Brain-training tasks related to trading in one way or another. Theorist, game theory, etc. - page 4

 
Candid:

For the original (ideal) formulation of the problem this is so. But in reality (as many have written above) the key factors are the spread and the finiteness of the capital. In this sense, as a next step towards reality, it would be interesting to include a commission in the form of a fixed fraction of the rate. The question may be: how much does p differ from 0.5 to make the mathematical expectation remain positive at this commission?

Do you have in mind a random independent process with the probability other than 0.5? There are guys who trade without spread and commission and their capital is virtually unlimited. For such a process they will pay you more than the whole DC you trade in. I can introduce you, if there's a real reason...

But something tells me there isn't. Only the fantasies of Reshetov's triplets about the ever-positive expectation of money.

 

Wow what an interesting topic has been raised!!! I'd like to share my results of a game of equal odds.

I tried with a friend to test the game manually on equal odds: as in real life, if you use a normal random number generator in the form of flipping a coin (and it is exactly what we used for the tests), the probability of falling out eagle / river is distributed approximately equally, in the absence of number zerro on the roulette can well have an average win.

Based on the observation that from time to time red and black numbers on the roulette wheel form a series and that a series of two or more numbers of the same colour fall out more often than alternating colours, we have formulated the rules of the game:

- In the beginning we look at the number of what colour has fallen out (let for example has fallen out red)

- We put a minimum bet on red.

- If red has fallen out (we win), then again we put a minimum bet on red.

- If black has fallen (we have lost) we bet double on black.

- Whenever there is a colour change (when we lose), double bet on the colour that fell out last (that is, always bet on what fell out - we follow the colour change, wishing to catch the next series of one-color numbers).

So, armed with these rules, taking a coin and a piece of paper, we started testing. In all cases the deposit was growing by the end of the table. At the same time the losses were not so big. Maximum drawdown was eight losses in a row. With a minimum bet of 1 cent this means:

1-2-4-8-16-32-64-128

So you should always have 512 cents to spare.

Needless to say, all casinos have a limit on the maximum bet. I have personally played at an online casino that has no such limit. Minimum = 1 cent, but there is no limit on the maximum. There are not many such casinos on the internet, but they do exist. Unfortunately, the program knows on what colour I bet, and the casino owner is not a fool to do a completely random probability distributions. So I ran into another series of 13 losses, I left the casino and never went back. But a program is a program. If you sit two against each other and play with a coin, then 1 can beat the other - it is quite realistic.

Now look: in forex, price has inertia - it's no secret. It is the equivalent of a roulette series of numbers of the same colour. Moreover, there are no zerro type numbers in forex - this increases the chances of winning. You can try to concoct an EA and see how it trades. Terms are the same as for roulette. You may try to start with stop and take equal to two minlevels. For example, for longs this way:

MinLevel=MarketInfo(Symbol(),MODE_STOPLEVEL);

PR=Ask; PR=NormalizeDouble(PR,Digits);

SL=PR-(MinLevel+MinLevel)*Point; SL=NormalizeDouble(SL,Digits);

TP=PR+(MinLevel+MinLevel)*Point; TP=NormalizeDouble(TP,Digits);

 
You're spoiled :)
 
is this martin experts' PR?
 
drknn:

- If black has fallen (we lose), double the bet on black

- Whenever there is a colour change (when we lose), bet double on the colour that came out last (that is, always bet on what came out - we follow the colour change, wanting to catch the next series of one-colour numbers)


What if you don't double up?
 
sever30:

What if you don't double up?

And then how do you get out of a drawdown? Instead of doubling, you can multiply the lot of a losing order by some coefficient that is less than two, but still allows you to bring the deposit to breakeven at least. This will minimize the drawdowns, but still, with each subsequent loss the lot must be increased.
 
drknn:

And then how do you get out of a drawdown? Instead of doubling, you can multiply the lot of a losing order by some coefficient that is less than two, but still brings the deposit to breakeven at least. This will minimize drawdowns, but still, with each subsequent loss lot must be increased.

Not a connoisseur of casinos, but there is an opinion that in your example, but without doubling rates, we would catch all the fallen out colours, and the loss on their alternation. And end the cycle at a given + and start again.
 

drknn:

However, the losses were not that great. The maximum drawdown was eight losses in a row. With a minimum bet of 1 cent, that works out to:

1-2-4-8-16-32-64-128

Therefore, you should always have 512 cents to spare.


If with such restrictions, when the maximum losing streak does not exceed 8 spins, then in this case any fool and unequal odds will win martin, without any manipulation of colours. Let there be at least 36 zeros on the reel. It is important that the losing streak does not exceed 8 spins. It's just a matter of finding a casino that kindly restricts losing streaks to its gamblers. And we'll be in the sweet spot.
 
timbo:

Do you have a tradable random independent process with a probability other than 0.5 in mind? There are guys who trade with no spread and no commission and their capital is virtually unlimited. For such a process they will pay you more than the whole DC you trade in. I can introduce you to them if you have a real reason.

If this also applies to the MO profit per trade for the TC at a constant lot, then I'll remember your suggestion just in case :). Although it's likely to be almost impossible to prove such a thing, no matter what the test results show.

Actually, I'm just pulling in the direction of a real test: run it through history and you get it all :). Intuitively it seems that such a test should be based on constructing an empirical distribution of increments.