Data Study

[Far Fetcher]

I took results from 7-13 on EUR_EUR and ES_SPY, using the 1.02 format.
There is the data itself, as well as the trades, and merging took a
bit of modification to JArb. The trouble is date and time format
mismatch. A couple of mods later had the data and reports relating.
Then since I had a grip on ES vs SPY, and no knowledge of EUR vs EUR,
the results below are only ES V SPY.

I went with the idea that there is a usual spread or slippage in
flipping positions of the pair. In my data set of 14590 entries, 6844
of them had a value of - $17.50 This makes some sense as .25 ES points
= $12.50 and .01 * 500 shares of SPY = $5.00.

Next, ignoring commission for the moment, I sorted the records into
spreads of less, equal, and more than -17.50. I found 3282 less, 6844
equal, and 4464 more than. Then I happened to note that a number of
the "more than" group had equal ES bid and ask. A filter was needed!
Of the 4464 "more than" 1061 had the ES bid >= ES ask. Of the
remaining 3403, all had SPY bid > ask. Thus about a third of the
entries are not good. This seems very high.

There were an average of 5.95 (286/48) data points in the same second
a trade was made.

I have more but it's getting late.

Larry