Open source/donations are mediocre. C# is toolable for proving in coding. What data set sweep rate are you using?
agreed re open source stuff, idealistic ventures. Sweep rate is not faster than what a human can do really, should not be a limitation at all. I was venturing into C# territory but currently am in full retreat, Multicharts and their script are soon going to be enabled to do all that I need. The plan here is not particularly ambitious, trade one future contract with a bot and most likely it will be a semi automated bot that needs input from stuff drawn on the chart unless I can automate the geometrics.
I agree with you wholeheartedly on the sweep rates. The computer does have the ability to sweep faster than humans and that factor does straighten out of the games humans play while looking at the screens. The computer has the ability to "look around" at all of the degrees of freedom that are worthwhile to generate each sweep. As you said trashing what is not important allows for specificity. Computer applications can do the same quickly and in slightly different ways. Making tooling that is not directly related to market price and volume can be made useful by computers. Then, the results can be backed out to the visual appearance of the markets. It is as if a shadow is cast into the near future where the shadow is as solid as the present regarding trader activity. It appears to me that the larger the capital the trader is using, the more he is likely to execute his plan for the market to come to. In other words, his die is cast before the present and subsequently the market could come to him. I am dilligent about watching the extent to which these plans are not fulfilled and what remediation can take place at what time and what price. A humorous non commodity example is the CYGT position that sunk a fund. Daily people make money trading CYGT, but the plans of other, they think, precludes their trading it in the price range now present. CROX is another example in another thread. So in the DOM of ES many players show their hands and it is done on two significant levels: thousands and hundreds. At turns, a lot is left on the table by these people. They are seen simply because they have the capital (margin) to place the orders. They do plans that must deal with the capital they have either in the market or on the sidelines and not working for them. I find that it is possible to have the next two trades geared up for execution as a consequence of computers having the capability that they now have. certainly it can be done manually because of the pace of markets at this time and for the foreseeable future. Going down to the tens and units values of the DOM is similar and not so significant. The tick to tic time out of dwell time is often determined by the tens position. Here, the counterintuitive nature of reasoning prevails. ACV ratios are, of course, not applicable; they are non specific and so do not relate to shorter term trading like scalping and expert high velocity trading. What does apply is important to be able to monitor and keep track of. To trade effectively a person cannot do CYGT's or CROX in any market. "This is "hope" trading and the market operation has simply moved away from the positions these people hold. For short term and high velocity traders tic dwell time and who leaves what on the table is very important. Losers got left behind and they are going to only endure the pain for so long in time and so deep in "drawdown". We see the quantities and they are tabbed for us by the computer. We watch time pass. Two things happen: pain limits are reached and are typified by "internal formations" We note these as well as hitches, stalls and dips. Each closes the "history" of the sums of money left on the table in the tens, hundreds and thousands places of the DOM; they ground down to zero. Price translation in tic pairs shows the losers in pain and the winners taking profits. Translation is "slowed" by both as tic pairs dominate. All of this gives a person the trading sereis that will unfold during the day. Always having the next two trades more ot less determined means that trading is fairly relaxed because it is determined by fairly simple observations of trader's intentions and trader's required efforts to either preserve capital or take limited profits. I you look at the sales profferrings of money mangagement firms, you see the corporate book being played out through the slogans they use to capture clients with their mediocre track records. The trading that they do in instruments they use to protect capital shows, often, in their coming off sidelines or "quitting" "drawdowns" that are beyond computerized corporate limits. Look at Greenspoon's record at his financial house for a limiting case. He makes only 2 to 3 million a year handling all he is capable of in the ES, namely 400 contract units traded 60 times a day at less than a net tic profit per turn. He does entries and exits, unfortunately.
Jack, Thanks for answering a question I had regarding the ACV. There is the theory that the market moves toward activity, which is supposedly indicated by the standing limit orders - it moves toward taking out the orders. Guess it's a version of the 'stop running" view. This is contrary to the minority control view, but it seems the differences may be depending on the "where" you see these two different things happening - points of change vs. when the market is running. Must be related the the T&S somehow. But I'm getting off topic here.
Back to posting briefly, nitro has given a great insight, You have to get passed developing models, solely, from backtesting historical data. Though, you need to be able to backtest, very well to do so... and still do so...
been a little more than 6 months now. lessons learnt: . had a few more environment-related glitches in the first 3 months (H/W, S/W, lines, broker API, etc) but only few & far between and some won't reoccur -> i really don't need to check, if anything breaks, i get an auto-warning from the main box or its twin . the vol of returns in Aug and Nov notably has been such that i've had to express it against a designated benchmark to pacify febrile institutional type asset managers who can't stomach much, thats why they are in those jobs... but all going well since . meanwhile and as a consequence of my marketing effort i got pulled into a bank to advise as to how to get their execution algos to "work"... too funny what i am witnessing there... obviously J.Kerviel has set a v.high standard but... these guys are spending tons of dough on useless stuff, at the same time they have no backtester (boing!), they are "testing" live on prop books, and that's because they got a rebuke from the regulators for overfills when they initially were testing against client orders (unbeknownst to said clients, if you were going to ask...) but basically their algos suck, might as well toss a coin on every tick and with just a few reasonable entry/exit rules i'd get similar crap results if not better... the sort of bots that just lose you (the client) money, but in a way you can't understand )))))))))) and its not even a crime........................ mmmmmhhhh....... come to think of it...... what's wrong about robbing gullible investors??? shit, there i go again, thinking like a banker.......