BlowFish

Sadly they (like Ninja in it's current implementation) only store historical data to second resolution (better but no cigar). I think I have made them aware that this is an issue, I am not sure I managed to get across how important it is to a whole segment of their target audience. Whilst currently they offer a slightly better product in many different ways to TS there slavish devotion to maintaining what is essentially a TS clone is a double edged sword. On the plus side you get easylanguage...... not only is it easy but lightening fast to develop stuff (I too have a vast library of code and code snippets in EL) on the downside they are emulating what is essentially and archaic platform including numerous weaknesses. Sounds like there will be quite a lot of re-architecting involved to provide some of these features. Hopefully they will use this opportunity to start with a relatively clean slate. P.S. I still have a soft spot for MC & TS if they could correct a few core issues they would not come with a whole heap of frustration too!

Certainly agree about the ms time stamping. Hard to take TS seriously as a platform until that is introduced. The trade objects sound like a great idea but I was talking more about converting EL to a fully OOPs implementation. Whether you instantiate a new trade object or invoke a function that adds a trade record to a linked list doesn't really matter too much (to me). Incidentally I meant to ask if they have made an official announcement about ms time stamping? Obviously it will come if they want to remain competitive just wondered if they are trumpeting about it yet. As an aside Neoticker has had all this stuff since......I dunno... when it was launched way back when. Still a great choice for retail traders who require tick precision and a whole lot more. (I have no connection with TickQuest).

I wonder if you fully understand compounding? The wonderful thing is that you can compound less aggressively as time goes by, and so reduce your risk whilst still achieving geometric growth (as opposed to arithmetic growth). This is what the race is all about. Even if you had a tiny edge and risked a minuscule amount on each trade you would be very hard pushed to take that long. Of course if you have no edge or are prone to risk too much the outcome will be consistent and much quicker than 20 years! The only place luck enters the equation is in 'streakiness' of results. As time goes on you can be more resilient to streaks by decreasing your risk a little.

Not really 'using' it but certainly not 'done' with it either. Have you found any merit in the principle?

It will be interesting to see if they retain the "easy" part of EasyLanguage whilst introducing this. Whilst OO does provide some advantages I am still not convinced that it's benefits outweigh the dis-advantages for smaller projects. Things that take minutes and a dozen or two lines in EL seem to take hours to day(s) and pages of code in more OO languages. As I have said before I am a dinosaur....perhaps that is something to do with it

:thumbs up: .................

@tams it dosen't really matter what you put in, Tradestation is likely to mangle it into garbage. You can get round this to some degree but it is messy too. It just does not have the architecture and internal precision for accurate testing in lower time frames. Using 1 tick bars for testing can help but it is far from ideal. @tresor Matlab is a great product but aimed more at modelling & simulationrather than back testing. Takes you towards a similar place but from a different direction. The big issue with it is that it does not handle large data sets very well (even with all the wizbang addons like the parallel computing toolbox) the main reason being that it needs to load the complete data set it is working on into memory, that can be quite a lot. Some while ago I had the fortune to attend a do that MathWorks hosted for the great & good in the financial industry. A straw pole of the attendees clearly showed that by far and away the predominant use of Matlab was modelling portfolio risk. I did meet a couple of guys that where trying to do interesting stuff at 'higher frequencies' and they where having issues with large vectors. It is possible but it requires a fair bit of 'fudging'.

Total is what is commonly used.

Do it your way then. Just trying to help. I take no pleasure in the system blowing an account albeit a small one.We can agree on one thing, the trader should determine risk based on their account size and their tolerance of risk (risk of ruin to be precise). Any system that gives absolutely no control to the trader with respect to position sizing is one that I would personally not trade.

Sure the Junk ones do, one should just discount those. BTW risk is not (or should not be) determined by the system. That is the key parameter that the trader has control over through appropriate position sizing. Ericbubb did not post enough info to determine whether the system he was using was junk, however the way he was trading was clearly over leveraged. Risking 5 to make 1 is absolutely fine (contrary to conventional 'wisdom') provided your winning percentage is higher enough of course one must carefully look at risk of ruin but with high % systems the RoR is actually relatively low (compared to more 'normal' 40-60% risk 1 to make 2 or 3). They tend to have much smoother equity curves and much lower drawdowns. Probably one of the best pieces of advice one can give too a new trader (even more so one that wants to use an automated system) is to thoroughly understand RoR. My favourite site for an intro is TradersCALM - Calculating risk of ruin . Anyone trying to evaluate these systems without a thorough understanding is pretty much doomed to failure. Any vendor that does not provide comprehensive enough performance details to allow one to calculate this metric should be crossed off the list (as I mentioned earlier in the thread). That's most I guess. If a discretionary trader is risking 20% account equity a trade chances are they will be gone before too long. I don't know why people expect that risking that much with a robot will be Ok?

That should just not happen under any circumstances. You need to have circuit breakers in place and more importantly use appropriate position sizing for your own tolerance of risk of ruin. in short, you bet too large. You will always have strings of 5 (or more) losers. RoR is absolutely critical with automated systems. Any vendor that does not give you enough info to calculate suitable position sizes is one to cross off your list.

I believe they are not though might be mistaken. This appears to cover a lot of the ins and outs So is Spread Betting really tax-free?

There are ways of doing things using 3rd party software. A lot of charting apps have auto trading through (fairly) simple scripting. Multicharts, Ninja, Amibroker spring to mind. The biggest problem is that you want to trade stocks. There is more sophistication and choice on the futures side. If you plan to do unattended operation you might want to use a broker solution (strategy runner for example). In that case the broker runs your script on their servers. Never used it myself but at least you don't have all the infrastructure worries. Again dunno if anyone offers stratergy runner and stocks, you'd need to talk to some brokers.

What does your algorithm trade? Amp & Mirus offer demo accounts for futures that would probably fit the bill. Do you execute your algorithm manually?

Most of the points raised here are equally applicable to all sorts of trading products and services. The catch 22 is that the people that all the junk is targeted at are unlikely to have the skills to evaluate the claims made. I guess 'black boxes' (EA's) are particularly tricky to deal with.

The charts are the same as the instrument you are betting on. It's pretty much like the FX bookies. You bet X pounds a tick/point on the movement of a financial instrument. So for example I could take a long bet on the ES at 5 quid a tick. The spread bet company will be the counter party though may chose to hedge their book in the real market if it gets too lopsided.

In a later thread you will see some discussion on this. PVP is a statistical measure whereas POC is a heuristic approach. Whilst they look similar the ways that Jerry suggests using these statistical measures don't really share that much with the way people tend to use MP. The short answer to your second question is no.

I think most would agree that it does not matter whether they are 'commercial' however I'd go further. You mention 4 criteria - 1) large - not so important to me which is fortunate as while you can detect patterns you can not know for sure whether it is 10 x 100 lot traders or 1 x 1000 lot traders. As you say liquidity is what it is all about. Fortunately things like intensity are pretty effective at detecting immediate liquidity requirements. 2) Getting into - again not so important perhaps - whether they are getting in to the market or out of the market. Fortunate as pretty hard to detect (imho). Arguably participants getting out of the market are equally worth coat tailing as they often cause greater movement (more urgency). 3) Initiating new trade - again not so important (imvho) similar reasons to above. Hard to know if the buyer/seller is buying/selling to open/close. Mind you I do like how you use cumulative delta as a sort of proxy for OI. 4) Directional trade - not bothered - hard to detect for sure especially when you consider that some of the largest participants are not speculating on market direction at all. The key thing is liquidity in fact you can look at markets fairly simply as systems to match the liquidity requirements of participants. I think my criteria would be a bit more general, determine where liquidity might be found (simple S/R would do or your resting inventory model seems pretty robust). Monitor the order flow (demand for liquidity) particularly in these areas. Do not pay much attention to the who is operating and why (or do so just for fun if that floats your boat). Fortunately you can always (leaving aside dark pools) see someone is at it, and by using tools based on V@B/A and or intensity see how aggressively.

I wonder if any do too. Of course none will get close to matching the extravagant claims of the vendors but I wonder if any will actually run profitably with suitable parameters. My hunch is that there must be one or two but I don't really have the inclination to put the (large) effort in to test and locate them. It doesn't help that such ridiculous claims are made I guess if you discount all those that would narrow the field....possibly to none!

Indeed. Whilst browsing the interwebz for interesting material for the kindle I came across this Powered by Google Docs One of the premises is that intense trade is a function of urgency with respect to available liquidity. It is focused on defaults but seems relevant to any situation where immediacy is required. The paper suggests a few reasons for urgency (avoiding defualts), there are others that occur to me that are not about 'defaults' (last leg of an arbitrage trade for example). I haven't got my head around the maths yet (I wish it wasn't so rusty) but it seems to seek to quantify intensity and its short and long term impact on price. (amongst other things). Quantifying what is 'intense' seems like an important thing for an agent.

The exchanges have a pretty clear definition of what a 'commercial' is. The emerging discipline of market microstructure has pretty clear definitions of types of participant too (more than simply commercial/ non commercial). The thing is they would both appear to be different from yours.....maybe that is where the confusion lies. It might be helpful if you offered a concise definition rather than anecdotal 'definitions'. Is it simply a matter of size (10000+)? speed (seconds)? Account size (millions)? Is automatic execution mandatory for you to consider it commercial? Finally, it doesn't really matter where the inventory comes from as long as you can track it, for the sake of discussion maybe it is better to leave aside the who and why. To be honest this all smacks of Tradeguiders 'smart money' rhetoric which is a shame as to me it devalues some pretty interesting work. I would suggest that anyone that is interested in why & how different market participants trade, should take a look at one of the books on market microstructure. (Harris or O'Harra spring to mind).

If you value completeness over timeliness. OFA seems to be geared towards timeliness (it does not load any historical data at all for example). Also bear in mind that using bid/ask differential as a proxy for trade direction/order flow is only about 80% efficacious (several studies come ina round this number). I am not sure that 'not usable' is the correct conclusion (clearly people are using it). Tainted results sure I'd accept that, not ideal for cumulative work sure I'd accept that, better for 'scalping' (where arguably timeliness is more important that completeness) sure I'd accept that too.

I agree. Which is a crying shame as some pretty solid ideas are being obscured by hyperbole and hubris. There are many market participants that have a wide range of modus operandi depending on their immediate objectives and current market conditions. To lump a co-located, market making HFTs with someone accumulating a large hedge over days or longer (so motivated by temporal stability rather than 'profit') seems ...well pretty much what you said Still if you are able to set aside the who and why and concentrate on the what there are some neat ideas to pursue. Anyway, back on topic, I wonder if anyone has experience of tools and applications to aid in this sort of work? In the very distant past I used Neuroshell Daytrader, pretty promising though ultimately not successful (probably due to my inadequacies at the time). Worth a look. One thing I do recall is that some quite unexpected things made good inputs and if you let the genetic optimiser rip on a bunch of inputs and there parameters it would often pick rather counter intuitive inputs and settings.

Thats the same one I linked a couple of posts ago. You can find a couple of charts I posted in the original trade intensity thread using this code. I actually rather like the indications it gives (small bars represent intense trade). Always thought intensity needs some sort of smoothing (which again was discussed before) it can go from very close to zero to pretty darn massive in one tick. It would be pretty trivial to convert to intensity. Isn't time taken pretty close to the reciprocal of intensity any way? Too early to think about for now:)

I don't know AFL but it seems like looking a bar further back for the signal and triggering on the current bar would do the trick. cond1 = ref(h,-2) > h; cond2 = ref(l,-2) < l; cond3 = ref(c,-1)-ref(o,-1)>0; signal = cond1 and cond2 and cond3 if signal then..............................................//Dont know the syntax for AFL if c > ref (h-1) then buy.........................//buy on this bar Not sure about the syntax but that logic should do it. assuming you can buy on the current bar. There are alternative ways of doing it that might be more appropriate. It is a question of logic rather than syntax of AFL. As Tams would be the first to say try to set out the logical steps and then try to code it. If you want to check for the break out on the next bar you will need to set a flag and make note of bar numbers.