Trend line polynomial function

[FFpotaters]

Having trouble figuring out what I am missing about the trendline function as displayed on a scatter plot chart and why using the function does not generate results in line with the R2 value. Here are the details. I have the following data set:
2331 99
2262 96
2259 96
2235 94
2175 87
2163 85
2067 66
2049 62

I scatter plot this data set, and insert a polynomial trendline, displaying the equation on the chart. Results are the function y = -0.0004x2 + 2.0072x - 2253.2 with Rē = 0.9999. Looks promising. However, when I duplicate the function as a formula to get predicted results, my answers are no where close to the original Y values. Instead, I get the following computed Y values for each X:

2331 252.16
2262 240.43
2259 239.83
2235 234.80
2175 220.21
2163 216.95
2067 186.69
2049 180.19

What am I missing? I guess it is possible that I have made a rudimentary mistake, but I have been through it a couple of times and keep getting the same problem...

[shg]

You're using one significant digit for the x^2 term, which has values in the range of 4 million.

	A	B	C	D
1	x	y	Fit
2	2331	99	98.96	C2: =$A$13*A2^2 + $B$13*A2 + $C$13
3	2262	96	96.17
4	2259	96	95.96
5	2235	94	93.97
6	2175	87	86.84
7	2163	85	85.04
8	2067	66	66.23
9	2049	62	61.83
10
11
12	x^2	x	b
13	-0.00043	2.007212	-2253.19
14	7.83E-06	0.034139	37.17581
15	0.999897	0.170569	#N/A
16	24227.41	5	#N/A
17	1409.73	0.145468	#N/A

Select A13:C17, paste =LINEST(B2:B9, A2:A9^{1,2}, , TRUE) in the formula bar, press and hold the Ctrl and Shift keys, then press Enter.

The red value is R2.

[FFpotaters]

shg - super helpful. thanks. I need to read up on the linest function, but this definitely solved the computational disconnect.

[shg]

You're welcome.