Flash Crash SEC Report
15 views
Skip to first unread message
nonlinear5
Oct 1, 2010, 4:10:40 PM10/1/10
to JBookTrader
Yesterday, the SEC released the May 6 Flash Crash report. It's a 100+
page document, and there are references to limit order book and market
depth all over the place, so I thought this would be good reading for
the JBT users: http://sec.gov/news/studies/2010/marketevents-report.pdf
page document, and there are references to limit order book and market
depth all over the place, so I thought this would be good reading for
the JBT users: http://sec.gov/news/studies/2010/marketevents-report.pdf
Shane
Oct 1, 2010, 4:19:59 PM10/1/10
to jbook...@googlegroups.com
on a related note:
FINRA Sanctions Trillium Brokerage Services, LLC, Director of Trading, Chief Compliance Officer, and Nine Traders $2.26 Million for Illicit Equities Trading Strategy
http://www.finra.org/Newsroom/NewsReleases/2010/P121951
Excerpt:
WASHINGTON — The Financial Industry Regulatory Authority (FINRA) today announced that it has censured and fined New York-based Trillium Brokerage Services, LLC, $1 million for using an illicit high frequency trading strategy and related supervisory failures. Trillium, through nine proprietary traders, entered numerous layered, non-bona fide market moving orders to generate selling or buying interest in specific stocks. By entering the non-bona fide orders, often in substantial size relative to a stock's overall legitimate pending order volume, Trillium traders created a false appearance of buy- or sell-side pressure.
This trading strategy induced other market participants to enter orders to execute against limit orders previously entered by the Trillium traders. Once their orders were filled, the Trillium traders would then immediately cancel orders that had only been designed to create the false appearance of market activity. As a result of this improper high frequency trading strategy, Trillium's traders obtained advantageous prices that otherwise would not have been available to them on 46,000 occasions. Other market participants were unaware that they were acting on the layered, illegitimate orders entered by Trillium traders.
FINRA Sanctions Trillium Brokerage Services, LLC, Director of Trading, Chief Compliance Officer, and Nine Traders $2.26 Million for Illicit Equities Trading Strategy
http://www.finra.org/Newsroom/NewsReleases/2010/P121951
Excerpt:
WASHINGTON — The Financial Industry Regulatory Authority (FINRA) today announced that it has censured and fined New York-based Trillium Brokerage Services, LLC, $1 million for using an illicit high frequency trading strategy and related supervisory failures. Trillium, through nine proprietary traders, entered numerous layered, non-bona fide market moving orders to generate selling or buying interest in specific stocks. By entering the non-bona fide orders, often in substantial size relative to a stock's overall legitimate pending order volume, Trillium traders created a false appearance of buy- or sell-side pressure.
This trading strategy induced other market participants to enter orders to execute against limit orders previously entered by the Trillium traders. Once their orders were filled, the Trillium traders would then immediately cancel orders that had only been designed to create the false appearance of market activity. As a result of this improper high frequency trading strategy, Trillium's traders obtained advantageous prices that otherwise would not have been available to them on 46,000 occasions. Other market participants were unaware that they were acting on the layered, illegitimate orders entered by Trillium traders.
--
You received this message because you are subscribed to the Google Groups "JBookTrader" group.
To post to this group, send email to jbook...@googlegroups.com.
To unsubscribe from this group, send email to jbooktrader...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/jbooktrader?hl=en.
John-Crichton McCutcheon
Oct 1, 2010, 4:30:25 PM10/1/10
to jbook...@googlegroups.com
Thanks. I wonder if they stop spoofing the order book maybe JBT will
get better results now?
get better results now?
nonlinear5
Oct 1, 2010, 4:33:41 PM10/1/10
to JBookTrader
> Thanks. I wonder if they stop spoofing the order book maybe JBT will
> get better results now?
>
Have you looked at the performance of the sample strategies included
> get better results now?
>
in the last version of JBT. They are pretty decent. I am about to
start trading live again.
Klaus
Oct 3, 2010, 9:49:33 AM10/3/10
to JBookTrader
Looks extremely good.. - are the algorithms shown the same ones as
contained in the distribution?
On 2 Okt., 00:28, Eugene Kononov <eugene.kono...@gmail.com> wrote:
> Here are the backtest results for my top-performing strategies. They've also
> proven to be well-performing in forward testing. These results are for the
> entire period of June 2009 to September 2010.
>
>
>
> On Fri, Oct 1, 2010 at 4:33 PM, nonlinear5 <eugene.kono...@gmail.com> wrote:
> > > Thanks. I wonder if they stop spoofing the order book maybe JBT will
> > > get better results now?
>
> > Have you looked at the performance of the sample strategies included
> > in the last version of JBT. They are pretty decent. I am about to
> > start trading live again.
>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "JBookTrader" group.
> > To post to this group, send email to jbook...@googlegroups.com.
> > To unsubscribe from this group, send email to
> > jbooktrader...@googlegroups.com<jbooktrader%2Bunsubscribe@googlegr oups.com>
> > .
> 57KAnzeigenHerunterladen
contained in the distribution?
On 2 Okt., 00:28, Eugene Kononov <eugene.kono...@gmail.com> wrote:
> Here are the backtest results for my top-performing strategies. They've also
> proven to be well-performing in forward testing. These results are for the
> entire period of June 2009 to September 2010.
>
>
>
> On Fri, Oct 1, 2010 at 4:33 PM, nonlinear5 <eugene.kono...@gmail.com> wrote:
> > > Thanks. I wonder if they stop spoofing the order book maybe JBT will
> > > get better results now?
>
> > Have you looked at the performance of the sample strategies included
> > in the last version of JBT. They are pretty decent. I am about to
> > start trading live again.
>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "JBookTrader" group.
> > To post to this group, send email to jbook...@googlegroups.com.
> > To unsubscribe from this group, send email to
> > .
> > For more options, visit this group at
> >http://groups.google.com/group/jbooktrader?hl=en.
>
>
>
> backtest.png
> >http://groups.google.com/group/jbooktrader?hl=en.
>
>
>
> 57KAnzeigenHerunterladen
Eugene Kononov
Oct 3, 2010, 10:31:20 AM10/3/10
to jbook...@googlegroups.com
Looks extremely good.. - are the algorithms shown the same ones as
contained in the distribution?
John-Crichton McCutcheon
Oct 4, 2010, 12:43:39 PM10/4/10
to jbook...@googlegroups.com
Thats good to hear. I never upgraded the Market Depth calculation
change from last summer. That is , in MarketDepth.java you changed
the calculation from the mid point of the high balance / low balance
for the interval to the EMA of the balances recorded during the
interval. At the time, it seemed to me that that change made the book
data much less volatile and so I found that it generated fewer
signals for the strategies I was using at the time such as Balancer.
Anyway, I'll see what those new strategies do against the "old depth"
calculation.
Since you say the performance looks good, I'll record with the current
depth calculation as well as the old.
change from last summer. That is , in MarketDepth.java you changed
the calculation from the mid point of the high balance / low balance
for the interval to the EMA of the balances recorded during the
interval. At the time, it seemed to me that that change made the book
data much less volatile and so I found that it generated fewer
signals for the strategies I was using at the time such as Balancer.
Anyway, I'll see what those new strategies do against the "old depth"
calculation.
Since you say the performance looks good, I'll record with the current
depth calculation as well as the old.
Klaus
Oct 4, 2010, 1:02:09 PM10/4/10
to JBookTrader
Dear John,
this is very interesting to hear. It would really be interesting to
see a side-by-side comparison
of identical strategies with deep and shallow book data. The only
thing is, we would then need
also all the data recorded with deep and shallow book.
Klaus
On 4 Okt., 18:43, John-Crichton McCutcheon
this is very interesting to hear. It would really be interesting to
see a side-by-side comparison
of identical strategies with deep and shallow book data. The only
thing is, we would then need
also all the data recorded with deep and shallow book.
Klaus
On 4 Okt., 18:43, John-Crichton McCutcheon
nonlinear5
Oct 4, 2010, 1:04:05 PM10/4/10
to JBookTrader
I'll make a qualifier for my comment about the performance. When I
looked at the days when my strategies had significant losses (like
$1,000 or so per day), I noticed that these days had suspicious data.
For example, on these days, the recording didn't start until 10:30am,
so the trades placed immediately after that time didn't have enough
price/balance history for the indicators. I ended up by simply
excluding these days from the recorded data file (there were 2 or 3
total). On the other hand, there was also May 6, the flash crash day,
on which my sample strategies made astronomical gains (in the $5,000
to $10,000 range), and that single day skewed the backtesting/
optimization results so much that I've dropped May 6 from my data set
as well. All in all, my working data set excludes what I believe to be
either erroneous or non-representative market data. As I've also
mentioned, the forward test results look very decent, too, which gives
me certain amount of confidence. Finally, I am also happy with the
fundamental aspects of them, that is to say, I am comfortable with
understanding why and how these strategies actually work. I am
starting live trading in about a week.
looked at the days when my strategies had significant losses (like
$1,000 or so per day), I noticed that these days had suspicious data.
For example, on these days, the recording didn't start until 10:30am,
so the trades placed immediately after that time didn't have enough
price/balance history for the indicators. I ended up by simply
excluding these days from the recorded data file (there were 2 or 3
total). On the other hand, there was also May 6, the flash crash day,
on which my sample strategies made astronomical gains (in the $5,000
to $10,000 range), and that single day skewed the backtesting/
optimization results so much that I've dropped May 6 from my data set
as well. All in all, my working data set excludes what I believe to be
either erroneous or non-representative market data. As I've also
mentioned, the forward test results look very decent, too, which gives
me certain amount of confidence. Finally, I am also happy with the
fundamental aspects of them, that is to say, I am comfortable with
understanding why and how these strategies actually work. I am
starting live trading in about a week.
John-Crichton McCutcheon
Oct 4, 2010, 3:15:12 PM10/4/10
to jbook...@googlegroups.com
You mean the 5 level book versus the 10 level book?
By the way, There was an article todaytalking about how 10 level book is
not deep enough and more is needed for market transparency.
http://www.bloomberg.com/news/2010-10-04/may-6-stock-crash-may-lead-to-new-rules-for-brokers-cftc-s-gensler-says.html
I think JBT strategies have a bright future.
By the way, There was an article todaytalking about how 10 level book is
not deep enough and more is needed for market transparency.
http://www.bloomberg.com/news/2010-10-04/may-6-stock-crash-may-lead-to-new-rules-for-brokers-cftc-s-gensler-says.html
I think JBT strategies have a bright future.
Reply all
Reply to author
Forward
0 new messages