it all depends on how far back you go when looking at the historical data.
we do it all the time in TA but usually no more than 6 months for current direction though resistance/support levels can be guessed at from longer data.
the problem with looking at data that is older than 6 months is that times change....fortunes change. One year ago the commodity/service that the company sells might have been in higher/lower demand than now. The company may have had a tougher/easier financial time then current. The market itself might have been in a bear/bull run compared to now and the company's fortunes might be tied to the overall market reaction more than others.
So you really have to be careful at attaching too much significance to really old historical data....past glories do not portend future results....just look at Nokia as an example.
However, the more recent the data...the better the prediction...you are extrapolating the current situation but 6 months would be an arbitrary limit I would say though that can increase or decrease depending on the company and the market status.
inhabitants is a pass no longer a static variety. Mules ought to pass the two procedures. drugs over, money and weapons lower back. the U. S. helps this pass as a results of fact it provides a huge Democratic balloting block in a popular Republican element of the country. Many human beings of Mexican/Indian extraction who've been here on the grounds that Ca, TX grew to become states have confronted hostility as a results of their background, yet are US voters. maximum do no longer in undemanding terms like the illegals as a results of fact they only tarnish their tricky won claims of popularity. The Republicans, regrettably have not been all too common with unlawful immigrations. Mexico in the time of the 1980-1990's used unlawful flows northward to rid itself of it extra mail uneducated inhabitants yet busing them North. Republican exploited this to pass the marketplace with lower priced workers. whilst those human beings moved to the North States --- Union observed the prospect for a skill grab. There are approximately 20 million unlawful interior the U. S. from Mexican or Latin American aspects. so long as a results of fact the U. S. tolerated the polices of the State of Mexico we can proceed to have this undertaking. we don't enable Canadians flood over the border. We deliver Haitian lower back. as a results of laxity human beings governance percentis now permitting export of chinese language to the U. S.. they could be spies or psychological aspects brokers yet they are going to plead persecution and non secular liberty and a few damn team of liberals will foyer to maintain them.
a) Risk - People (including me) use historical data to estmate future volatility all the time. As underexposed says, the data becomes less and less relevant as time goes on so the farther back you get, the less relevant it gets. True story - RiskMetrics is a premier risk package and they spent tons of time trying to come up with a general methodology for estimating future volatility. The spent big bucks (i.e., much bigger bucks than any academic could spend) testing various formulations of estimators of future vol from past vol and decided that among all the wildly fancy estimators anybody could think of that none worked better than the exponential weighted smoothed estimator they give in their public doc. I talked at length to the guy who did that project (Allen Malz) and I am a believer that he did it right. Want to estimate future vol (at least for risk management) - use RM approach.
b) return - You are just an idiot if you say past return predicts anything about future return. Just silly.
c) Covariance - Lots of research on this too. You can use an exponentially weighted kind of estiator here too (and RM does) but there is lots of good research out there that says you should "shrink toward 0". High correlations in one period are usually less in the next period.
Nothing's "wrong" with it. As long as your able to also determine the risk management points where action needs to be taken. You must be able to act on the hard or soft stops you determined (before you took the position). You must have a plan to exit on the upside.
Answers & Comments
Verified answer
When one companies majority product revenue source loses share to a new competitor..Palm, Nokia,Blackberry, Droid, IPhone/Pad..
it all depends on how far back you go when looking at the historical data.
we do it all the time in TA but usually no more than 6 months for current direction though resistance/support levels can be guessed at from longer data.
the problem with looking at data that is older than 6 months is that times change....fortunes change. One year ago the commodity/service that the company sells might have been in higher/lower demand than now. The company may have had a tougher/easier financial time then current. The market itself might have been in a bear/bull run compared to now and the company's fortunes might be tied to the overall market reaction more than others.
So you really have to be careful at attaching too much significance to really old historical data....past glories do not portend future results....just look at Nokia as an example.
However, the more recent the data...the better the prediction...you are extrapolating the current situation but 6 months would be an arbitrary limit I would say though that can increase or decrease depending on the company and the market status.
inhabitants is a pass no longer a static variety. Mules ought to pass the two procedures. drugs over, money and weapons lower back. the U. S. helps this pass as a results of fact it provides a huge Democratic balloting block in a popular Republican element of the country. Many human beings of Mexican/Indian extraction who've been here on the grounds that Ca, TX grew to become states have confronted hostility as a results of their background, yet are US voters. maximum do no longer in undemanding terms like the illegals as a results of fact they only tarnish their tricky won claims of popularity. The Republicans, regrettably have not been all too common with unlawful immigrations. Mexico in the time of the 1980-1990's used unlawful flows northward to rid itself of it extra mail uneducated inhabitants yet busing them North. Republican exploited this to pass the marketplace with lower priced workers. whilst those human beings moved to the North States --- Union observed the prospect for a skill grab. There are approximately 20 million unlawful interior the U. S. from Mexican or Latin American aspects. so long as a results of fact the U. S. tolerated the polices of the State of Mexico we can proceed to have this undertaking. we don't enable Canadians flood over the border. We deliver Haitian lower back. as a results of laxity human beings governance percentis now permitting export of chinese language to the U. S.. they could be spies or psychological aspects brokers yet they are going to plead persecution and non secular liberty and a few damn team of liberals will foyer to maintain them.
They are all different questions:
a) Risk - People (including me) use historical data to estmate future volatility all the time. As underexposed says, the data becomes less and less relevant as time goes on so the farther back you get, the less relevant it gets. True story - RiskMetrics is a premier risk package and they spent tons of time trying to come up with a general methodology for estimating future volatility. The spent big bucks (i.e., much bigger bucks than any academic could spend) testing various formulations of estimators of future vol from past vol and decided that among all the wildly fancy estimators anybody could think of that none worked better than the exponential weighted smoothed estimator they give in their public doc. I talked at length to the guy who did that project (Allen Malz) and I am a believer that he did it right. Want to estimate future vol (at least for risk management) - use RM approach.
b) return - You are just an idiot if you say past return predicts anything about future return. Just silly.
c) Covariance - Lots of research on this too. You can use an exponentially weighted kind of estiator here too (and RM does) but there is lots of good research out there that says you should "shrink toward 0". High correlations in one period are usually less in the next period.
Nothing's "wrong" with it. As long as your able to also determine the risk management points where action needs to be taken. You must be able to act on the hard or soft stops you determined (before you took the position). You must have a plan to exit on the upside.
In short you must have "rules".