Sunday, October 11, 2009

CME Tick Data Changes

Allrighty, so on the 5th, the CME started reporting more granular tick data. Oh joy! In a nutshell from the marketdelta crew:

:: Summary ::
The CME has begun “unbundling” some ticks/trades that were previously being sent as a single tick. From inspecting charts of the ES, it appears that the net result of this change is that approximately 2.5 times as many ticks or trades are being sent on average, WITH a much smaller average size per trade. The change does not effect the volume, just the number of trades (more trades with a smaller average size).

:: Here is what it means and what you will begin seeing ::
1) More Tickbars: If you are using tickbar charts, you will start seeing approximately 2.5 times as many tickbars per day as you were before because there are approximately 2.5 times as many ticks per day. The CME claims this will give the user a more accurate tickbar picture.

2) Slower Backfill: Backfill or downloading of historical data will take longer since there will be more ticks/trades to download.

3) Larger Database: The amount of the space in the database required to store a day of tick data will increase. Thus, the time it takes to backup or verify the database will increase.

4) Fewer Big (large lot) Trades: If you are filtering trades based on volume (in Time and Sales windows or using the Volume Breakdown indicator), you will see fewer large trades since many of these larger trades have been replaced with a few smaller trades. This change is rather substantial. The number of large trades (> 199 lots) per day on the ES dropped from approximately 1800 to approximately 400.

Now where my input comes in is with the adjustment needed to alter your current trading timeframe to the new adjusted timeframe. The TTM folks came out and said "multiply by 2.5x's" so a 233T is equal to a 582T but like most things out of the TTM folks, it sounds easy, simple, and sassy, but delivers nothing but epic failure. So I did my own analysis using a barcounter in TradeStation.

Taking the last 10 days prior to the conversion and finding the average amount of bars per day in two different timeframes I had been trading and comparing them to the average of the new amount of bars post October 5th the following results were collected. I then adjusted the needed increase in ticks until similar results were found.

What I found was that the conversion was between 1.5X's and 1.75X's as much data to correctly convert your old timeframe to a comparable new timeframe. But without any doubt it certainly isn't a conversion factor of 2.5x's as was suggested by others.


E-Mini Player said...

Good info!

M said...
This comment has been removed by the author.
Naresvara said...

Logical. But it would be different for various instruments. No?

Where does one find a bar counter for TS? Please advise

Adam said...

Interesting! Thank you for the heads up :)