|
PureBytes Links
Trading Reference Links
|
I have to agree with Andrew. After 6 mos of 6 to 20 round trips per day, I
can only recall One time where I had negative slippage on a stop loss. I
did have a few where I had positive slippage of 1 tick. And they do send
stop limits directly to Globex.
David
----- Original Message -----
From: "AndrewN" <tradera@xxxxxxxxxx>
To: "Omega List" <omega-list@xxxxxxxxxx>
Sent: Tuesday, February 11, 2003 9:01 AM
Subject: RE: IB TWS problem with stops
> IB has a great trading platform for very many reasons that I won't go into
> now. As far as queuing of stops is concerned, I can't remember the last
time
> I had any slippage on a stop. As for any concerns about stops not being
> elected by IB, just place a stop limit order. These ARE sent to Globex.
> Problem solved.
>
> Andrew
>
>
> -----Original Message-----
> From: Simon Campbell [mailto:simtrader@xxxxxxxxxx]
> Sent: Tuesday, February 11, 2003 7:18 AM
> To: Omega List
> Subject: Re: IB TWS problem with stops
>
>
> Why any futures trader uses IB is beyond me given the way they handle
> stop orders. Unless they've finally made the long-awaited change of
> sending stops native to Globex (judging by the comments it seems not)
> then in my opinion, it's a mess because your stop is only held on IB
> servers which means:
>
> - your order will only get sent to Globex as a market order when the
> stop price is triggered. In the best of cirucmstances (when it works)
> this means that your fill takes the lowest priority. All stops at
> the same price held directly on Globex will be filled before your
> order is. Orders held native on globex are timestamped and queued.
> Even if you sent your IB stop order four hours earlier, Globex
> doesn't even know about it until after the price is hit ...because
> only then does IB actually send it off. Bottom line? time lag in
> getting the order off (even if <1 second) and bottom of the pile
> priority, all mean increased risk of a tick slippage. At $12.50 ES,
> what good is your extra $1 commission saving?
>
> - Your stop relies on IBs own datafeed to trigger it and send it to
> Globex. A problem with their datafeed means your order can go
> unexecuted. My understanding is that it's not a true streaming
> datafeed but rather a snapshot every second or so. What if the
> snapshot misses your stop price? Or what if the feed goes down for
> 10 seconds around the time your stop would be triggered? Your stop
> goes unfilled that's what!
>
> Contrast this to the situation where your stop is held directly on
> Globex. It's been timestamped and received by the entity actually
> doing the execution. It will receive preferred filling. Your PC
> could blow up, your brokers servers could explode ...and nonetheless
> you are guaranteed to get executed. The worst that happens is that
> your brokers order software goes iffy (it happens to them all) and it
> says your order is still working. But if you KNOW your order was
> sent to Globex you have total peace of mind that the truth is you
> were executed and they just aren't showing it yet. You can relax that
> any software glitch has no material impact on you. Contrast that to
> IB where you are sitting there with a "working" order and wondering:
> "was I filled?", "is the system just slow?", "am I still in the
> market?". Who needs that hassle in their life?
>
> To me, it's such a basic fundamental to have your stop orders only
> rest on Globex servers ...trading with them on a brokers' servers or
> keeping them on your PC is just insanity. Pay an extra buck or two
> and use a J-Trader platform* that holds stops directly on Globex.
> IMHO the commission savings of IB just aren't worth the extra risk of
> $12.50 slippage, the extra risk of unexecuted orders, and the mental
> stress it puts on your ability to successfully trade, when problems
> do occur. There's enough stress in trading without worrying whether
> your orders are filled or not.
>
> S.
> *Not all versions of J-Trader can send stops directly to Globex (some
> versions only hold them on your PC), but stand alone version
> 2.8.0.6.D does offer this capability.
>
|