a little trading thought I've had

[arabianights]

Ok, so conservative estimate (and I know it does a lot more), assume FTSE does 100k lots in a day.

And say it has a 100 tick range usually.

And average trade is 10 lots, say (again an overestimate)

So we have 10 000 moves to make 100 ticks (and that's if it begins and ends on the extremes)

So in theory every move is 1/100th of a tick.

So make a 1 tick market consistently and we should have a 99% edge, right?

 

[arabianights]

so, then we notice a flaw in this theory, best illustrated by an extreme example: all trades are lifts of offers.

Leaving us 100 000 lots short, an average of 50 ticks offside.

So how to refine this?

Simple, if we fancy it up, we make a bid constantly, so long as we fancy it up. This way we capture the spread and get the right direction. So we have 199% edge.

I hope this illustrates something.

 

[arabianights]

But then we notice another problem, we have far too big a postion, even if it is mightily onside, to exit.

So what can we do about this?

Simple: We only bid some of the time, so we only get a size we're comfy with, and we do that by feeding the order in slowly and measured... and we do more at lower levels.

 

[arabianights]

And that, ladies and gentlemen, is why averaging is so powerful when done properly

 

[elitejets]

Very neat. Even I understood that.

 

[GammaJammer]

One of the more common algorithms in use these days in institutional circles is an accumulator type, where if you're trying to buy some stock (or euros, or whatever), your bid is 'pegged' to the top of the book, so you're always there or thereabouts, but not in market dwarfing size. There are all sorts of different ways to splice and dice this, different parameters for how aggressive you are etc, but in basic terms a lot of people are already doing this, just not necessarily for the reasons you're stating.

One offshoot of this however, is that these algos can kind of end up chasing each other round in circles a bit at times. That definitely happens in crosses in FX.

 

[arabianights]

Yup, I've seen them quite a bit in short sterling, quite useful as it happens

 

[GammaJammer]

Yeah totally - discernable patterns are a good thing.

You can sometimes see them in FX when people are trying to leg into crosses. Happens fairly often in cable / eurusd (people trying executing size in eurgbp). Personally I always find that one a bit surprising. cable liquidity isn't great these days, and eurgbp is more solid, but nevertheless it still happens.