You're done.
The Ultimate DAX Scalping Discussion Thread!
- Thread starter Scientist
- Start date
You're done.
What precise ratio would light your fire? Would it be 1.5 or would it be higher? Do you require that another symbol also have the same ratio at the same time? e.g. ES Asks are 1.5x larger than bids, and simultaneously likewise in NQ = Buy the Dax?
I have a mental block when it comes to such graphical displays as you know. What about this as a simpler idea: Take your threshold ratio (say 1.5 or whatever) and monitor the next 3-5 ratios that come in ...if the next 3-5 ratios are succesively higher e.g it goes 1.5, 1.55, 1.6, 1.7 ...then auto enter a market order (or set up something flashing). Then again, maybe the ratios don't need to be succesively higher, maybe the next 3-5 numbers just need to be higher than the threshold trigger, so 1.5, 1.7, 1.6, 1.6 ...would trigger a flashing light? What specific paramaters would you like to see?
What version of the program will that be on? Maybe he'll like my successive numbers idea for a flashing light and put it in the program, then give me a big fat wet kiss to say thanks - you know he loves me really
As a variation on a theme, what about Last Traded Size as a concept to form a ratio off? Is the following of relevance in your opinion or is it too historial already:
Suppose you've got ES staying at 1025.00 BID and 1025.25 ASK for one minute alternating between the two values. Until such time as we get either a new Bid or new Offer (when the slate wipes clean), we keep a running total of the total number of contracts traded at the Bid and a running total traded at the Ask, and express them both as a % or Ratio of the total contracts traded at either price. So, as the market ticks betwen the two values, we may reach a point when we see that 1.5 times more contracts have now traded at the Bid, and maybe it's time to Sell?
Relevant, or past news already?
Hi All,
Now that we are in the midst of throwing numbers around and (perhaps timidly) advancing decision mechanisms based on these, let's recall that any kind of knowledge on this field is predicated onbeing able to express in numbers what you are talking about. This goes both ways though:
(1) Don't talk about things that you can't express in numbers;
(2) Don't talk about numbers that you don't know what they relate to.
Both these constitute your entry ticket into what I have called already several times "scientism" (having picked up the term from some very smart troll).
Did anybody among the followers of this thread already try to collect realtime time series (data) about DOM? Anybody heard about sombody having done this. Would this be of any use? If so what did people discover?
As for myself, I already have thought about this, but I admit that I am not too sure yet what I would do with these to distill some little money recipie out of this. In fact, I have discussed this problem recently with some specialists in automatic process control identification (modelling) who seem to confirm the very complex nature of all this.
A casual reader must admit that over the months, you are confronted with quite some halfbaked "authoritive" sounding statements on this. Several threads trying to pull people together on this did not create much clarity. In fact another set of guys have a thread going right now on level II in which you encounter babble about esoteric math technology.
Little nononsense , not having put a foot on Mars yet, is simply trying to get a bit wiser about the whole thing.
Be good you all,
nononsense
Now that we are in the midst of throwing numbers around and (perhaps timidly) advancing decision mechanisms based on these, let's recall that any kind of knowledge on this field is predicated onbeing able to express in numbers what you are talking about. This goes both ways though:
(1) Don't talk about things that you can't express in numbers;
(2) Don't talk about numbers that you don't know what they relate to.
Both these constitute your entry ticket into what I have called already several times "scientism" (having picked up the term from some very smart troll).
Did anybody among the followers of this thread already try to collect realtime time series (data) about DOM? Anybody heard about sombody having done this. Would this be of any use? If so what did people discover?
As for myself, I already have thought about this, but I admit that I am not too sure yet what I would do with these to distill some little money recipie out of this. In fact, I have discussed this problem recently with some specialists in automatic process control identification (modelling) who seem to confirm the very complex nature of all this.
A casual reader must admit that over the months, you are confronted with quite some halfbaked "authoritive" sounding statements on this. Several threads trying to pull people together on this did not create much clarity. In fact another set of guys have a thread going right now on level II in which you encounter babble about esoteric math technology.
Little nononsense , not having put a foot on Mars yet, is simply trying to get a bit wiser about the whole thing.
Be good you all,
nononsense
I think there is merit to the Scientist's basic concept of measuring the buying volume vs the selling volume and trading as a contrarian. How one uses that information may vary from one person to another. Scientist may scalp with it. I might just want to put it as a filter on a stop order at a price I would trade at for other reasons. It is important to define what basic numbers (or ratios) may be useful, and that's whay I'm hoping for clarity because I also can't be dealing with the complex algorithm theories that sometimes these threads degenerate into.
There's merit here. I just need to JRify it for my simpleton's brain(!) while maybe offering a couple of thoughts along the way to not just be a taker. The good news is that while Scientist & I disagree about Jack ('Grob' to the uninformed) i.e. I find him a babbling fool 100% of the time ....he still asked Jack to take his thread-destroying babble somewhere else
That person was likely me! I do see it as an issue that not 100% of anything (last prices, bid & ask sizes etc.) are coming through. I wish to god one of these developers would make their program work with a good tick-by-tick datafeed. There is probably still some use to this analysis with the fast IB snapshot feed with the logic that it's better than nothing.
The area that concerns me (if we want to add up the sum of the five bid sizes for example): how would the developer cope with the scenario of levels 1,2, 4 and 5 changing while level 3 does not. Do we just add the old/existing level 3 size into the new total?
Quote from John Lydon:
I think there is merit to the Scientist's basic concept of measuring the buying volume vs the selling volume and trading as a contrarian. How one uses that information may vary from one person to another. Scientist may scalp with it. I might just want to put it as a filter on a stop order at a price I would trade at for other reasons. It is important to define what basic numbers (or ratios) may be useful, and that's whay I'm hoping for clarity because I also can't be dealing with the complex algorithm theories that sometimes these threads degenerate into.
There's merit here. I just need to JRify it for my simpleton's brain(!) while maybe offering a couple of thoughts along the way to not just be a taker. The good news is that while Scientist & I disagree about Jack ('Grob' to the uninformed) i.e. I find him a babbling fool 100% of the time ....he still asked Jack to take his thread-destroying babble somewhere else
John,
Of course there is merit to "measuring the buying volume vs the selling volume". A lot of theories are based on this.
However, John, coming to our DOM and level II data coming out of the feeds, tell me, how do you extract "buying volume" and how do you extract "selling volume". These terms are even vaporous. How do you define buying and selling volume? Do we know what we are talking about? This is the question that I am struggling with. This is what I referred to in my previous post.
A lot of advice has been offered by the many, often contradictory, on DOM and level II usage. Let some genius come forward to explain this to the ET people. I am listening. Of course John, there is lots of merit to all this. Like you, I happen to be only endowed with a "simpleton's brain" as you put it. But don't forget, true knowledge only germinates in simpleton's brains! People like you and me have to continuously look over our shoulders so as not to be ensnared by the siren songs of the more enlightedned ones.
As to your skepticism towards some software, if I understand you correctly, I myself, I can't program anything unless I posses perfectly unambiguous definitions and understanding about what I am doing. Showing people some flashing height bars and colored widgets in some way derrived from DOM data is to me mumbo jumbo. Some people call this also spaghetti. This I can easily do, but I fail to see how I can make money with this, besides selling it to "true believers"of course.
nononsense
No precise ratio. It depends on the overall market situation and the correlative depth situation on other indices. As said, it's discretionary. It can be roughly quantified, but only roughly, because there are many situations, times of day, and market velocity states.
JR, I don't like the "market order" idea. Please remember that cumulative depth is only a guideline - another tool in your box. That's all. You can use it as a filter or determinant, but for godsake not to fire trades! Also, please consider that cumulative depth often shifts according to price change. If price moves up 5 levels, cumulative could be decidingly different. So no, this really won't work. Of course you are very welcome to email the people at ButtonTrader with your ideas. It's up to you.I have a mental block when it comes to such graphical displays as you know. What about this as a simpler idea: Take your threshold ratio (say 1.5 or whatever) and monitor the next 3-5 ratios that come in ...if the next 3-5 ratios are succesively higher e.g it goes 1.5, 1.55, 1.6, 1.7 ...then auto enter a market order (or set up something flashing). Then again, maybe the ratios don't need to be succesively higher, maybe the next 3-5 numbers just need to be higher than the threshold trigger, so 1.5, 1.7, 1.6, 1.6 ...would trigger a flashing light? What specific paramaters would you like to see?
What version of the program will that be on? Maybe he'll like my successive numbers idea for a flashing light and put it in the program, then give me a big fat wet kiss to say thanks - you know he loves me really
If you mean following last displayed size on depth and volume memory, that is exactly what ButtonTrader already does. And so does X_Trader and any other advanced discretionary trading platform.
Please remember that all this does involve discretion to be refined enough to extract money with high performance. From a rough estimate I'd say you'd need the analytical power of several beowulf clusters of Cray2's just to roughly approach some basic simplifications of the discretionary power of the human brain as applied to trading, particularly scalping. I don't care what the advantages are on computers being void of emotion: Computers to date remain dumber than any living creature on earth, and superior trading, IMHO, requires some intelligence. Computers have none. Note that we're dealing mostly with people and emotions in the markets, not with machines. Machines can not quantify emotion and profit from it.
Quite frankly, I agree with velocity_trader on this: I don't think that it's possible to program this into a computer.
But I welcome any attempt to prove me wrong.
You don't need to extract anything. All you do is add up the cumulative volume on each 5 levels and weight your bias according to the cumulative volume ratio. If the cumulative volume is 7,000 contracts on one side and 4,000 contracts on the other, then you have a situation. It shouldn't even be hard to measure with a computer program that monitors it. You can't use it as a sole entry determinant, but you could indeed use it as a filter, at least for "do not trade here" decisions, to increase your relative hit rate for a mechanical system. Considering the probable ease of implementing this, it might well be worth the programming. If the ButtonTrader dude can do it, so can you...
Best Regards and Good Luck
Right now I'm interested in adding up the five Bid levels, adding up the five ask levels and having each one espressed as a ratio of the other. It's hard and it's quantifiable. If the Bids total 1500 contracts and the Asks 1000, the ratio will be 1.5. I'd be interested in exploring this further to assess it's usefulness, so I'm looking forward to Scientist's reply to my earlier questions and thoughts. I too can only deal with well-defined stuff.
John, please realize that it's got nothing whatsoever to do with the developers. The snapshot feeds come this way from the broker (IB), not the platforms. No developer can do anything about it.
And if you go to blame the broker now, you can forget that, too. The broker can't do anything about it, either, because they receive their snapshots from Eurex, and Eurex, like any other exchange, only delivers data in snapshots. I do not see how it would be possible to deliver any amount of depth data refreshes on a per-change basis the way you describe it, because:
1) It would take too much bandwidth (what, 30Mbps connection?)
2) The ping/pong to the exchange couldn't be fast enough so why bother
3) Who needs to see every piker's contract subbed' and added to the depth anyway?
The snapshot system works fine the way it is. Get used to it. I hope that's the end of this discussion.
Yes.