Although I do not trade the bund outright, I hear people complaining about how it is owned by algos. I suspect that these algos (well, at least the most aggressive are using the order book flow to place false volume, moving their orders to the back of the queue to avoid being hit. In conjunctions with upping the size on they do not want to get executed they ensure that their orders get filled and everyone else playing for small ranges gets forced out.
From this one can deduce that looking at the size will, a lot of the time, give you the wrong signal.
So what can you work out. Well, if they are cycling their size, then when they are about to get hit on a side they do not want, they will move to the back of the queue. They will retain their size on the side they do not want to get hit on.
Option 1, which is too complex to most people would be to analyze the order book like they do, and produce a filtered ladder showing real size and chances of being hit on bid and offer. Effectively filter out the crap to produce a ladder showing real order flow.
Secondly, unless they are being really sophisticated, the changes in bid and offer size on the ladders should give away their intent. The side which is changing the fastest, should indicate the ladder on which the ordered are being cycled, the side on which they do not want to be hit. The order size should not matter too much, but the rate at which it changes should.
Thus, the side with the least volatilty in bid/offer size changes should be the side with the least order flow.
Just an idea, would be interested if anyone thinks there is any value in this.
From this one can deduce that looking at the size will, a lot of the time, give you the wrong signal.
So what can you work out. Well, if they are cycling their size, then when they are about to get hit on a side they do not want, they will move to the back of the queue. They will retain their size on the side they do not want to get hit on.
Option 1, which is too complex to most people would be to analyze the order book like they do, and produce a filtered ladder showing real size and chances of being hit on bid and offer. Effectively filter out the crap to produce a ladder showing real order flow.
Secondly, unless they are being really sophisticated, the changes in bid and offer size on the ladders should give away their intent. The side which is changing the fastest, should indicate the ladder on which the ordered are being cycled, the side on which they do not want to be hit. The order size should not matter too much, but the rate at which it changes should.
Thus, the side with the least volatilty in bid/offer size changes should be the side with the least order flow.
Just an idea, would be interested if anyone thinks there is any value in this.