actually there is a systematic way of doing this and it's basically the discrete version of the derivative
Define the forward difference operator Δf(n) = f(n+1)-f(n)
the binomial coefficients xCk are the monomial analogues of the continuous monomials x^k /k!
in the sense that Δ(xCk) = (xC[k-1]), much like how D(x^k /k!) = x^[k-1]/(k-1)!
so you can use this to "differentiate" any polynomial expression and obtain the first, second, third, etc. "derivatives"
now observe that in classical calculus that for any polynomial (of degree n) P(x), you can write the polynomial in the following form:
P(0)x⁰/0! + P'(0)x¹/1! + P''(0)x²/2! + P'''(0)x³/3! + ... + P^(n) (0) x^n/n! (this is just the maclaurin series of P(x))
The same thing is true in discrete calculus, that you just evaluate each difference at 0 to obtain the coefficients of the binomial coefficients.
So we proceed, first defining f(x) = x^5 + x^3 + 2x
Δf(x) = 5x⁴ + 10x³ + 13x² + 8x + 4
Δ²f(x) = 20x³ + 60x² + 76x + 36
Δ³f(x) = 60x² + 180x + 156
Δ⁴f(x) = 120x + 240
Δ⁵f(x) = 120
and notice that the constant terms of each of these are the same weights as i have from my previous answer.
change x to n to recover the problem (i kept it the same just to make the analogy clearer)