Spread calculation discrepancies

[checkraise]

it is my understanding that eSig calculates spreads for intraday charts using the 1-minute OHLC data of each spread component to derive the OHLC of the spread for that minute. Is this correct?

I noticed several suspicious bars on the 1-minute chart of (SPY - IWM)., so using Tools/Data Export, I put the data from the individual stocks into a spreadsheet, and calculated the spread OHLC for each 1-minute bar by subtracting OHLC of IWM from the corresponding 1-minute OHLC of SPY. I then compared the datapoints of the spreadsheet spread to the datapoints of the eSig spread.

There is quite a discrepancy between the eSig spread data and the spreadsheet data. The Open and Close for all 390 i-minute bars from 7-12-06 match perfectly, no discrepancy there, but only 205 of 390 Highs matched, and just 208 of 390 lows matched. Differences ranged from 0.01 to 0.48 .

Why do OC match perfectly but HL do not?

8.0 (Build 779) here.

[checkraise]

OK I see what's going on here. After performing the spread calculations on each 1-min bar, it is possible, and in the case of this spread, likely, that H will be less than L or C or O, or that L will be greater than H or O or C. So the H of one component minus the H of the other component is not necessarily the H of the spread. This accounts for the nearly all of the discrepancies between the spreadsheet spread and the eSig spread, because the spreadsheet did not check for the abberations of the inequalities.

So what I'm left with are the several huge spikes each day which do not reflect the 1-min prices at the time of the spike. They cannot be fixed by the bar editor and are usually corrected by the following day, but when using a 1-min chart, the next-day corrections are not very useful.