uexkuell

May I just ask two questions (if it is ok to be curious): - Which is the science you are familiar with? - What's the reason for heading towards 100% automation?

In my view this is a very important point that makes a difference between traders - staying alert, watching what happens. I think it is so important, because the information you can get through watching does not later appear in a conventional chart! If you do not watch it while it happens there is no way to see it later on the chart. On the other hand this is a close to impossible task: Only a few will be able to stay alert for hours, watch every little move and recognize what it means. There is only a minor number of variables in any data stream: - price, time, volume - with regard to either trade happened or order level in the book Putting as much of this information together into a graphical presentation in a way that you can still digest makes a difference.

Sorry but you are missing the point. The quote to which the reference was made: So it is not about showing bid/ask at some moment which many charting software does. It is more about defining a distinct kind of charts that is different from the well-known types that take their basis from time, tick count, range or overall trade volume. For all kinds of charts there must be a point when a new bar (or candle) starts. For time based that is after some time (e.g. 5 min), tick based some number of trade ticks (e.g. 100) and so on. I understand the posting so that it is (theoretically) possible to start a new bar dependent on how much volume was traded at bid or ask. The reaction of AgeKay to my request for further information makes me believe that he wrote his post rather theoretical - perhaps inspired by the other posting also. Two things make me doubt that such charts exist or are of value: - I cannot see depending on which condition exactly a new bar should be started - Trades at bid and ask are highly manipulated. If you look even at a high-speed data stream you find that quite often there are subsequent trades that come in so fast that the quote source has not enough time to adjust bid or ask. Therefore it is easily possible that a trade that happened at bid is seen as a trade that happened at ask and vice versa. The manipulation/disguise makes sense because if just from "hammering the ask with trades" you could know with a high chance that price is/will be going up it would be much too simple to make money from this.

Certainly an important point and not a begginers question in my view. Quite frequently you can see price move with only very few volume (very rarely even without volume - this only in a not so liquid instrument when heavy news are appearing). Anyway moves on low volume are to be noticed. Unfortunately low volume moves can mean very different things and that's why people get confused using volume (and sometimes even refuse to use it). Just as in your example. The person that refuses to make the deal can have completely different reasons for doing so: 1 "Commodity A is so valuable, I won't give it away for cheap." 2 "This guy desperately wants A, let's see, how far he will go! The more I wait the more anxious he will become to get it." 1 Is at the beginning of a move 2 Is at the end Isn't this the same? No, because "the end" may be quite some time and $ away. If you ever entered into a trade at the last exhaustion spike and the spike extended for some time and hit your stop loss only to turn some time after you will know that mentally it is completely different.

May I ask which quote feed you are using? Big boys have sophisticated infrastructure at their hands that allows them to disguise whether they traded at bid or ask, especially (but not necessarily) if you look at it with a feed that sometimes lags. Just imagine that their software can shoot impulses of trades that are coming in so fast that the quote feed cannot adjust the dom stack (especially level 0 - bid and ask) fast enough. This way every software or trader that relies on bid/ask-relations can get highly confused. Perhaps also reasons for strange observations: - Rollover close for ESTX (block trades happening) - Some major change in the market on the way

That is indeed quite interesting. So what do you think this means? Are they just simulating the trades somewhat like a CFD-shop?

Could you please specify what interesting data you found / might be worth looking for (service, commissions..)?

Brownsfan has made already all the important points. I'd like to repeat only a few things in a less technical way: For me it made a big difference to look at the charts with more common sense. If there was a big move upwards the normal thing would be to stay there for some time. If a big downmove is following this would mean: Many people come to realize that they got lured into buying but there is no substance to it so better get out quick. This is not the normal thing, you should only think this situation is going on when you have strong hints that it is so. The big boys are hard working on their screens. Therefore at lunchtime they get hungry and normally market activity stalls. If they continue their trading up to 12:40 they must have a good reason - probably news that have to be priced in. But surely at 12:40 they think: Enough is enough. And they don't want to engage in a new battle downward. Some technical thing: "Indicators" are generally not so very helpful, volume is (would show you the stalling). To make it short: When you want to trade trends try to start from some quiet region.

If you want an "indicator" that can only vary within a certain range (e.g. 0-100) you can always use the stochastic formula as a basis. As input you would use e.g. the put/call ratio or the tick value (instead of price, as usually done). But: Please think if it is wise to do so. In my view it is important to get as much information as you can from the data you have access to. If you notice that P/C ratio is "leaning" then this is valuable information which should not be eliminated by a formula that forces values into an artificial range. For me it is clearly the task to get as much (valuable) information as possible not to cut it down.

If it is consistent, why not enter short then?

The absolutely central question is: What are you intending to use your code for? The answer to "where to start" will vary strongly depending on your intention. There are several main directions where programming in connection with trading might go: A Statistical evaluations (means, regressions, time series analysis) B "Old-style" indicators (like variations of MAs, Stoch etc., using price and/or volume to compute new values. C Extracting new / additional information from high-speed/high-bandwidth datastreams (like CQG), doing computations on bid/ask/trades, Dom-data etc. Please add more that I forgot. Since you said you want to interface to CQG I guess you want to go for the more innovative/high-speed things. It can be useful to consider several parameters of programming languages: - Speed of execution - Learning curve for the user - Support for the user (environment supporting the development, language helps with error checking, debugging helps, language straightforward) - Mathematical / statistical analysis supported - Can try truely innovative approaches - Forums and example programs available Some people recommended the use of trading-platform dependent languages like EasyLanguage. This restricts the user to the possibilities of the platform. There are always big limitations if the program runs in a "sandbox". Also there may be some day where suddenly there is no more access to the platform (company shuts down). E.g. many OpenTick users are looking in the net for new homes. This approach is mainly suited if you want to go for "old-style" indicators. Speed -, Learning+, Support+, Math+/-, Innovations-- C++ Sure it is the fastest high-level programming language out there and you can nearly do anything with it. But WARNING it has a very steep learning curve (some of the posters already admitted that he never got beyond "hello world" - this is just the usual outcome for newbies & C++). Also there are tons of traps hidden in the language and the user has to do many things (checking) that in other languages the compiler does for him. Many examples how you change your code just very slightly and it means something completely different. Good news: You don't need C++: With modern fast computers mostly you do not need the ultimate speed, fast is good enough. And: When dealing with price, volume and so on you do not need complex language features. Speed++, Learning--, Support--, Math +/-, Innovations++ C# Is very different from C++! It does error checking, it is much more straightforward in the language construction. C# is pretty much the same as Visual Basic. You can just go to MSDN and find hundreds of pages where you see that the language constructs are just a bit different (between C# and VB), the essence is the same. Example: BitConverter Class (System) Speed +, Learning+, Support+, Math+/-, Innovations++ Visual Basic (VB) In my view better to read (Yes, I admit - I started with Pascal 25 years ago and still like it) and more free support out in the net. Otherwise with only few exceptions like C#. Modern processors in connection with VB can easily keep up with a real-time data stream at 100kB/sec and still do quite funny things. Speed +, Learning+, Support+, Math+/-, Innovations++ Java Learning curve pretty much like C# or VB. Additional advantage of being able to let programs run on different systems (linux). Personally I prefer anyway the (free) development platforms from Microsoft. They support the user in many ways. Speed +, Learning+, Support+, Math+/-, Innovations++ Python Can be learned pretty fast, intelligent language. But relatively slow, not so much support. May be hard to connect to some data provider API (because you have to write an adaptor in the first place). Speed -, Learning+, Support +/-, Math+/-, Innovations- Mathematical languages (MatLab, R..) The language is usually quite easy to learn, the mathematics behind you must know before! You are limited to quant-type strategies. Speed -, Learning+, Support+, Math++, Innovations- Excel Dead slow. Very easy to learn in the first place. If you try to do something a little complex you end up with unreadable code. But good for a quick look at numbers. Speed --, Learning++, Support+, Math+, Innovations-- Quote from CQG site: "Samples of how to use the CQG API are provided for C#, C++, MatLab, Microsoft Excel VBA, and Visual Basic 6.0." In this case just follow the precedence CQG has already implied: C#! Btw if they say C# surely VB is possible also but I don't know if they have sample code. Sorry for getting lenghty. It is a complex question.

That sounds convincing and interesting. Which software would do this? Could you please post a chart?

Hi shortbleu, thank you for appreciating my words. Extremely helpful to know where one is standing. If you never traded some of these (DAX, ESTX) before live I strongly recommend papertrading them until you get profitable at least for a month pretty much every day (if intraday-trading which you seem to be up to). If starting to trade right away without sufficient training unfortunately loss of precious liquidity is sure. When you papertrade you can just go right for the DAX it sure is more fun than the slower-moving things. And if you succeed there nothing else is going to scare you any more. After papertrading for a sufficient time-span you by yourself will exactly know, what trading style suits you best, which instrument, how much margin you need, what your odds are and many more things. Nobody else can tell you because they don't know how you trade and how your personality is structured. Yes, sure, for me BUND is boring too. FESX and DAX are mostly absolutely parallel. Exception are spikes that happen in the DAX sometimes and take it away from the mainstream in seconds. If you don't have yet so much liquidity and experience these can make you behave irrational.

Ok, sounds as if you found your personal style and what you do not want, which is very important. This is unfortunately incompatible with your first statement cited above: If you prepare yourself to go for 5 points this means, that you are heading towards short term daytrading. In these days DAX makes sometimes 200 points a day easily. Once again incompatible with your opening statement. If you go for (large) interday moves commissions will only be peanuts for you. There are other things that have higher priority. Alltogether: DAX is a wild beast that will rip you apart in seconds if you do not know what you do. Therefore it is essential to know what kind of trading style suits you well and where you are profitable. If you don't have quite deep pockets DAX is nowadays only for short term trading (seconds - 3 hours) and you have to be ready to take actions immediately. Otherwise (as said above) ESTX is perhaps more suitable.

Can only underscore this. "Smart money" uses quite sophisticated software for working their orders. They do not just hit the "buy 200" order button. Orders of big size show that the buyers are more in the category "big moose, but (at most) medium sophistication". Therefore, their importance in determining important market turning points is not really that big. Real smart money has btw already gone beyond pulsing orders.

What does it mean to "exit the call leg of a straddle"? Am I right that this means just keeping the puts? If you do that you change your mind from expecting a big move in any direction to betting that the underlying will be loosing ground?

Thanks for bringing this to our focus. Interesting approach that they use, pretty clever. They reward people that make good prognoses. The better you get the more the system provides you with better information. But: If you don't (yet) have an edge by yourself the thing is worthless because you get delayed information. If you make good prognoses it would really be stupid to give them away to them because you can be 100% sure that they will take advantage of it thereby ruining your edge. And after your edge is gone (the market always adapts if you use a strategy on large scale) they will just move to watch the next user who has discovered a working strategy. Guess they thought about that too. So they know they will never attract people that already have a special edge. They probably count on statistics: Weighting the opinions of many mediocre traders that are just good enough (or consistably bad - special jewels) shall give them an advantage. For that reason you can be sure that they will give you information that is less valuable to you than the information that you provide to them. So it might be better to spend time on searching for a real edge. The idea of looking for the commitments of other traders and where they enter into trades is a good starting point.

I have done quite some programming with the IB API. My rough estimation about your goals (in lines of code) would be: 1) historical data: >1500 (!!) - if done correctly realtime 200 2) 200 3) ??? depends on the strategy 4) 50 5) 50 6) market: 50 limit: depends on how to compute it - perhaps >200 In my view this is a major project and many unexpected problems will arise from the API. Any program that gives some support would save lots of time. Therefore Tradestation or Multicharts would certainly be a good framework especially because they can handle the historical backfill thing.

This can save you much money: Start paper trading and do it until you are consistably profitable over a longer time period (if daytrading say 2 weeks in a row (that's still short), if interday trading much longer). When you have found a way of becoming profitable on paper you will know EXACTLY how much downside you have to expect from your strategy. Then you can easily answer your question by yourself. As long as you cannot answer the question by yourself this is a failsafe indicator that you better stay away from real money.

Why not have both? Several reasons: - Sometimes you may be very glad to have a backup-broker. Just imagine you have a big open futures position with one of them and the broker's connection goes down (not so rare with IB, can't tell about Amp). You may then hedge with a position on the other. - The wide-spread possibilities of IB may be an advantage in some special situations. - Better data feed is a plus for Amp.

There will for sure be no way to directly connect to IB without TWS because that would mean that their security is broken. You need password and token to enter their system. After this the data is SSL-encoded. Please specify what it is exactly that you want to do, perhaps I could tell you more about bypassing TWS. If it is the idea of receiving IB data without having an account with them - impossible.

Just my thoughts: The kind of charts you see mostly are a graphical presentation of three kinds of information: price, time, possibly volume. Think about how to present best this data in just one chart (I count a chart with an "indicator"-window below as two charts) you get into trouble because you normally loose something. In my view, this "one chart" thing is quite important, because for our small brains it is very difficult to look at one picture, keep some piece of information in mind (eg volume) then look at another one (eg time & price) and set the two into a relation. Constant time charts: At times with few trades you get strong distortions of your picture, especially your trendlines (mainly with futures that are trading around the clock). So constant volume or tick charts are much better for that. (whereas tick charts are no good for Interactive Brokers users because you cannot get appropriate backfill.) Constant volume charts: You loose the time information - it cannot easily be seen how long a movement took which is sometimes surely valuable. To answer your question: I think with range charts you loose the time and the volume information, only price is left. Range charts do not have a constant x-axis spacing relating to time or volume so both these informations cannot be seen with one look (though you may get them back through some indicators you define, but then it is not just in one picture). For that same reason trendlines that may be very useful for your trading cannot be drawn correctly sometimes. On the other hand, this is only theoretical thinking. I just used range charts on some of the instruments I am trading and there I see that trendlines drawn on the basis of range charts "explain" some turning points very nicely whereas other turning points fit much better into trendlines drawn on volume charts. I guess that comes just from the fact that there are traders out there who use tick or volume charts and others use range charts. In an ideal setup you would probably watch constant volume and constant range charts in parallel.

Right. German bond futures are relatively good opposite indicators for FDAX. May I just add that FGBL (Bund) is moving only in big steps and thus does not give so clear signals. You may consider to add FGBS (Schatz) or - best - FGBM (Bobl) to your watchlist. The latter is moving quite smoothly and has the biggest range (in ticks) of the 3.

Some confusion arising here about tick and historical data download from IB. There are 3 different kind of quote data available form IB: 1. Tick data (realtime) is coming (as stated) in 0.2 to 0.3 sec intervals. It is mostly truly realtime but it can (rarely) be lagging max. a few seconds. It comes as price (of trade) and volume information. If any trades at price levels away from the given price took place you will never know (this could happen only for very liquid instruments). Good thing is you never loose track of volume traded. 2. Historical data (to be downloaded). It can be ordered in various granularities (1 sec=minimum NOT 5 sec as was stated, 5 sec, 10 sec ... 1 day). The data comes as: Open, High, Low, Close, Volume (OHLC). In one data request you can get (depending on interval chosen) about 2000 data points (33.3 minutes if 1 sec interval chosen). Data download is throttled: If you make several data requests, you can only have one data request every 11 secs or so. That makes downloading of 1 sec data of a longer time interval pretty slow. 3. Realtime Bars (realtime at 5 sec intervals, OHLC bars). Currently it is only possible to get 5 sec realtime bars. Data comes (as with historicals) as Open, High, Low, Close, Volume. The advantage over tick data is that you can be sure that you will see if a new High or Low has been made. Tick data and Realtime bars from one symbol can be ordered in parallel. Main question is in what kind of program you want to use the data. If you want to make your own program, IB will be fine. I am not aware of any commercially available program that could use all the possibilities that the different IB data formats give you. Program to download data from IB in text form: http://home.cidadevirtual.pt/jtwsdump/ (I am not the programmer and do not use it).