Traders' Roundtable

This is a newbie question and I apologize if it's been discussed previously.

The issue is which data series to use - electronic vs. floor+electronic - if I'll be trading only the electronic market.

Using the 10 Year US Treasury Note contract as an example - It looks to me that Interactive Brokers offers only the electronic market (symbol ZN). CSI offers both ZN (electronic only) and TY (floor+electronic combined) data. I notice that the default data series in Trading Blox is TY. On one hand I can see the benefit to using the floor+electronic combined data; on the other hand, if you plan on trading only the electronic market is there some benefit to "testing what you trade/trading what you test." Are the data so similar that it doesn't really matter?

Any thoughts/comments will be greatly appreciated.

Bubba,

Most TY activity does take place (at least for now) during traditional day session hours.

Contract: since IB is set up for electronic markets that is your only practical choice.

CSI: combined contract has 'pit' session close but carries the max(pit, electronic) and min(pit,electronic) for High Low.

L

Thanks.

So, assuming trading with IB (electronic markets), which is best to use for backtesting?

One advantage I can see with the combined electronic+floor data for backtesting is longer history (for the US 10 Year Note the combined data go back to 1982 vs. 1995 for the electronic only data).

Combined (pit + electronic) does give you longer history BUT availability of electronic sessions changed even pit behavior!

If you can, develop your strategy using a mix of inactive markets.

CSI has a huge selection of inactive markets.
Make a separate portfolio "dead" markets during the electronic age. Export the files and delete the portfolio ONLY but KEEP the files. This way, your CSI account is not charged for the 'extra' markets.

Once the strategy is to your satisfaction, test on 1st half of target markets and refine your model. Cut the second half data set in two, get a baseline value (that is half to 3/4 of data). Compare out of sample, last quarter to in sample.

L