Hi GeminiSimba,

judging from your statement, you're bold -- and boldness is maybe the most important character trait you need as a programmer. The second important character trait IMHO is a high frustration tolerance. As developers, most of the time we have to face an overwhelming lack of information, incomplete and even misleading information, a messy setup, half broken tools, crappy documentation and combatants, whose best skills aren't the social ones.

As a first step, I'd recommend that you try to understand why your goals (becoming a crypto programmer) are very ambitious. Unless you're an absolute genius, you'll need years of constant, actual practising to contribute to that area without doing any harm. Thus it would be a very good idea to define yourself some intermediary goals to aim at. Kind of  like stepping stones. For example, how about being able to code a web site, which might be for some crypto currency related service? A next, somewhat more ambitious goal would be the ability to contribute to a standalone client application with a nice GUI, learn GUI programming and improve the usability of such a GUI. A next, even more ambitious step would be the ability to work in backend / server processes. Like, e.g. the ability to contribute to a trading engine which matches bids and asks and executes orders. And -- while you proceed on that part, I'd highly recommend that you always try to read and understand other people's code. So that you get a feeling of the actual problems contemporary coding has to deal with. Thus, instead of just using frameworks and libraries (as most programmers do), you could add the additional challenge for yourself to look into the source code of such frameworks and libraries and try to understand why they are the way they are.

The good news is that with today’s vibrant open source world, it is very much possible to get your hands on "real stuff", without the need to be part of an university or industrial organisation. I'd wish way more people would build on that unique opportunity and engage into OpenSource development, instead of wasting their time and mental capacity with gaming and social networks. Because it is very much possible that we're headed towards a world again, where computers are completely dongled and -- unless you sign a non disclosure agreement -- all you get your hands on will be toy stuff, not real stuff.

Regarding a starting point for help: have a look into stackoverflow.com ! There is a huge number of beginner questions for each and every language there (including C). This might give you a feeling about the problems other people have to deal with. Plus, especially from the answers to beginner questions, you often get links to interesting blogs.

Beyond that, the definitive, hard-core forums for any widely used programming language are still in the "usenet" (the "news" protocol, which is a very ancient, text based message protocol from the old days of the internet). This is still the place where you find the old farts with years of experience and lots of battle stories, the people who create programming languages and build compilers. There are some sites which bridge the news groups into the Web, so you can read them like a web forum. My recommendation is to look at gmane.org. From there, look into the news groups in the "comp.lang.*" hierarchy.

hope that helps
and: welcome in the community of people actually programming computers!

-- Ichthyo

Thanks for the OCO orders.
very much appreciated.

One detail question: Is there some kind of headroom added to the current ticker price?

E.g., currently the lowest ask is at 619, highest bid 618

Now, if I want to open a long position and thus enter limit = 614, stop = 620, I get a error popup "OCO stop rate is below ticker". The lowest stop value accepted by the system is 627, which seems quite off the current spread IMHO.

Basically we had this very situation last year.
A critical flaw in a just released new version of the standard client was discovered, which would have led to transactions being judged differently by miners, and thus to a split of the chain. It took the core devs just some hours to negotiate with some large miner pools to revert back. Before 12:00 PM UTC next day, most of the bitcoin network had followed this move, and the few problematic blocks where obsoleted this way.

Initially this created quite some turmoil and a lot of bad press, but in the end consensus was that this incident showed the Bitcoin Community's ability to deal with critical situations professionally.

This whole thread is built upon a fundamental misconception

You assume that there is a thing called "The Law" or "Legal System", which you conceive as being eternal, "underlying" as made out of stone.

Yet in reality there is no such thing.


You can go out on the street, pick some random person, force him by pointing a gun at his head to go to a "court" you have set up, and read to him the verdict, that his money is seized as illegal, because it belongs to the poor hungry people of Africa and was taken away from them by capitalism and imperialism. Chances are that that person will yield and hand over pocket money to you.

But in no way this makes your action legal. Nor did the person hand over the money because you where in right, but because of the gun.

Please note the important point:
You must not draw conclusions about the nature of the money involved here.
You may only draw conclusions about the nature and the meaning of the gun.


In a similar manner, you can imagine some state or legal system in the world declaring Bitcoin as illegal for reason of [insert random ideology here]. Then this state or legal system can use its mechanisms of power to seize Bitcoins whenever a user interacts with some entity under the control of this legal system.

Such actions can be quite effective for people entrapped in such a legal system, but this fact doesn't turn the provisions of such a legal system into an absolute truth. You might be drawing conclusions about the nature of such a legal system itself, but you can't use it as foundation for drawing conclusions about the nature of things like the Bitcoins involved.


The decision to handle and classify Bitcoins in a specific way is a political decision. It is pointless to use logic to derive what is the matter of fact in this respect.

nice try.

now tell me: why should anyone put money into infrastructure or act as ISP, if the ability to make money by selling one's customers is taken away?

this thread is a piece of real satire

just some page back the whole herd was yelling because you could took out loans below 1 hour without charge. And now we're starting to yell because of that charge?

ROFL

after reading the backlog in this thread....

one thing which seems to be blatantly absent with many traders quite vocal in this thread is common sense.

If you want to achieve something, you need to be willing to use the appropriate means: If you want to trade quick swings over several $100 in a matter of seconds, you need a guaranteed reliable market access, and you need a trading link which responds in terms of milliseconds. If you attempt the same with just some website service, you're lacking common sense. Or to put it the other way round, you aren't acting rational, you're gambling. If you gamble, you loose on average. If you do it for the adrenaline, then please go for it, but don't whine afterwards.

You can apply rational strategies even with an unreliable slow market link. Just use more long term strategies then. For example, if you'd traded some kind of market-follow strategy, maybe even some dead simple "average crossover", then you'd be switched over to shorting days ago when we went below $800 and you'd be taking home money now. Or at least you'd be stepped out and watch the drama from the sidelines.

The trick is to find a pattern of behaviour which you are really capable of keeping up long time, and which is profitable on average.

There is a much deeper question involved here.

If we want to protect people against each other actions, if we try to prevent every possible way of "manipulation", we end up at a regulated market. In fact, the traditional markets became more and more regulated exactly this way: Whenever someone got burned, he immediately starts yelling, asking the "authorities" to protect him against losses. But if the "authorities" then place taxes on the gains, everyone is pissed off.


That being said -- an order book just shows offers, not more. It is not binding.
Fake orders and real orders can be placed and pulled any time. This shows some current sentiment, not more, not less. But for a reliable judgement, you should look at the actual deals, and more importantly, you should look at the actual trading volume.


Others have already pointed out, that it can be convenient for a trader to have long term orders sitting there, and temporarily place short term orders closer to the spread. At times you'll even place several buy and sell orders at the same time.

Now the proposal is to prevent or limit this pattern. Personally I think this shows a questionable mindset: basically you assume that it is necessary to watch and guard people's actions. This is nannying.

If we we consider placing such fake walls harmful, the proper way of defence would be not to "regulate", but to make them expensive: simply deduce the trading fees from any order considered for execution, even if the order will be discarded due to lacking margin.  

In trading, if someone panicks, he deserves to be burned hard.
If someone can't control his emotions, he'd be better of putting the money into some kind of pension fund at his local bank.

Nice proof that you're clueless.

Some people will behave according to your "algorithm". These are those kind of noobs which buy at all time high and sell at the bottom, just before the market turns round. And yes, the market needs a herd of clueless noobs. Those are there to allow skilful traders to make extra profit.

Some points you need to understand

• market equilibrium is driven by real forces, not fake power.
• people do buy / sell at a given point, because they want to do so, for whatever reason.
• when your strategy is built on cheating other people, be prepared to be gamed yourself.

technical question to the Bitfinex team: how do you handle now the situation when you loose the API link to Bitstamp? Will you still remove Bitstamp orders from the orderbook temporarily, until you re-establish the API link?


EDIT: Bitfinex -> Bitstamp

No it refers to the order book, or the "market depth display" if you want.

A single exchange has typically orders at a bid side and an ask side and a gap in between, called the "spread".
Whenever someone places an limit buy order above current ask price or a limit sell order below current bid price,
this order can be trivially filled and thus will be filled immediately.


But for a service like Bitfinex, which aggregates several actually independent markets, this structure doesn't hold anymore,
since not every limit buy matches automatically with any applicable limit sell order. If you just overlay the offers of two markets
into a single 2nd-level orderbook, it can well happen that the lowest ask order is below the highest bid order, whenever the behaviour of the underlying markets diverges substantially. This situation is known as "crossed orders". This can lead to confusion as to when limit or stop orders are actually triggered by the trading engine.


The change implemented now by the Bitfinex people is that they now don't display such "crossed orders" anymore. Rather they detect them and execute them in a best match way until the crossing has resolved and bid is below ask again. That is what I infer from the quoted statement and from observing how the trading engine works now. (Please correct me if I'm wrong)

for me it's mostly related to market psychology.
Esp. during quick movements, the BFX internal market behaved significantly different than the "large public" markets Bitstamp and Gox.
My (surely subjectively biassed) impression is that "actions" tend to start rather from Gox and sometimes now from Bitstamp. Anyway, I often just watch the Bitstamp chart and when I see e.g. 2 big Ask walls inserted, then I'd like to identify those on the Bitfinex orderbook as well.

Regarding "psychology", Bitfinex had tremendous growth, but for all foreseeable future BFX and BSTMP are two audiences, and most notably one of these two audiences (BFX) can see and watch the other one in action, while the opposite is not true; action on Bitfinex appears as sudden market activity on Bitstamp (unless someone has an Bitfinex account and cares to look into the BFX orderbook).


Regarding the feature itself, in the past it had three distinct aspects:

• the ability of routing orders to Bitstamp or limiting orders to Bitfinex
• the fact that the Bitstamp orderbook could disappear any time temporarily (due to connection problems), so you'd better not rely on a big Bitstamp order sitting there
• the convenience to see which offer comes from which audience

Order routing is gone now. Not sure about the handling of connection problems. But the thrid point is surely just some convenience, since you can look on a Bitstamp chart directly.

If you need to get just the data (as opposed to do trading via the API): getting data is easy, since it is just a HTTP request. You get a string as response, which is formatted in JSON format. You can try out retrieving data from the API just with the "wget" command of unix. JSON format is human readable.

Not sure if C++ is the most clever choice, since there aren't so much easy to use libraries for doing this kind of stuff (i.e. pulling some web page via HTTPS and parsing a JSON string). Chances are that you'll end up writing hundreds of LOC for things you can do in other languages with just 10 lines of code. If you're familiar with C++, maybe you should consider to learn a bit Java, since Java is also a compiled and statically typed OO language, but offers an abundance of ready-to use libraries and tutorials. (Unless of course you have other reasons for using C++ beyond just familiarity)

Even that is a simplified view of running a business.
The most relevant part your argument misses is to account for all kinds of debt.
Especially when you "bootstrap" a business, this situation is far from the equilibrium. The charges you can take are basically dictated from external circumstances, plus typically you need to attack the competition. Then, as the organisation grows, you start accumulating organisational debts and you start accumulating technical debt. If you want to survive, you're bound to create increased income either/or by expanding into new areas or markets and by rationalisation. This additional revenue typically goes towards the pay of debts.

Additionally, every growth incurs an increase in entropy, which means, the more you grow, the more you have to fight for revenue through rationalisation. A classical example is the difficulty for a startup funder to delegate and "let go" at the point when (s)he can't manage everything alone anymore -- which often results in an outburst of "internal politics" (and even kills a lot of successful start-ups) Thus, personally I would not expect rationalisation gains to be passed on to the customer, with the exception of very old and stable enterprises. Or as a marketing gag.

Unfortunately most people's knowledge about running a business is shaped by the images and misconceptions fabricated by the marketing departments of our industry. Which are clean and slick and not especially related to reality.

While it is possible to exploit this, it doesn't seem to be done regularly. On a second thought, from the traders POV, it isn't worth the hassle. You can earn way more by effective trading, but you can also loose way more than the typical order of magnitude of the swap.

I'd guess that this unfortunate gap becomes largely visible only during flash crashes like the last days, where many people open positions during a very short timeframe. But there is the danger for an experienced and planful attacker to exploit this gap systematically, and probably the Bitfinex team is already aware of that.

Without knowing any internal details of the implementation, my guess is that it's non trivial to calculate the swap precisely in a way that scales out on a growing distributed system, since a complete trading position is comprised of N position changes multiplied with M involved loans.

I observed several times when it looked as if stop orders weren't triggering, where, at first glance I thought they should have.
But when investigating closer (including data available via the API), it turned out to be correct.

Kitkat007, please re-read this response again, and try to judge it as an outside observer.
Just a hint: talking this way is very telling about your mentality.

---

I'd like to add a note regarding that issue of stop orders seemingly not triggered. (I am a user, and not affiliated to Bitfinex in any way) I observed that behaviour multiple times too, and thought there might a be a bug or problem in the engine. But when watching carefully, I couldn't find anything actually to complain.

The point is, matters are far more complicated when you take a close look (as in e.g.: try to implement a trading engine or algorithm).
There is not just "IT" went down. Actually, there are several rates involved.

• you can create a rate value function by connecting the price of each deal happening on the platform (and then making that into a continuous function of course)
• you can do the same by connecting only the deals done on one of the connected platforms. This function looks entirely different
• next, you can use some heuristics and combination with other data, to create for each of the above a bid rate and a ask rate function
• and besides all that, you can create similar functions based on the offers currently in the system. The reasoning is that such would be the rate the next buy / sell would happen at
• and last but not least, you can (and indeed should) use a heuristic formula to combine all those data into two functions, which could be titled "effective nominal current bid / ask rate".

Note, the last one, this "effective nominal current rate" is what drives the events of the engine. There is no canonical solution for this, and there can't be, since it is at the very core of the specifics of any trading algorithm or trading engine how to build those functions. Also there are somewhat non-deterministic effects involved, since every engine runs at a specific sampling/update interval and only captures data with a certain granularity.

If you look closely, you'll notice that the nominal rate, as displayed in the UI, is often slightly different than what is visible over the API functions as last trade, or what is visible e.g on Bitcoinwisdom (which, as I take it, also pulls the Bitfinex API)

If, after shaving away all those complexities, there remains a situation where the nominal rate is lower than the trigger point of a given stop sell for more than one update period, then this would actually be a bug and I'm sure Bitfinex stuff will be very interested in getting a full documentation of such an incident.

Unfortunately not.
But this is a very frequently posed feature request.
(Typically the request is to combine a limit and a fixed stop)

This would be very helpful indeed, since it supports the standard situation when you try to "get into a position".
My guess is that implementing such a feature will create a challenge for both the internal data representation and the visualisation in the GUI


@Bitfinex team: are there any plans or have you considered such a feature?

Hi.. also this topic was discussed last spring in detail here

For one, this "Bitfinex Sentiment Index" is deliberately crafted in a way to give an rough indication,
but without being able to derive precise numbers (which could reveal internal data which was at that
time deemed dangerous).

But beyond that, the "Sentiment Index" always reflects the current sentiment of the people doing margin
trading on Bitfinex. Of course this can change in a very short time span.


Such can happen, and will happen regularly. The tricky part now is how to proceed.
Basically you have two options: jump on the rolling trend, to get at least part of a market movement, or just let that opportunity pass and use the next opportunity.

Please note, there is no mechanical solution to answer this question. Especially please don't think "Technical Analysis" can answer you this question. (TA gives you only tools to see some trends, not to judge them). You need to rely on your own judgement as a trader -- and exactly this is what differentiates between traders and portfolio managers likewise. The "better" traders/portfolio managers are better in judging the situation, and thus are profitable above the average; the lesser traders are profitable below average and thus are the source of the formers profits.