Experiment - page 21
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
We can conduct a speculative experiment - let there be one TS on one currency pair. TP equals SL, the probability of SL realization - 48%, TP - 52%. Initial deposit is $1000, leverage 1:100, we enter into trade at $100. If we carry out 1000 such trades then we obtain the deposit change trajectory, in fact it is the value of points gained during the whole set. If we carry out 500 such sets, we obtain the following picture:
If to use this same TS for 10 different pairs and break the deal volume, i.e. to enter $10 at each pair, the total deposit load in a moment will be the same, but balance changes look much more beautiful.
Of course. The expediency of maximizing the number of outcomes (diversification) follows from the paradox of increasing the bet for the player with a negative expectation. A player with a positive expected payout should do the opposite: increase the number of outcomes and decrease the bet size, because the probability of winning grows with the size of the series.
The problem of ruining a player
Of course. The expediency of maximising the number of outcomes (diversification) follows from the paradox of increasing the bet for a player with a negative expected payoff. A player with a positive expected payout is reasonable to do the opposite: increase the number of outcomes and decrease the bet size, because the probability of winning grows with the size of the series.
The Gambler's Brokeback Problem
A lot of small volume trades with a positive expectation of winning?
You don't know if the stock will rise immediately after you enter. There may be a pullback that throws you out of the market, then the stock will continue its expected rise, only you won't participate in it anymore.
Look at the chart on the previous page, there are many realisations that have gone to 0 on the balance sheet, despite the positive MO. That's the answer to the question why you shouldn't gambling on the whole cutscene, even having probability of winning >0.5. It's just bad luck, and the deposit is already over. Had the bet size been smaller - some of those losing realisations would have come back into positive territory, the bigger part the smaller the bet size would have been.
We can conduct a speculative experiment - let there be one TS on one currency pair. TP equals SL, the probability of SL realization - 48%, TP - 52%. Initial deposit is $1000, leverage 1:100, we enter into deals of $100. If we carry out 1000 such trades then we obtain the deposit change trajectory, in fact it is the value of points gained during the whole set. If we carry out 500 such sets, we obtain the following picture:
If to use the same TS for 10 different pairs and break volume of transaction, i.e. to enter $10 at each pair, total load on deposit in a moment will be the same, but balance changes look much more beautiful.
come to your senses, start thinking with your head.
the above charts have nothing to do with diversification and trading on 10 pairs. None other than personal cockroaches.
We don't have 10 independent pairs. And most importantly - the chart itself is number 2 in the way it is calculated, it is about the choice of lot size when trading one pair (you just got a sample in 10 steps).
Hence the result: if the probability of winning is higher than 0.5, then when trading with a smaller lot it is more difficult to lose. That's not fantastic.
You don't know if the stock will rise immediately after you enter. There may be a pullback that throws you out of the market, then the stock will continue its expected rise, only you will no longer participate in it.
Look at the chart on the previous page, there are many realisations that have gone to 0 on the balance sheet, despite the positive MO. That's the answer to the question why you don't have to gambling on the whole cutscene, even having a winning probability >0.5. It's just bad luck and the deposit has already run out. Had the bet size been smaller - some of those losing realisations would have come back into positive territory, the bigger part the smaller the bet size would have been.
That's weird. I mean, for example, I want to buy a stock on everything without leverage. I buy vtb shares for 50,000p on the market. And the share price is all at once?)
"On the whole cutlet" is not without leverage, it implies high risk and large possible profits. And all sorts of things happen, for example companies sometimes go bankrupt and instead of money the shareholder gets a lump of oil and 3 cents on the dollar.
Anyway, it's a general principle and probabilities for a series of outcomes, not a particular case.
We can conduct a speculative experiment - let there be one TS on one currency pair. TP equals SL, the probability of SL realization is 48%, TP - 52%. Initial deposit is $1000, leverage 1:100, we enter into deals of $100. If we carry out 1000 such trades then we obtain the deposit change trajectory, in fact it is the value of points gained during the whole set. If we carry out 500 such sets, we obtain the following picture:
If to use the same TS for 10 different pairs and break the deal volume, i.e. to enter $10 at each pair, the total deposit load in a moment will be the same, but balance changes look much more beautiful.
Not about that at all.
Yes, and a very dirty "experiment", as a result the "theory" will be very far from practice. The behaviour of trading instruments is not the behaviour of white noise. Neither is the behaviour of the mutual-correlation function (cross-correlation) of any two instruments.
He doesn't know what diversification in trading is and what it's used for - what more speculative experiments....
:)) no comment...
we do not have 10 independent pairs. And most importantly - the chart itself is #2 in the way of calculation, it is about the choice of lot size when trading one pair (you just got a sample in 10 steps).
It is clear that in the second case, the balance graph is built on independent rows, it is banal-obvious. The first and second pictures, are two limiting cases - all other cases that can be built on series with different degree of correlation lie between them. If for the second chart take series with 100% correlation, then the picture will degenerate to the first one - open 10 deals with small lot and one large lot simultaneously on one pair, there will be no difference between them.
If we take the average of all sets for the first and second picture, they will approximately coincide, because they are determined by initial SL/TP probabilities. As the number of rows increases and the lots split accordingly, the picture will narrow down to that average.
Yusuf has 34 pairs, of course they are correlated to a different extent, first of all via dollar index. But due to the great number of pairs, if the forecasting system had had some rationality, the trading would have become positive.
Nikolai Semko:
It's not about that at all.
Yes, and a very messy "experiment" as well, with the result that the "theory" will be very far from practice. The behaviour of trading instruments is not the behaviour of white noise. Neither is the behaviour of a mutual correlation function (cross-correlation) of any two instruments.
In what way is the dirtiness expressed?
In my opinion, on the contrary, very useful modelling from a practical point of view. Suppose you invented some super-system that can predict the direction of closing of a 4-hour bar with a decent probability of 54/46. Then you improve it and it is able to make predictions on 9 other pairs with approximately the same probability. You need to understand how much volume to enter on each pair, in order not to fail to lose. You calculate the correlation of pairs, volume weights and build a similar simulation.
It is clear that in the second case the balance graph is plotted on independent series, it is banal-obvious. The first and second pictures are two limiting cases - all other cases that can be built on series with different degree of correlation lie between them. If for the second chart take series with 100% correlation, then the picture will degenerate to the first one - open 10 deals with small lot and one large lot simultaneously on one pair, there will be no difference between them.
If we take the average of all sets for the first and second picture, they will approximately coincide, because they are determined by initial SL/TP probabilities. As the number of rows increases and the lots split accordingly, the picture will narrow down to that average.
Yusuf has 34 pairs, of course they are correlated to a different extent, first of all via dollar index. But due to the great number of pairs, if the forecasting system had any rationality at all, the trading would be in the plus.
Once again, the main thing: you do not have pictures of diversification. you have pictures of trading with different lots at P!=0.5
Only in the upper picture (#1) the lot is big and X number of consecutive trades, while in the second one (#2) the lot is smaller and X number of tens of trades.
correlation is a second matter...
There are a lot of drawbacks in such "experiments". They don't take into account simultaneity/non-simultaneousness/different density/length of trades. They don't take into account that 0 is fatal not at closing, but in the process (stop-out).
When trading on 10 tools it is necessary to consider the margin-call. We cannot do without it - equity has fallen, but we are still in the plus, and the margin is gone, and we cannot open a new order.