Can bad ticks trip stops?

Discussion in 'Order Execution' started by jftrader, Sep 29, 2005.

  1. From what I have seen, there are several possible causes of bad ticks:

    1) Bad aggregation, almost all of the aggregated feed providers (Thomson, HyperFeed, Comstock, etc) would, in very high data rate times / product (say eCBOT or high rate equities / options / futures), time slice. So the system would essentially take say 4-5 snapshots of the data every second, for instance. In this scenario, it is possible that very short spikes got filtered out, so the user / end-user system would miss data. The few well publicized cases in equities (when Yahoo! showed a price of 9000), is when the provider (Reuters in Yahoo!) would just have bad data slices.

    2) Bad multi-origin data, many products (options, equities, etc) are traded on multiple exchanges, if one of them produces a bad quote. PCX was / is notorious for sometimes providing bad Options data, so the aggregators would have make a decision. Sometime the system gets it right, some it gets wrong.

    3) Bad source data, much more rare, for instance the OPRA feed has been known to be off at times, as well as NYSE book does have bad sequences. But there are usually exchange-wide notices for this type of errors.
     
    #21     Oct 3, 2005
  2. You guys can pull up a "Time and Sales" list, sort it by exchange, and then you will some of the "out of sequence" ticks, etc.

    Don
     
    #22     Oct 3, 2005
  3. Here's a paper put together by the folks over at Tick Data on the topic of scrubbing data.
     
    #23     Oct 3, 2005
  4. This is definitely a problem for Nasdaq NNM stocks....I see this all of the time....and the Lightspeed platform I am using does not differentiate these "prints" other than to highlight them differently if outside of the BBO.
    IMHO, multiple approaches are needed to effectively program automatic stop orders:
    1) 2 or more consecutive prints occur at that price (remove out-of-sequence prints)
    2) price delta from prior print is less than "X" % (eliminate spikes caused by whatever)
    3) print price is within x ticks of best bid (buy); x ticks of best offer (sell)

    Naturally, all of these "tests" combined would eliminate erroneous stop orders (reduce false positives), but increase the chance that a valid stop order would be missed (lower reliability).

    Best approach is to make the stop order configuration GLOBAL for all symbols with the ability to OVERRIDE the parameters on an individual symbol basis.

    Does anyone know of a trading platform that either does this NOW or can be programmed to do this other than Tradestation ?
     
    #24     Oct 3, 2005
  5. bad ticks can surely trigger stops. i had QQQ bad ticks triggering my stops more than once. i complained to my broker at that time and he offered to put me back in the position, although it was not his fault - it was amex.
     
    #25     Oct 4, 2005
  6. That's a good point. Are brokers supposed to take responsibility for exchange errors? It seems to me like we always placate you retail investors without question...and the problem just gets worse as you keep asking us to take responsibility for exchange problems.
     
    #26     Oct 8, 2005
  7. and it's impossible to fight a bad bid/ask spread. here's one this morning on a Naz stock LIHRY:

    sell stop at 23.87, triggered off a bid of 23.83 with an ask of 24.67 (almost a dollar wide at the open) , which was RIGHT AFTER the stock opened at 24.25. the worst part was the execution wasn't until 9:31:20 at 24.09!
     
    #27     Oct 25, 2005
  8. have had the same happen to me. usually the AMEX will "mistakenly" keystroke the wrong big number and it will trigger ARCA stops.
     
    #28     Oct 25, 2005
  9. Only in a crappy software. Any good software should check the time & sales AND cross reference the bid or ask. This double check should virtually eliminate and bad stop triggers.
     
    #29     Oct 25, 2005
  10. #30     Oct 25, 2011