Math/statistics/data science/econometrics question for a million dollar

P

prototech

New member
Joined
Jun 14, 2023
Messages
4

Hi. Tricky math/data science question for a million dollar, for geniuses with iq 160+
?
​

We have 4 series of data - graphs that go somewhat randomly, we know that they are correlated and cointegrated, i.e. in some time frame they move almost the same/very similar - i.e. they go up/down together and move away and get closer to some reasonable distance, that's what we have calculated by the Johansen test, etc. (so it is almost 100% certain that they were and will be co-integrated and correlated) For example, the prices of similar / dependent commodities.

01. The question: How do we find out that one of the graphs/price has moved significantly away from the others, but at the same time there is a high probability that the distance will no longer increase significantly and, on the contrary, it is taking the opposite course - that is, it is already starting to approach the other 3 graphs.

02. If we were to use classic percentages, and the distance in percentages, how do we calculate it since all values are dynamic?
constantly with each other are not static numbers. If we also normalize the graphs to the percentage normalized range 0-100, it is always for the selected time period, but by changing the time range-period, the proportions will also change and it will be scattered. At the same time, how do we determine which of the prices / graphs is overpriced and which is underpriced compared to the others?

03. If we enter a trade, how do we know that the deviation is sufficient and will not increase further? of course, we never know exactly, but how do we determine a statistically significant deviation and at the same time a statistically high probability that the deviation will no longer increase, but on the contrary decrease - so we can enter the trade.

well thank you, if needed we can communicate via e-mail, thanks

Martin
 
prototech said:

?
​

We have 4 series of data....

well thank you, if needed we can communicate via e-mail, thanks
Click to expand...
That's not actually how it works here. Any communication takes place openly, on the forums.
 
prototech said:

Hi. Tricky math/data science question for a million dollar, for geniuses with iq 160+ ​

?
​

We have 4 series of data - graphs that go somewhat randomly, we know that they are correlated and cointegrated, i.e. in some time frame they move almost the same/very similar - i.e. they go up/down together and move away and get closer to some reasonable distance, that's what we have calculated by the Johansen test, etc. (so it is almost 100% certain that they were and will be co-integrated and correlated) For example, the prices of similar / dependent commodities.

01. The question: How do we find out that one of the graphs/price has moved significantly away from the others, but at the same time there is a high probability that the distance will no longer increase significantly and, on the contrary, it is taking the opposite course - that is, it is already starting to approach the other 3 graphs.

02. If we were to use classic percentages, and the distance in percentages, how do we calculate it since all values are dynamic?
constantly with each other are not static numbers. If we also normalize the graphs to the percentage normalized range 0-100, it is always for the selected time period, but by changing the time range-period, the proportions will also change and it will be scattered. At the same time, how do we determine which of the prices / graphs is overpriced and which is underpriced compared to the others?

03. If we enter a trade, how do we know that the deviation is sufficient and will not increase further? of course, we never know exactly, but how do we determine a statistically significant deviation and at the same time a statistically high probability that the deviation will no longer increase, but on the contrary decrease - so we can enter the trade.

well thank you, if needed we can communicate via e-mail, thanks

Martin
Click to expand...
Please show us what you have tried and exactly where you are stuck.

Please follow the rules of posting in this forum, as enunciated at:

READ BEFORE POSTING

Please share your work/thoughts about this problem
 
khansaheb said:
Please show us what you have tried and exactly where you are stuck.

Please follow the rules of posting in this forum, as enunciated at:

READ BEFORE POSTING

Please share your work/thoughts about this problem
Click to expand...

it is long and difficult to show here but spreads, z-score, bollinger etc applied to spreads. But still not 100% sure which stock is overbought and oversold when calcullating spreads because spreads are calcullated from normalized data = they are normalized to 0-100 from specified time range and ratios are changed when you change time period of calcullation. But there are many big hedge funds using this / similar technique, and very very sucesfully.
 
Hello, I wrote this here in the summer when there probably weren't many people on PC, so I'll try to remember if there is anyone who is interested in this area. Well thank you
 
prototech said:

Hi. Tricky math/data science question for a million dollar, for geniuses with iq 160+ ​

?
​

We have 4 series of data - graphs that go somewhat randomly, we know that they are correlated and cointegrated, i.e. in some time frame they move almost the same/very similar - i.e. they go up/down together and move away and get closer to some reasonable distance, that's what we have calculated by the Johansen test, etc. (so it is almost 100% certain that they were and will be co-integrated and correlated) For example, the prices of similar / dependent commodities.

01. The question: How do we find out that one of the graphs/price has moved significantly away from the others, but at the same time there is a high probability that the distance will no longer increase significantly and, on the contrary, it is taking the opposite course - that is, it is already starting to approach the other 3 graphs.

02. If we were to use classic percentages, and the distance in percentages, how do we calculate it since all values are dynamic?
constantly with each other are not static numbers. If we also normalize the graphs to the percentage normalized range 0-100, it is always for the selected time period, but by changing the time range-period, the proportions will also change and it will be scattered. At the same time, how do we determine which of the prices / graphs is overpriced and which is underpriced compared to the others?

03. If we enter a trade, how do we know that the deviation is sufficient and will not increase further? of course, we never know exactly, but how do we determine a statistically significant deviation and at the same time a statistically high probability that the deviation will no longer increase, but on the contrary decrease - so we can enter the trade.

well thank you, if needed we can communicate via e-mail, thanks

Martin
Click to expand...
Did you succeed?
 
You must log in or register to reply here.
Top