So as you may already have guessed the bot is <b>not</b> "mimicking" the behavior of a single trader, but of an entire <b>team of reciprocally independent traders</b>. The function of "coach" is exercised by the bot (based on the user's strategy), which "coordinates" all these traders in order to achieve the best ratio: AvgProfit --------------- | maxDD | This is why we can have at the same time traders who are doing a CT (countertrending) game and traders who are doing a T (trending) game. Clearly, one watching at the IB account or TWS activity will not easily "understand" the game because what he sees is actually the <b>"overlay"</b> or <b>"projection"</b> on a single account of all games played by all traders. What the bot does is taking the orders from all traders simultaneously (<b>just "like if" each trader were playing on a different account</b>), and <b>merge</b> them in the resulting single actions. Is this understandable ? In other words you can picture it as a sort of "multithreading" system with a <b>"supervisor" channelling the result of all this activity of several independent traders</b> into <b>1 account</b>. This is a crucial concept (to me). In my experience (which can be of course wrong), it's not possible to create a systematically profitable system with "single-threaded" approach, because in that case we would return to case 1 of 3 analized in a previous post, where one uses stops on single trades. And that never turns out to be <b>systematically</b> profitable (at least, i have never seen that). Tom
Have been continuing the experiment (44 days). Metals (si, gdc, ...) have been quite wild, causing some sharp drawdown (but still very far from simulated worst cases) Si moved 1594 ticks. Probably a wild season on oil is going to start. I am going to change the parameters to anticipate inversions, and redistribute an update to all the people helping with testing. I have added a complete rollover feature. I am going to quit this session and restart a new one because have done massive changes in the engine and added some features. The bot is now also being traded with real money. Best of luck to these fund managers with balls! ) I might also start real trading myself: i am just studying a folio with "micro" futures. ;-) Tom <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3103812" width="1000" />
I have restarted trading. I have changed the so called "ct t ls e" strategy (need to find some better name ) to anticipate possible reversion. These 3 days have been ideal for trading (expect that yesterday there was a technical problem with IB paper trading datafeed, which now has been solved by IB staff). The bot has taken advantage of favorable trading conditions. I had the idea to add a couple of new good strategy "rules" (will discuss in next post). <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3108932" width= "1000" /> Tom
<b>Discretionality and autotrading</b> Hi friends, After watching for a while the most recent strategies with "trenders" and the "extreme entry" setting, i have to say that the latter conceptual device seems effective, at least after some time has elapsed, to reduce the possibility of getting in trouble with large drawdowns (dd). Actually when i introduced it, i probably overlooked it's usefulness. In fact, as trading progresses, the newest BUYs will alway be executed at the lowest places, and the SELLs at the highest places. This is a good countermeasure to avoid, as time goes by, taking "kamikaze" position ;-) which may cause large dd; on the other hand also trades become rarer and more prudential. Imagine that you are trading and, at a certain point, you get close to the <b> historical maximum</b>. Would you feel confortable repeatedly taking a LONG position? We might take a long position and pull it out with scalps several times. But it's not unlikely that one of this long position will be the start of a long drawdown/hedging effort while the prices plunge down. So the question is: <b>why should we put ourself in such a potentially risky situation, when within a folio there are plenty of other opportunities to make money ?</b> Isn't Better to stay out of troubles ? The "extreme entry" rule contributes to avoid this situation. BUT, what about we are <b>just starting out</b> and the bot has no range information ? Is it right that we let it start totally <b>blind </b>? We might be near an hystorical minimum and the bot starts out with short positions. Would we like that ? We, as traders, for sure might have some range information which we might want to somehow convey to the bot for a more "meaningful" startup. Clearly, anything can happen, including going further down from an hystorical minimum, but if we can give some "suggestion" to the bot, should we not contemplate this possibility ? So this brings out the issue of <b>using the bot "discretionarily"</b>. Discretion has certainly a role even when using a bot. Infact, apart the strategy selection, the <b>selection of the folio</b> is another important area where we exercise discretion. And certainly, if we can stay out of trouble, by possibly avoiding instruments which are expected to have some wild and deep move, it may be a good idea. So, in order to give the trader the possibility to "suggest" entry zones (in case he wishes to) to the bot (to the bot, long or short initial entry does not really matter) i have added a feature to provide entry bounds. I have actually been discussing this new feature yesterday with one of the fund manager currently trading with real money and he has actually given smart suggestions such as also to represent the bounds <b>visually</b> in the trading window, so that it would be easier to understand immediately their position, relative to the price. I have quickly implemented this feature, which seems promising, and this is a screenshot of prototypical implementation (bounds can also be repositioned immediately by simply clicking on prices). Let me know your thoughts or suggestions ... Tom <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3110206" width="1200" />
... trading of the <b>mini folio</b> is enjoying so far a very lucky run with more than 24K in less than 4 days. I have started also a <b>"micro" folio</b>, to take a look a these kind of instruments too. They might be a good option for relatively small capitals. I have been able to identify, so far, some 7 "micros" which seem suitable for autotrading. Let me know if anyone knows some more candidates worth testing. <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3110828" width="1200" /> Tom
Mini folio still ok. It has recovered to 28K (8 days trading) after some dd, mainly due to the still crazy SI. It took a couple of stops in the attempt of reversing in the zone around 36.50, but price pulled back. Various instruments are going to expire soon, i will probably start a new instance, instead of bothering to roll over. In the meantime I have done some further change to strategy rules (an entry blocker for countertrending entries). <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3115102" width="1200" /> I am also running a parallel run on a microfolio with 7 instruments M6A FUT 201103 GLOBEX 10000 AUD.USD Forex M6B FUT 201103 GLOBEX 6250 GBP.USD Forex M6E FUT 201103 GLOBEX 12500 EUR.USD Forex MSF FUT 201103 GLOBEX 12500 CHF.USD Forex MCD FUT 201103 GLOBEX 10000 CAD.USD Forex MJY FUT 201103 GLOBEX 1250000 JPY.USD Forex MGC FUT 201104 NYMEX 10 E-Micro Gold which is doing nicely (clearly PNL and DD are expected to be in the order of less than 1/10 with micros). About $ 670 in 8 days with little max DD ($ -300) Does anybody have other possible candidates to test ? [i am not seeing any data for next contracts for mcd, mjy, msf (??) ] Tom
Hi Tom, That's not surprising, as micro have very low volumes and FX futures usualy roll the last day of the contract (first notice monday 13th for most of them) Did you notice any difference between folio in trading the mini or the micro FX contract? Micro are generally 1 or 2 tick wider.
Hi unco, thanks a lot for the picture. As to the shape, the price curves of mini and micro contracts they seem to me "practically coincident". Spread for "micros" seems relatively larger (although, clearly, much smaller in absolute value) and values ranging from 1 to 4 are pretty common. Commissions are also relatively higher, i think (i am seeing a figure of 2.4 per contract (?) ). Probably also margins are relatively higher. Well, clearly trading with smaller leverage it should be normal to have trading expenses relatively higher. Pretty similar to the "savings" you make when buying a "family pack" at the grocery store ;-)) Anyway, i think that for relatively <b>small capitals</b> or for people who wish to <b>start autotrading</b> or just get a feeling about it, these micros may not be bad at all. After all the bot does not "care" much of spread and commissions, as it includes them in the scalps. Here we have a tickvalue of just $1 and i used a scalp of $25. So the whole game is scaled down by 1/10. Thanks for the data info: i was suspecting that ) strange though that we need to wait for the very last day (the one already receiving some data are: m6a, m6b, m6e, clearly with large spreads). Tom
I have restarted with a new folio. This is a detailed report of trades (autogenerated). [Edit: don't see the attachment, PM me in case interested in single trades: i will email it]
Hmm, i am not sure why the zipped attachment is not showing in the previous post. Anyway here is a pictorial summary (1.25 days): <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3116502" width="1200"> I have not defined "discretionary bounds" (latest feature introduced we have been discussing), as for now i am more interested in examining the behavior for worst case drawdowns (dd) and deep moves. Clearly, the discretionary bounds (defined at instrument level) may help greatly avoiding troubles and deep dd. Tom