If you can look under the hood (i.e. at what data points you are averaging), just run the average for each 20 samples of x60 (60 minute sampled data point) and verify if it equates to what they display/calculate.
As prior poster suggested, that x60 data point might be o,h,l,cl or some average...