timeslice algo

[fxmisticat]

Now lavafx and citi and few other banks/ecns are providing a timeslice algo. When your algo gives you an entry or exit signal you want to drip feed your order in small pieces do not spook the market. I have been back testing various timeslice routines with various times chosen based on the time of day or liquidity.

BUY 30m EUR/USD over 60min.
min slice size: 100k
max slice size: 2m
buffer: 5m (first tier)
throttle: 1 (10%)
stop if >: 1.4
slippage: 1.5 (pips)
% change: 1%

no I have tried many different ways to apply this over a period of time.
During high volatile times and moving against your entry position is when the slicing is getting crushed.

I will go into detail on the different variations of the signals if people are interested in something like this.

I would like to hear your thoughts?

 

[MrGecko]

Isn't this just an Iceberg entry?

I think you will find most pro front ends will do something similar... especially if FIX is available.

 

[fxmisticat]

Isn't this just an Iceberg entry?

I think you will find most pro front ends will do something similar... especially if FIX is available.

not really. If you had 200m to get rid of you would move the market. A timeslice would allow you to drip pieces of it into the market and adjusting the pieces based on market signals to help your weighted average.

 

[MrGecko]

OK then - an Iceberg with style... - I still can't see what's so special about it; there are already systems for VWAP order entry etc... and FIX should be able to handle it if you wanted to do it yourself (which you would if you were regularly doing deals of 200m)...

But if you actually had 200m to unload I suspect you wouldn't need an algo to do it for you...

Am I missing something?

 

[fxmisticat]

hehe... I think your missing the whole point

My desk moves yards in almost every pair in the G10 not that it matters but you're wondering about unloading 200m euros on EBS and see what the market reaction is, that's why the timeslicer was created. Majority of our clients are wanting to use the timeslicer on orders over 30m... i keep a running avg on the bid mid offer and based on market conditions the final fill is sometimes really good or sometimes really bad. Obviously I would like it better than the SMA on Reuters to take more of a vig on the fill.. anyways hope that helps

I thought I would post on here to see if anyone can put in some of the algos they are using into the timeslice..

 

[MrGecko]

well of course you're not going to do it all in one clip, that goes without saying... I would have thought, however, that you could do 30m in an hour without any difficulty at all; infact I'd be blown away of 200m over 30m was an issue...

.. what I don't understand what all the fuss is about / why it took so long?! Splitting larger trades into smaller clips is par for the course... getting a computer to do it for you is still par for the course... and even then, if it's a client order I thought any execution trader could do it without any real difficulty at all.

... moreover, Equities traders have been doing this kind of thing for yonks. Is this kind of thing revolutionary in the FX market?? If so, why hasn't someone build it in-house - i reckon someone who knew what they were doing could knock this up relativley easily using FIX...???

(NB: just curious, but I would have thought that any wholesale desk would have something similar to offer their clients anyway - do they care which liquidity pool it comes from?)

 

[fxmisticat]

well.. you're right this is nothing new. A trader manully doing a timeslicer, yes i have done it a million times, try doing 2 of them at once and picking up the phone for another order.

We already have a timeslicer but I would like it to have more behavioral logic in it for when the markets moving your way or against you or put market condition parameters around it.

And yes putting 200m out in 30min in todays market will push it.

I'm not trying to be rude here but what you're telling me has nothing to do with what I asked. I feel like we are discussing the business need for a "timeslicer" and getting an out of the box solution (which i already wrote). cheers..

 

[MrGecko]

Is liquidity that bad..?? Bl00dy hell...

OK mate, I understand your point - I am merely trying to gain an insight into a side of the business that I'm unlikely to get exposure too, didn't mean to p!ss on your parade.

FWIW, this kind of order can be done relatively easily in most futures markets through an ISV + a bit of jiggery pokery in excel... (and I know you can connect to Hotspot through TT Pro.. currenex / lava et al I'm not sure)

and I think you are unlikely to find anyone to share ideas with when it comes to execution algo's... most here are retail traders spreadBetting the currencies / indices, though I'd be delighted to be proven otherwise (provided you let me interrupt with the odd question )