Grand net a large amount of
Guide language
The article is with NYMEX crude oil exemple, analytic international crude oil and domestic chemical industry are tasted (bitumen, PTA, PE, PP, methanol, PVC and SC) dependency and linkage sex, from metric angle analytic crude oil and domestic chemical industry taste what the likelihood exists between the underlying concern between and breed to cover profit opportunity.
The active for the futures of each breed contract of our specific analysis bid (brunt agreement) , sampling length is (2013/01/01, 2018/05/31) , example number is 1275.
Dependency reachs regression
A
Correlation coefficient:
Pass the computation of correlation coefficient, we discover the effect of international oil price to domestic SC, BU, PTA is very distinct, remarkable to the influence of PE, PP, and not remarkable to the influence of MA, PVC.
[note: The extraction of correlation coefficient R is worth for (- 1, 1) , absolute value is more adjacent 1 state two variable dependency are taller, among them R=0 expresses two variable are nonexistent dependency, at the same time correlation coefficient is meant higher go because variable accuracy is taller,be being forecasted with independent variable, correlation coefficient does regression high to analyse ability significant) ]
B
Come loose bit of graph and recursive equation:
The computation that passes correlation coefficient and come loose the observation of bit of graph, we are independent variable X with NYMEX crude oil, taste with domestic chemical industry because of,be variable Y, build a yuan of linear to return to a model: Y= α + I of β Xi+ μ ;
[disturb randomly a μ is contented 3 big hypothesis, namely μ obedience normal school, Ε (μ ) =0, δ 2= constant disturbs randomly an oneself is not relevant, namely Cov (μ I, μ J) =0; Disturb randomly with independent variable not relevant, namely Cov (Xi, μ I) ] =0; [as contrast, we also are opposite the MA of dependency on the low side, PVC does regression]
... (here we omit T examines)
β coefficient is BU respectively (37.62) , PTA (46.88) , l (36.66) , PP (60.23) , MA (11.1) , PVC (12.17) , SC (8.1) ; Mean NYMEX crude oil every go up drop one is nodded, domestic chemical industry is tasted can corresponding go up drop 37.62, 46.88, 36.66, 62.03, 11.1, 12.17, 8.1 bits.
Plan to add up to goodness (but definitely coefficient) : R^2 is BU respectively (0.88) , PTA (0.83) , l (0.72) , PP (0.76) , MA (0.46) , PVC (0.17) , SC (0.85) ; Show BU, PTA, PE, PP, of SC it is good to plan to add up to goodness, MA, of PVC plan to add up to goodness difference. Namely of crude oil walk along potential energy to explain domestic chemical industry tastes BU(88%) , PTA (83% ) , l(72%) , PP (76% ) , SC (85% ) go up drop reason, but can explain methanol only (46% ) with PVC(17%) go up drop reason.
[note: Plan to add up to goodness R^2 to state regression drafts the level that close with observation point-blank, r^2 takes a cost: ≤ of 0 ≤ R^2 1, r^2 is close to 1 specification to plan to add up to degree to had been jumped over more. ]
