CSI Eurodollar data

Use this forum to discuss data providers like CSI, charting, or other non testing software.
Post Reply
Corn Elius
Senior Member
Senior Member
Posts: 28
Joined: Mon Jul 21, 2003 12:33 am

CSI Eurodollar data

Post by Corn Elius »

Hi. Is anyone using CSI to construct continuous contracts for the Eurodollar? My CSI keeps switching from Sept 2005 and March 2006, and to top it off, the resultant continuous contract has wrong data. I've talked to another person using CSI and he has the same problem. I'm using version 2.8.3 and he's using 2.8.1. CSI hasn't answered my emails yet.

Anyone have a fix?
JackR
Full Member
Full Member
Posts: 14
Joined: Sun Jul 31, 2005 7:39 pm
Location: Washington, DC

Post by JackR »

What is the CSI product number or symbol? What method are you using to join contracts? C++, FORTRAN, Close-to-Close, etc.
RedRock
Roundtable Knight
Roundtable Knight
Posts: 944
Joined: Fri Jan 30, 2004 3:54 pm
Location: Arizona

Post by RedRock »

Mine did it too. I think there is a switch that will only roll after two days of higher OI. To follow that however would mean always rolling ++ after the 'trade' has made its move.

Perhaps download a fresh database file for ED and rebuild your output files if they are corrupt.
JackR
Full Member
Full Member
Posts: 14
Joined: Sun Jul 31, 2005 7:39 pm
Location: Washington, DC

Post by JackR »

CSI provides a number of ways to roll futures. I have never traded the EuroDollar and that is why I wanted the CSI product number. I thought I'd look at the problem and see if it was related to selecting different roll triggers.

Jack
Corn Elius
Senior Member
Senior Member
Posts: 28
Joined: Mon Jul 21, 2003 12:33 am

Post by Corn Elius »

ED is 141 and ED day only is 269. I got a reply from CSI that says to swithc from the C++ back adjuster to the older Fortran Back adjuster. Better, but the back adjusted data that ends with the March 2006 contract is very different from the actual March 2006 contract. Roll Trigger is Open Interest.
sluggo
Roundtable Knight
Roundtable Knight
Posts: 2987
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo »

I trade "ED2" (CME Eurodollars, day session only) in a real account with real money, using a mechanical system that reads CSI continuous contracts to generate its daily signals. I haven't been seeing any problems, screen capture image below. The ninth column of data is the contract month; it smoothly transitions from Dec to March without chattering back and forth between the two of them. I use Roll Trigger = Date and I use the C++ Back Adjuster exclusively.

Sometimes when I meet a trader who triggers rollovers using open interest, I ask why. Why open interest rather than volume, for instance? The answers usually are surprising, at least surprising to me. The most surprising answer of all is: "I was in a hurry so I just copied the way some guy did it. I don't feel like experimenting with different rollovers, how BORING is that! I'd rather spend my precious research time looking at different systems and position sizing methods."
Attachments
ed2.png
ed2.png (29.7 KiB) Viewed 6057 times
RedRock
Roundtable Knight
Roundtable Knight
Posts: 944
Joined: Fri Jan 30, 2004 3:54 pm
Location: Arizona

Post by RedRock »

SLUGGO SAYS:
Sometimes when I meet a trader who triggers rollovers using open interest, I ask why. Why open interest rather than volume, for instance? The answers usually are surprising, at least surprising to me. The most surprising answer of all is: "I was in a hurry so I just copied the way some guy did it. I don't feel like experimenting with different rollovers, how BORING is that! I'd rather spend my precious research time looking at different systems and position sizing methods."
It seems OI rolling does a fair job of capturing the move to the new top month. In the thinner markets rolling on some arbitrary date before expiration could roll you into an illiquid exp. month. Volume could be a faster way of capturing the top month change than OI. But typically OI does a pretty good job. Would seem most OI that does roll is not interested in delivery. Thus would be large traders and funds whom constitute the bulk of any given markets depth. Why not follow them. Testing each method would be the only way to say this for certain as Sluggo says. If CSI had some intelligence built in to know 1st notice date in advance. I would like to test rolling the day before 1st notice.

Concerning CSI and the switch back roll. Perhaps the exchange or csi published an error in OI for 3/06 which caused the roll. And was ultimately corrected resulting in the fallback to sep 05. When mine rolled, I followed it. But did not switch back to sep 05 when csi did. A few days later the trade really did move to mch06 and so I’m back in synch.

I have just recently added ED2 to my portfolio and am not experienced with rolling ED2 using OI specifically. I know the Eurodollar market has some peculiarities in terms of out month volatility and liquidity vs the close months.

Sluggo. I'm curious about how you selected a date to roll your contracts. Is this a date standard which applies to all markets in your portfolio equally? Or does each market have its own specific date formula???

rr
stancramer
Roundtable Fellow
Roundtable Fellow
Posts: 60
Joined: Sat Feb 07, 2004 10:34 am
Location: Washington DC

Post by stancramer »

If you're saying that large Open Interest means high liquidity, I don't agree. Volume is the number of contracts that traded on a certain day. Open Interest measures the number of contracts that didn't trade on a certain day. If volume=100 and open interest = one million, there were 999,900 contracts that didn't trade that day. In my opinion, a day with volume=100 contracts is extremely illiquid, regardless of the size of open interest. I myself sure wouldn't want to execute a rollover trade on a day with volume=100, no matter what the open interest was.
JackR
Full Member
Full Member
Posts: 14
Joined: Sun Jul 31, 2005 7:39 pm
Location: Washington, DC

Post by JackR »

CSI offers a very large number of ways to trigger a rollover. I think we can assume that they would not have made so many choices available if there were not valid reasons for each method. Here is an extract from the CSI help file:

Please note that in some instances of the triggers you may select the roll to be effected on the 1st, 2nd, 3rd, or 4th instance of the trigger event, e.g. roll on 2nd instance of OI of back montrh exceeding front month.
----------------------------
Roll Trigger

This input determines when UA switches from one futures contract to the next in creating your back-adjusted series. You may choose between using volume and/or open interest or a specific date as explained below.

When creating back-adjusted continuous files that use volume and/or open interest as roll triggers, the result will depend upon whether the raw data series is viewed forward or backward in time. The "Standard" (Fortran) back adjuster views a given market backward in time when it builds a back-adjusted file. The "C++" back adjuster views a given market forward in time. In certain circumstances, the results of these two methods will be different. You can change the back adjuster setting for this portfolio through the Back Adjusting tab in "Edit Portfolio Settings" from the Portfolio menu.

· Open Interest - Click here to exclude the nearest contract when heaviest open interest shifts to the subsequent delivery month.

· Volume - Click here to exclude the nearest contract when heaviest volume shifts to the subsequent delivery month.

· Open Interest and Volume - Click here to exclude the nearest contract when BOTH heaviest open interest AND heaviest volume simultaneously shift to the subsequent delivery month.

· Open Interest or Volume - Click here to exclude the nearest contract when EITHER heaviest open interest OR heaviest volume shifts to the subsequent delivery month.

· Date - This option lets your computed contract roll from one contract to the next on a specific day of the month. Click here and type the day of the month you wish rollforward to occur. Enter the roll-forward date (1 to 31) within the rollover month. Use 31 to roll on the last trading day. This prompt coordinates with another prompt (at right).

Start/End of Month - The default rolling criterion is for the day of the month (above) to be calibrated to the beginning of the month. An alternate rolling option is to select End of Month here, and enter the number of days prior to the end of the month to change contracts in the box at left. When "End of Month" is selected, enter the number of calendar days before the month is over that you want to roll out of a contract (1 to 28). Think of this as an inverted calendar. Enter 1 for the last day of the month, 2 for the next-to-last day, etc. This option is useful for traders who want to avoid risking delivery of a commodity by rolling out of a contract on the first notice day, which is often calibrated relative to the end of the month.

Months Prior - In the box to the left of this prompt, enter the number of months prior to the contract's expiration month when rolling should occur. Enter 0 for rolling within the expiration month, 1 for one month prior, etc.



· Strictly by Days Before Expiration Date - Click here to keep the lead contract in your continuous series until a specific date before expiration of the appropriate contract. To calibrate an early roll date relative to the expiration date, enter the number of days before expiration for the rollover to occur. Enter 0 to roll on the expiration date. The "expiration date" used here is based on the typical expiration date for the specific commodity, and may not be accurate at all times.

Roll Timing (Not offered if "Date" is used as Roll Trigger)

This determines when the near contract is dropped and replaced by the next in your back- and forward- adjusted contracts. The choices are:

· Anticipate - With this choice, the last close of the expiring contract is included on the day before your specified rollforward date. The first opening price of the new lead contract appears in the series on the specified rollforward date.

· Aligned with Price Data - With this choice, the last close of the expiring contract is posted on the day of the specified rollforward. The first open of the new contract is posted the day after the specified rollforward date.

· When Known - With this choice, the final close of the expiring contract appears the day after the specified rollforward date and the first open of the new contract appears two days after the specified rollforward.

©2002 - 2005 CSI

--------------------------

Jack
RedRock
Roundtable Knight
Roundtable Knight
Posts: 944
Joined: Fri Jan 30, 2004 3:54 pm
Location: Arizona

Post by RedRock »

8)

More bellcurves remain
sluggo
Roundtable Knight
Roundtable Knight
Posts: 2987
Joined: Fri Jun 11, 2004 2:50 pm

Post by sluggo »

stancramer wrote:Open Interest measures the number of contracts that didn't trade on a certain day. If volume=100 and open interest = one million, there were 999,900 contracts that didn't trade that day.
RedRock wrote:I would like to test rolling the day before 1st notice.
I agree with both of these sentiments. In my opinion the most important aspect of rollovers is their timing w.r.t. first notice and expiration day; the 2nd most important aspect is volume; and the 3rd most important aspect is backtested system performance. 4th place goes to "which contract months will I / won't I roll into?" (Example: September soybeans, yes or no? My answer: no) Open interest isn't nearly as important as these aspects, at least in my opinion.

viewtopic.php?p=12399&highlight=crude+contrac%2A#12399
viewtopic.php?p=6203&highlight=cl%2Aeven%2A#6203
Old European
Roundtable Knight
Roundtable Knight
Posts: 144
Joined: Mon Apr 19, 2004 3:04 pm
Location: Old Europe

Post by Old European »

Sluggo,

I still think that rolling-over on open interest cross-overs is a very acceptable thing to do.

1. It does a very good job as far as timing wrt first notice day is concerned
2. The higher the volume the lower the slippage usually is. But I have noticed that cross-over on volume most of the time takes place one or two days after cross-over on open interest (not always but generally speaking) and that there is an information time lag of two days, so that the cross-over really takes place exactly at the moment slippage is supposed to be very reasonable and that is what finally counts
3. Every component of every system has always to be backtested extensively. But in my own backtesting I have never seen a large advantage in prefering one roll-over procedure over another. And I'm very sceptic of a system when its edge relies too much on its roll-over method
4. I agree that it is worthwile to pay some attention to the contract months. Sometimes roll-over costs (commission and slippage) can be reduced substantially by only working with the most liquid months

Open interest is a very good indicator for market (month) depth and importance. When the price moves by one tick, you know exactly whether it is one thousand or a hundred thousand contracts that are going to be affected.

Of course other roll-over methods have their advantages. Rolling-over by date for example allows you to plan your roll-overs well in advance, which can be very handy.

Cheers,

Old European
Post Reply