Search
× Search
Friday, December 20, 2024

Archived Discussions

Recent member discussions

The Algorithmic Traders' Association prides itself on providing a forum for the publication and dissemination of its members' white papers, research, reflections, works in progress, and other contributions. Please Note that archive searches and some of our members' publications are reserved for members only, so please log in or sign up to gain the most from our members' contributions.

I have some questions regarding automated order execution in a Bid/Ask trading environment. Anyone out there think they are knowledgeable enough to take a stab at trying to answer them?

photo

 William Schamp, President/Quantitative Analyst - Beacon Logic LLC

 Thursday, July 24, 2014

I have some questions regarding automated order execution in a Bid/Ask trading environment. Anyone out there think they are knowledgeable enough to take a stab at trying to answer them?


Print

5 comments on article "I have some questions regarding automated order execution in a Bid/Ask trading environment. Anyone out there think they are knowledgeable enough to take a stab at trying to answer them?"

photo

 Shawna Zhou, Product Manager at hundsun

 Monday, July 28, 2014



Your transmission must be the fastest in the market to earn the spread profit.


photo

 James Harris, Director, Equities Analysis & Trading, Standard Chartered Bank

 Monday, July 28, 2014



Dear William,

Very interesting! I wish you every success in developing this model. For what it is worth, I worked on a similar project a while ago but couldn't make the (attractive) model transcend into being workable live.

The main reasons for my failure:

1. You rarely, if ever, get positive slippage (i.e. better profit targets than planned), only negative slippage (i.e. worse stops than planned)

2. Negative slippage is catastrophic for systems with small profit targets relative to the bid/ask spread

3. My own limitations as a programmer: I couldn't code a satisfactory way of dealing with the queuing system. Scott’s ‘lean-on’ idea above will certainly help your negative slippage situation but beware: such a script essentially pre-empts/second-guesses your strategy which may mean some reduction in edge. Not all trades that looked like they would be losers would have turned out that way. For profit targets, I couldn't guarantee that my PT order would be filled within the spread for a whole host of reasons. That said, looking at volume traded at a price since the limit order was created versus the amount of volume in front of the order in the queue would help, but beyond me. I know now that processing time-stamped tick-data is probably the key element of any successful solution to this problem. (Side warning: if your data provider has a 99.9% reliability rating for time-stamped tick data and a security has 100,000 individual ticks per day then you will still have to cope with 100 instances of faulty data, every day.)

Amid growing frustration, for further back-testing / proof of concept reasons, I created a related model that was almost the same as the first but would be forced to wait until the trade moved in the anticipated direction by a full spread to take profit, at which point it would simply hit the bid or lift the offer. Accordingly, I would get the 1 tick profit near as dammit to 100% of the time the trade moved by a spread, but would miss out when it didn't. Of course this isn’t 'conceptually fair' on the first model as you will get a healthy percentage of fills within the planned spread, but at least I could trust that the results were replicable live. Sadly, the subsequent results were so far removed from those generated by the original system that I put the whole concept in the 'too hard' bracket, weeping a little but also pleased that I did not risk real capital. As you correctly say, for you this is a perfect outcome as you WANT your competitors to quit. To the victor, the spoils!


photo

 Scott Boulette, Algorithmic Trading

 Monday, July 28, 2014



@James - Let me clarify my post a bit. I wasn't suggesting someone lean on an order for exactly the reasons you stated. What I was suggesting is that this will work if you can find the correct order to lean on.

These orders are most frequently found in the physical commodities and bonds because you have firms that want the underlying at a specific price for whatever reason.

As you did, I moved to a nearly identical model with a few critical modifications. I also have the advantage of a direct exchange feed which has very few errors (if any), not to mention a whole host of additional tags that are traditionally not found in a retail data feed.


photo

 Ernesto Giorgi, indipendenza

 Monday, July 28, 2014



@William Schamp.

Really, I have created the 'feeling' with reasonable success: it is published on www.finraptor.wordpress.com for the results and the algorithm is explained in finraptor2 and finraptor3. You may perhaps be interested. I have decided to stay out and not to use it because I think that the market is too dangerous (and manipulated). If you think differently, I can help you with suggestions.


photo

 Tony L., President at Day Trading Logics LLC.

 Friday, August 1, 2014



All instruments including the ES have a natural fluctuation between Peaks and troughs. Divergence is a natural event it occurs many times during the day. a Direction is not a straight line, you must account in your X stop loss a additive "the natural fluctuation before price moves again in the direction".

If you notice over 30 years everyone most traders will tell you if you are going for the gold taking the entire swing you must for the ES maintain stop tick value of 10 to 15 ticks. If you catch the crossover into the beginning of a new direction be prepare for the first divergence to come up within a few bars kinda like a fake out..

There are certain things that happen naturally, does it happen always yes but does it happen enough that you have not prepared for it, you will get stop loss out of good trades, this occurs to often.

The stop loss can not be just how much you are willing to lose or bare, the stop loss must be this plus the amount divergence that naturally occurs and its value is different for different instruments. So the math should go like this Stoploss = Risk + Divergence.

Therefore it is not unrealistic in accept stoploss tick values of 15 to 20 ticks. Over 20 ticks I think you entered a good trade in a bad time, close to a news event, close to a institutional dump, or you really did a bad trade. Now this of course if your intra-trading, if you are doing long term then hold on to your paints.

As for Market Orders or Order Limits, I prefer Order Limits but as the famous jake bernstein said " Don't try to call your broker for a better price, just get in the at market"

The only time I would disagree with this if your foundation strategy is flawed, you might have a fighting chance with a order limit. I ran thousands of back tests and found that Market orders give better profits at the end of year, but it is not for the fainted hearted, expect larger draw downs...

Please login or register to post comments.

TRADING FUTURES AND OPTIONS INVOLVES SUBSTANTIAL RISK OF LOSS AND IS NOT SUITABLE FOR ALL INVESTORS
Terms Of UsePrivacy StatementCopyright 2018 Algorithmic Traders Association