My last thread had become long. I am starting a new fresh one with a new folio of futures. We have been watching a larger folio ( the so-called "KS") to get some idea of the "personalities" of the different instruments. (If one wants to see past history of our discussions and strategy development, the most recent thread is <a href="http://www.elitetrader.com/vb/showthread.php?s=&threadid=208265&perpage=6&pagenumber=23" >this one</a> ) In a recent free-seminar announcement from IB i read: <i>"It's critical for traders to understand that when they take a short-term position in a particular market, they are actually trading the "personality" of that market and not just the market itself. Each market is made up of individual traders and systems that trade in a particular way."</i> And i think i could not agree more. This folio here is a more meaningful diversified ensemble of instruments, including Currencies, Energy, Metals, Indexes and Agricultural. The picture reflects the last trading day. We trade 1 contract at a time (very prudently), as is our style. A large investor would be overlaying multiple delayed folios, in order to achieve higher frequencies. [no need to say that i am in no way associated with IB ;-) ] Tom <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3209488" width="1200" />
You may note that perhaps this folio has too much oil related product in it. But that's just because i prefer things to be well lubricated ;-)) Joke apart, this folio does have an excess of energy related product. But that is because personally i like energy futures, because of their large moves with large retracements. (So it reflects a personal bias.) Looking, at the "$ speed" metric (picture below), we have been discussing, in previous posts, we have the following realtime situation (each column represents how much in $ per hours a given instrument moves): <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3209507" width="1200" />
Tom, every day your Bot becomes stronger. I appreciate all of your efforts and look forward to continued learning.
thank you my friends, i work on performance improvement on a daily basis, and your encouragement is good fuel to carry out this quest: this project is growing everyday especially thanks to your your expertise and useful discussion and many features, each one, carry the name of a smart person who proposed (or contributed) to it ;-) Tom
So far, all proceeding without big problems. Most rewarding has been GOLD (followed by CL, YM, and RB). Most disappointing, the agricultural stuff. [Looks like they just dont have the "right" kind of fluctuations, or the right "personality" ;-)) ] Here are the single trades of GC: <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3211037" width="1200" />
Spectacular oil drop (right side of picture) in just few minutes! Fortunately wasn't able to hurt much the bot game at this point (which has already established some trading bounds). The chart shows the current folio situation (corn and a few others are in dd/investment). <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3211400" width="1200" />
Here is the situation now. The cyan arrows indicate 2 (additional) <b>"clones"</b> (for CL and HO), whose usage i will explain better in a next post. The corn was going too far, i have halted it with a "discretionary stop", for later resume. I am probably paying my ignorance about this instrument. Knowing the market, we would probably have used a "forced directional entry" (short) from the beginning, and realized a nice profit the way dow. Anyway, lesson for next time. (Not bad anyway, 20K in almost 4 days. We can't really complain.) Tom <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3212496" width="1200" />
Current situation (almost 4 days). I am now going to explain the <b>"clone instrument"</b> concept along with some suggestions. I hope that in a while i will receive *your* suggestions on how to best use them. <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3212621" width="1200" /> Tom
This trading approach relies essentially on risk spreading and diversification. This is attained by two means (in addition to the "folio" concept): - <b>Multiple folio instances</b> [that is the capability to start multiple instances of the same or different folios on the same or multiple accounts] - <b>Multiple instrument instances</b> within the same folio ("<b>clones</b>") The above can be used mainly for 2 purposes: - <b>Overloading</b> (increase the position of some instruments) - <b>Hedging</b> (use a folio or another instance of an instrument to hedge existing positions) The obvious advantage of this approach is that, through overlaying, one can reach any trading frequency and any investment size, without liquidity or execution problems, and not even being obviosly detected. In addition, the HF is not attained through a tiny timeframe, but through overlaying. Which allows all players to have a lower frequency. Clearly, a much safer approach (which does not even require expensive infrastructure). [It's pretty obvious that allocating large positions on a small timeframe game, can lead, sooner or later to some catastrophe. Not to mention that the game would be much less efficient, from the strategic point of view] Now let's see the "clone" concept and some practical use of it. For now, look at the following picture, which has all CL trades. Note after the crash, the sideways trajectory (will continue in a next post). <img src="http://www.elitetrader.com/vb/attachment.php?s=&postid=3212636" width="1200" />
So let's say something about "clones". A <b>"clone"</b> (instrument) is just another copy of a given instrument, which, being activated after the original, will perform entries which are out of sync with the original. When started, a clone (as any instrument), can (<b>optionally</b>) be instructed to <b>"force" a long or short entry</b>. When and how to activate clones is a <b>discretionary</b> thing, which clearly affects the immediate results of our trading. [ I have been asked several times: why do you allow discretionary intervention within a full automated system? The answer is that, while i make sure that the strategy performed has good "statistical properties", there are many times when can introduce strategic variants, without affecting the statistical properties of the strategy itself. So i allow discretionary intervention in such a way that it merely creates a strategic variations, which anyway does not change the general properties of the strategy (say in the space of possible price curves). Let's make a concrete example. For this strategy, not being based on indicators, it is irrelevant (statistically speaking) the direction of the first entry. So if i allow the user the possibility to choose the initial direction, it wont harm (statistically speaking). Similarly, i can allow the user to use discretionary bounds, to allows entries only beyond given prices ("support/resistance"). ] A clone is another example of <b>discretionary intervention</b>. Trading experience will tell the best way to use these discretionary possibilities, and frankly, even myself, being focused more on strategy development than trading, have still a lot to learn. And actually i do expect to learn a lot from the friends and fund managers who are exploring the app. An example of use i figured out is to resume scalping action on "stuck" instruments (after some large price shift). In fact, one thing which might occur (and this is what hurt more this approach) are large price shift with no sign retracement (for a long period). In this case we can start a clone to resume the scalping action on the new price level. This will allow us to scalp while waiting for a retracement. In fact after large moves the price will hit a support and resistance and begin a sideways period, which can be highly beneficial for the bot. Now there is an open question if the clone should force a directional entry or not. My current believe (which may prove wrong) is that it probably should. Let's take for instance the CL example we have seen above. After the large drop, i may decide to take advantage of the sideways oscillations by starting a clone. But i may want to force LONG entries only, because otherwise, in case of reversal i am going to "neutralize" the benefit of the reversal (at least in the first trip back). [As an alternative, perhaps, we could leave the clone free to make any entry, or even force it SHORT, provided that we shut it down (close) when we see signs of retracement.] My current guess is that the clone feature should not be abused, but used parsimoniously. The above CL example is not a real good one, because we should limit the use of the clones, to the cases where the instrument is really "stuck in an investment" phase and shows no sign of retracements. Anyway take it just to give the idea. When used too often, the clone has more an "<b>overloading</b>" effect, than hedging, and somehow an "excessive" (too many or too "close" in time) use would probably conflict with the principle of diversifying risk, and with the folio concept. So we should really use them, for "overloading", when we have good reasons (market insight) to expect the market will have a certain behavior. Let me know your thoughts about this topic... Tom