NetTecture

I just have to answer here. Seriously, get a little grip on reality. Trading is great, but a LITTLE more thinking is required. > What is the best website to start trading ? THE BEST? Google.com. There are a lot of ressources, but none is THE BEST. > What is the best website or place to learn trading ? Place? Your home, in your chair. Website? Google.com. SOME groundwork is needed from you, for example.... > What is the best method of trading ? Penny stocks , day trading ? Question, what is better - steak or pasta? A LOT depeneds on your style and priorities. For example, I trade no stocks - futures and forex only. > What is the process of trading ? do i use a broker or do i do it myself ? Question - how do you do it yourself? Standing in front of your house and trying to buy and sell stocks from people passing by? Because YOU can not go to an exchange without buying an exchange seat. Handling the orders is what a broker does. So, what is the question here? > And what website provide these services? None. See, websites offer no services. This is like saying "i want to eat pizza, which flyer offers these services": A website is made by a company to advertise and facilitate it's business - and the COMPANY provides the services. Which one? Depends what you want to start trading. There are literally hundreds or thousands of brokers in different countries. > Where do i learn to read charts ? We are back to your chair in front of your desk. Tons of approaches, tons of material on google. Find an approach and then - well, books, but mostly sitting and trying to work things out. Given your - hm - level if common sense demonstrated (by basically asking questions to generic that "google" is the only answer), I would suggest you open a tiny account with a decent reputable forex broker and walk through their introduction material. Oanda is good as you can trade super small sizes. This place here is too. And they have a policy about not advertising other websites, you know. Likely you never read the material you signed when signing up here? Now, let me be straight. Welcome here. And trading is great. But do not expect riches fast - it is a long road and it requires hard work from your side. Finding out what sources are good is part of the education.

The answer: ALL OF THEM. SImple like that. The answer as useless as the question. Because "will allow" does not mean efficient / cost effective. You need to do a lot more homework - and use your brain a little more - before even considering doing testing. For example you name a programming language (Python). As someone who has invested maybe 5-6 man years in developing the infrastructure to do what you want to do with C# - let me tell you that the language is irrelevant. As in Either you go with an existing framework, or the work you do will be LARGE; regardless of the language. Just visualizing and analyzing backtest data will take a lot of time. Writing an testing every single indicator you plan to use (and you will use a lot more than you have now) will take time. Writing a good exchange simulator will take time. Unless you have a SERIOUS programming background (which I assume you have not given the way the question is formulated) you have like zero chance to finish this to a usable state. So you are better of doing with a higher level / trading ready framework like NinjaTrader (as crapy as that is in parts)... simply because the alternative is "not finishing anything".

Reading. Before going into a tangent that totally will not work, read Robert Pardos book The Evaluation and Optimization of Trading Strategies. Too many people think they know what they do in a backtest and then are shocked that the real world is different. Also if you think of using Renko charts read this blog post about backtests and renko charts - and make sure your simulator handles them better (or get better renko that is appropriate for backtesting) - otherwise you basically waste your time. It does not start with throwing some ideas into a program and looking at the results - not if you want to make money. In this case, reading up on what you actually do in a backtest is not a bad approach.

It is possible - even "supported" - but a lot of work to replace the whole feed connector: Trade-Robots lower than that - I am not sure. I would contact MbTrading about this. If they support NinjaTrader as platform, it is their duty to go and make it work properly. There is no known / documented way to intercept the data without the full "custom feed" solution I have outlined in that blog post.

Define what you need. Help formulating? Coding? Syntax questions? Analysing why the numbers are not what you expect after a backtest? If that is a "hey, do all for me" - do you have a budget for someone doing your work?

Before you do any backtests I suggest you head over to Trade-Robots - Contents tagged with ninjatrader - Tags Most threads we have tagged for ninja refer to critical bugs in the backtesting. From ignoring configured slippage to returning random results on backtests. All of that can be dealt with - but I would assume 99% of the people are not even aware of those critical bugs in NinjaTrader, so they never look for the solutions (that area found via google). As few people seem to validate that the trading results are correct.... guess what

Before you even consider that this may be a good system head over to Trade-Robots - Renko and Backtests - it can be a very wrong result and check whether you fall into the same situation of "oh god, my results are all wrong because my backtest simulator and renko do not work together well". As far as I know MT4 does not to tick replay so - all results from renko are crap.

No, you will sort of have to do some research yourself. I have my own infrastructure developped and we did not really complete that bridge to a level I am willing to share. All the info is there - as I said, a month, maybe two, for a GOOD programmer. Sit down and do it, or hire someone to do it.

Yes you can. Even today. All you need is a bridge and NT thinking it talks to - TT. Trading Technology uses FIX and that is well documented. Not totally easy but a nice month project for an experienced programmer.

Yes, crazy, but they are essentially random. Days will be missing in backtests without a warning, as it seems. My team has documented the behavior at http://www.trade-robots.com/blog/ninjatrader-so-good-that-backtest-results-are-random There are sitautions where you can execute the same backtest on 3 machines and get different results. Results that even differ between executions. Same strategy, same settings, same timeframe, 2 backtest = 2 results. For anyone trying to come up with a good strategy this is a critical bug.

That is quite serious. I have a large UPS for now (large - 10kw output for an hour), do regular backups. Just sometimes is strikes. Preparation is everything. Desaster happened to me yesterday, all except possibly some emails in a spool was back this morning. Just when I wrote i used that for my first blog post on my own new trading side - Trade-Robots - Technical Risk. Are you prepared for a disaster? Thank heaven I start being paranoid by now - all servers not at home are backed up every 30 seconds Having good backups saved me from a lot of problems this time. Not the first time a RAID array dies

Do you guys run serious protection for your infrastructure? I was reconfiguring some storage yesterday preparing for some discs upgrades (and I did not have slots on that server for more discs, so I decided to disable the SSD cache). The 50gb data write dump - triggered 3 disc failures within 5 minutes, taking a Raid 6 down and destroying more than 1tb on data, mostly virtual machines. Pretty much - outside some large data stores (tick data, sql for analysis) all my small trading operation consists of. Thank heaven I take backups 2x per day, every 15 minutes for very time critical things (like emails) and we had the new discs (actually SSD) already ready. But that makes me wonder - how much are you guys preparing for contingencies? My background is IT, running a small IT company (not focusing on trading), so I have the whole thing totally set up (including replicating my trading machines down to my office every 30 seconds). Most traders I talk to are ignorant in this area (that reads down to "Hah? Backup? I have a good disc"). How do you deal with power outages? Internet breakdowns?

Try MIkrotik. They combine multiple links whatever you want (not being a modem, though) with a lot of high end features if they ever come in handy. Need to know what you are doing, though (i.e. not ane end user stuff). I use them to link 2-3 locations to a backbone in a data center with MPLS fast routing, load balancing over multiple connections (currently 3 in the head office where I also trade from), full quality of service and central management The price for the item we use int the head office was below 150 USD (450G model), the data center core uses an 1100AH for around 800 USD.

It is called brain usage. Any non-normal average says so. THis is why the "moving average" is the normal one, and guess what - the "exponential moving average" is - the exponential one. Normal means normal. Anything else is not normal. Some people think too complex and fail to see the obvious, and obviously the OP belongs into that category.

With all respect... an exponential moving average would NOT result in an ATR. Period. The "Average" is mathematically well defined. Same with a normal moving average - which is why the exponential moving average has a different name. Assuming the ATR is not a normal average is a little - sorry - dump. It says clearly it is the AVERAGE of the true range, which definitely IS officially defined.

Davia... You are SO under the wrong impression about me Hehe. * There is no need to form an LLC or something I operate an IT company that just opens a trading branch, hopefully moving into that direction. * Similar the cost of data for me is zero I only collect data I want to have anyway I strife into system development, and full tick data is a necessity. I basically need them for backtesting And as I do not know what I Will trade next month.... I collect complete exchanges. I have ot say I am not so interested in... hm.... open source per se. This is data collection for an enterprise. Being someone else open source arm would mean a commitment I am not sure I want to make at that point Plus, seriously.... the code is more trivial. The main problem is handling the data I am afraid I Will need a new server for the data - with another 24 discs or so Next year The real deal would be sharing the data - and that would not be free. Too much investment needed for that This is a tremendous amount of stuff going through the system Non-Redistribution is basically always for real time - and in that case... I will adhere. Point. An LLC would not protect against that anyway.... but I Think non-real-time is really valuable enough for many people for backtesting My timeframe is starting with CME and possibly FOREX 1st of October. We have some performance issues with our data collector so far in the provider code that works great under intel, but breas down on AMD processors. I hope they fix it next week It sucks if a CPU core (and their network code is inherently single threaded - which makes sense, as long as you only have one connection for one login) breaks down and uses way too much CPU. Totally no clue where that comes from - as I said, on Intel it works fine. Once that is done, we will start collection on CME and - as I said - possibly FOREX, CBOT and the ICE are next on the list. I basically will add them as I see I can handle the data volume And as I actually need them for my own operations. Once the full feed scalability is fixed.... my collectors start working Even if only in dump mode (write all out to a text file for later transversion). My programming focus is now on trade capture for internal reporting and getting the whole cleaning / accounting side into a database - so I can trade with another frontend and capture the trades with a software for internal reporting. Plus holding one's hand (new trader starting tomorrow after sim training - lets see how that goes) and looking for the first programmer to add full time to that project Btw., I am 60% done with defining a binary coded market data stream format Which WILL be open source Incl. C# code to read it

Who lets you do that? Rollover is manual to my knowledge. The price you get is the price you get

Yeah, but you dont look for a news FEED I really miss that electronic calendar type of thing

Double here, only that I dont are about Zen-Fire / NT (as in: I work in C#, the broker connection ahs no relevance to that). I also have strategies that behave different around news events Is there any way to get that stuff? Econoday - Economic Calendars, printed, online and mobile seems to hae it, but the pricing is - expensive, and I see no API / RSS access, which sort of is needed for proper programmatic access

Trigger orders (stop, market if touched) are not part of the market util the trigger is triggered an thus are not visible. Please alos note that you may not see the total orders. Iceberg orders - either even by the exchange, or the trading framework - may be part of that. What is that? Buy 100 cars, show maximum 10 Buys: 100 in lots of 10. Some exchanges support that, some trading frameworks (Zen-Fire, Rithmic - ZenFire only when using the RIthmic Api) support that automatically.

Because the order needs a matchuing opposite order. If you have a stop, it is NOT in the ist of orders to execute at the stop price. Once a trade (the first) happens at the stop price, the trigger is removed and the stop order turns into a MARKET order. It goes then into the END OF THE LIST OF ALL MARKET ORDERS. If at that point there are more market orders in the list than limit orders on that price, the opposite orders get eaten up, the price moves and then you execute. Example: You have a stop at (YM taken, nice points) 9500. There are 51 asks to sell there. Price is triggered - your stop (of 1 contract) turns into a market order, there are (without new asks coming in) a maximum of 50 now there (because one had to sell to trigger the stop. If you go into the end of the list of market and limit orders (and where you go there may depend how old your stop is - older stops have priority) at position 60 (all one lot, to keep things easy).... you dont fill at that price... because the price od 9500 is eaten up by the orders in front of you. Now, if the market is really thin, the orders in front of you may eat up a couple of ticks. Trick here is: * Your stop turns market order and goes into the end of the trade list. * Stops are evaluated in historical orders, so someone putting his stop in BEFORE YOU.... gets into the market before you. * I am not sure how that is handled for modified orders. Would be fair to put a modified order to the end of the list, at least when it contains a price change. Slippage naturally happens when the amount of orders triggered and processed in the market overwhelms the resistance price. Seriously, this is not something for a technical laboratory - that is total beginner stuff. Get a book like "Trading Futures for Dummies".

Well My RIthmic tests are "on hold" until they fix some issue in their API - it overloads on AMD processors with the full CME feed, works fine on Intel They got an AMD server from me now for testing I Hope they sort it out soonish. I am now working on order handling and subscribing only to specific instruments for now I hope to start full data capture next month (I really hope) Did my first trade today (on my app against CME test environment) - the whole low level API's are not exactly totally easy Now time for some front end and data storage I hear a lot of mentioning of OpenTick (defunct as they are) Anyone thinks there is a market for not necessariyl real time historical data (especially if it includes best bid/ask or even full bid/ask)?

Zen-Fire is a resell of the Rithmic API. It exposes only a small subset (interesting enough) of what RIthmic is capable of... and as the RIthmic API works on Zen-Fire and the Rithmic operations team know about Zen-Fire Guess who is the real deal. Capabilities of the Rithmic API missing in Zen-Fire: * Historical data Limited, but there. * Cross-Instrument-Triggers (stop on future triggered by forex price move, or by bid/ask move) * Subscribe to complete exchange data (HEAVY on bandwidth - CME has around 400.000 symbols alone) * Time based orders. I actually CAN put in an order to enter to the exchange at a specific time.

I fully have to agree with that. C# is pretty much a great langauge and it is done by really great people pushing it forward - a lot of stuff in now wihth 3.5 and coming with 4.0 is pretty revolutionary for a mainstream langauge - things like LINQ etc. have all more or less been there as side-things, but C# and .NET put them into mainstream. Finally. You can do a LOT with C# - contrary to (old) VB it is designed as a main langauge - powerfull runtime, feature rich. I have been working with it for about 10 years now, and only sometimes do I see the need to do some C++ (MOSTLY in the form of C++/CLI, which is the form emitting classes compatible with .NET/C# - like now wrapping up the Rithmic API for use from within C#, as it comes as C++ lib files). If you are serious about programming (not just for trading or so), make sure you learn MORE - some more language with different concepts. Real assembler may be interesting for a summer, as should be SMalltalk (the original object language), F# (functional .NET language) and maybe something else. Make sure you do not see all things from the point of view of C# ONLY. For trading, though, C# will be more than enough. Heck, one can write serious trading software in C#, and contrary to what the (not too intelligent, it seems) people at NinjaTrader did... it does not have to be slow and crash all the time This is a lot more bad programming than a .NET fault - as can be seen in the tremendous speedups they get for Ninja 7... by redoing broken architecture.

It is not an exchange decision. It makes no sense from the beginning. A limit order higher than the current prices will logically - regardless of exchange regulations - execute. It is non-sensical for it to exist longer than it takes to match it.