Did any of traders of 6E notice that since last Friday (11/30/2012) there seems to be a new market making algo in action. You must have noticed it if you watch DOM in combination with volume footprint. Around 6:15am CST the algo starts flashing often upwards of 500 contracts on best bid/ask but what's even more interesting is that it actually also executes a lot of volume. During its activity you can see many prints of >= 500 contracts at a single level on a one minute footpring chart (normally, it's quite rare for one minute footprint of 6E to show a single level volume larger than 500 contracts). This algo is completely screwing up one of my volume based signals... as it 'artificially' creates prints that seem significant but they are really not... So far, it's been 4 days in a row that this seems to be happening, let's see if they continue to operate. I will be interested to hear other trades if they noticed it. Attached is a picture of monday's (12/3/2012) with price graph along with best bid/ask size, which clearly shows when the algo started...
I watch footprint chart and know what volume is usually traded at a given price (within a given time frame-in this case one minute time frame). Now, without this algo in action, when you see x-number of contracts traded at a single level where 'x' is sufficiently large (depends on the contract's average liquidity) you can make some judgement as to whether there is a chance of trend reversal etc. What this algo produces is a lot of situations where this volume is very large but it's not caused by 'true' volume coming into market but rather this algo's market making activity.
That algo must be able to hedge his position in the spot market since spot market is much bigger than futures markets for FX. This means if the algo does prove to be profitable, they will not stop the algo - rather it will continue to keep working. You will definitely have to modify your signals in light of this development, though I am not sure what changes you can do.
So the entity which is executing the algo is taking both sides of the trade to make the volume seem to be elevated? Not sure how that's going to make anyone any money, but maybe I'm missing something. I don't think that deploying an algo with the sole purpose of "messing up" someone else's volume-based signals is actually a way to make money. If that's not what's happening and there are trades happening between two parties, I don't understand how that isn't "true" volume.
That volume is true since it actually happened, I probably shouldn't have indicated that I don't consider it 'true'... but it's purpose is to make a tick (or two) at the most and is not saying anything (at least not to me) about longer term movement of the price (as opposed to what I am used to seeing with other volume that is not related to this algo). I suspect they have a model that is good at predicting very short term movement and they get in and out of hundreds of contracts on each side, probably pocketing a fraction of a tick on average. They use the (quite deep) liquidity in futures if one is willing to sacrifice a fraction of a tick vs cash. If they are successful making money with it, they will stay there till it stops working... I have to rethink some of my strategy logic so it can filter it out to the extent possible...
good points all around. Please feel free to share either here or in PM, how you modify your strategy logic to deal with this new situation. We might be able to collaborate on this particular aspect as this is my interest area.
OK, I see a little more what your point is, although I would guess that there are many, many "tick or two"-seeking algos operating in almost every market, although with that type of algo, you better hope you don't experience any slippage. It would be surprising, though, if they really did have a good model for predicting short-term movements, but only operated at one specific time of day. A good model should be more robust than that. Of course, it can't be completely discounted that they are just lazy.
I think that it's just that they started testing it live at a specific time. Usually it goes on for about 30 minutes. Friday they started at 6:06am, Monday 6:16am,Tuesday 6:19am, today (12/5/2012) at about 6:13am.