He went E2-E4, the programme analysed all the variants and gave up, because if the man plays correctly, he will win 100%. So here too, you can't apply it head-on. Steel it does not sink either, but the biggest ships are made of it, not wood.

[Sceptic Philozoff]

St.Vitaliy: They once wrote a super-duper chess programme. I put it against a man. He goes E2-E4, the programme analyses all variants and gives up, because if the game is played correctly the man will win 100%.

This is probably just an anecdote.

There is a programme that is already beating former world champions quite successfully. Irregularly, but it beats them. But it beats grandmasters.

Steel does not sink either, but the biggest ships are made of it, not wood.

You must be an expert on nonsense.

qarabala: teoriya veroyatnosti eto absurd. ya kakematik ne rekomenduyu

Looks like there's been a full moon recently...

[Anatoli Kazharski]

Mathemat:

...

Looks like there's been a full moon recently...

There was a new moon recently. And the full moon is coming in about a week or so. ))

[St.Vitaliy]

Mathemat:

It's probably just an anecdote.

There is a programme that is already beating former world champions quite successfully. Not regularly, but it beats them. But it beats grandmasters.

You must be an expert on nonsense.

Looks like there's been a full moon recently...

You must be very fixated on some narrow sector. An example is a historical fact held at a university in the 16th century.

[Ratibor Shokhirev]

The algorithm for winning a coin toss game is simple: if it's tails, bet on tails; if it's heads, bet on heads. If the number of flips is infinite, you win.)