<<Brown's local manager said he is getting 40-50 complaint calls a
day from his customers on this issue. He spends a good amount of
time now trying to renegotiate executions with their MMs. My
concern is this is a new practice MMs have decided to institute
nationwide to counter the execution speed brought by direct
Internet and ECN interfaces.>>
I am in no way an apologist for the wholesalers, but there is a flip side to this coin. This entire 'basket' or 'bucket' thing (the automatic cutoffs on the autoexecute systems) came about after the internet hype-safari this spring, where stuff like ktel, marg etc. went up hugely in very short periods of time. Up until that time the big 3 wholesalers would almost never cut a stock off auto, and traders (market makers) would find themselves suddenly short 50k of a stock from autofills in a minute and already underwater when some hype-event happened in a stock they traded. Firms were literally losing a million+ a day on a given stock (this has a morale effect, since even though a market maker may only get 20-25% of the profit of his book, a million dollar hit can screw up his year terribly). This is when herzog, nite and mayer all basicly simoltaneously started saying they were redoing their systems with one type of basket or another. After a few months the coding was in place, and all sorts of anomolies started popping up. Traders (mm's), naturally, have learned to use the new system to their own advantage, to load up their book one way or another, though I think on some of the crazier stocks that do monstrous volume people overestimate the trader's ability to even notice their order, there is so much stuff going across their screen. The baskets were designed to control a firm's risk in a stock....to keep the trader from suddenly being short 100k and 5 points under on a stock that takes off....traders all over are getting hit on these latest hype-stocks but their losses are nothing compared to if they were wide open to their correspondents on automatic.
It is an interesting problem. |
