I personally have NO IDEA why this does happen. Bill actually gives his opinions on it a few posts back. I also don't CARE why it happens. Remember that the orders viewable in the limit order open book can represent stops, targets, scale-ins, who knows what - so I have yet to be able to decipher WHY this happens, but don't take anyones word for it - look at the screenshots Bill presented, then watch it in real time for yourself. Its a fact, believe it or not, at least on ES/ER2/NQ/YM. W
I have been looking into this for a little while and I am in no doubt that the DOM behavior called ACV is quite real on stock index futures. It is not nearly so pronounced on CME FX futures (Why ?). I too would like to understand what is going on here. I suspect that some of the order book volume is not for directional trades and this is one factor. Anyway, attached is a 1min chart of some of todays ESTX50 session that illustrates the DOM behavior. The second from bottom candlestick plot is derived from the ratio of total contracts at the ask at all five levels to total contracts (bid and ask) in the DOM. It is shown as an OHLC series. Because the data is so noisy, a second (blue line) plot shows a smoothed representation. The plot is generated by applying John Ehlers finite impulse response filter to a series of (H+L+C)/3 of the ratio outlined above. The FIR is probably a better smoother than SMA for this and certainly better than an EMA which has too much 'memory'. I have found this to be a very usefull way to look at the DOM.
alright lets wait for Bill then. I would be willing to accept the 2:1 mechanism as a fact, although I would find it hard to trade it without understanding what happens. What bothers me most is the earlier example in Bill's .doc saying "the sellers are trying to stop the downtrend". This implies some logic causal relation, also in Bill's head, that is totally unclear to me. Also, I just have been watching NQ and YM for a bit, and I do see that ratio's are skewed when movements happened. I set Buttontrader so that it signals whenever this happens and it did happen quite a lot in the last half hour. Nice. However, I see no cause/effect relation whatsoever. The movement and the skew happen at the same time, and to me it is evident that this happens because the trades are 'eating in the mountain', causing the eaten side to steeper than the side behind the trades, and this would of course generate a signal. If I would react on the offer:bid being >2:1 by buying I would start buying when the buying already started. I would agree that if a buy movement happens without the confirmation from the 2:1 ratio, then maybe stepping in is not such a good idea. So maybe it is useful as a filter. Still, is Bill, or anyone else able to explain the phenomenon, especially the 'sellers trying to stop the downtrend' bit? Ursa..
Hmm, in order for this to work, Neoticker would have to record every change in the order book. For this it must treat every change in the order book as a tick. My question is, does it actually do this? Or does it treat a trade as a tick, and every change in the order book between trades is missed? I'm curious as to how this is then plotted, be it against volume/tick/or time bars, as there would be a lot more data emanating from the limit order book, requiring a higher temporal resolution. I do not use Neoticker, so for all I know this might be possible - I am merely pointing out a possible flaw in the methodology. Good trading.
is there anymore info i can read about this method? does interactive brokers have the indicator? thanks, bgp
As I said in previous post some of the order book volume may not be from directional traders. Perhaps arbs of various types. Maybe somebody with more knowledge could expand on this or tell me it's a lot of nonsense. That this stuff seems more applicable to SIFs than currency futures is a clue here.
Neoticker records the orderbook as snapshots at a user defined interval down to every 1 second. So it's not every tick, but it gets the job done.
Dcraig, thanks for the chart. THIS is what we need more of. For any NeoTicker users, I have posted instructions on setting up ACV in Neo at http://forums.neoticker.com/showthread.php?t=1618 I have no connection with NeoTicker other than as a happy user. It is *always* helpful to add another data dimension to be able to analyze, and this is a valid and unique data point to utilize. The key issue now is just HOW to use it, how to interpret what it tells us. To just say "I do this to scalp for 2 ticks when the ratio is 2:1" may in fact be valid, but it barely scratches the surface of what COULD be done. Let's keep the ideas coming! W
thx w. and to bill for starting thread. i'm re-reading all posts a few times so i can better understand this concept. bgp