Let's take advantage of the holidays to take a look at the global report created by the broker for this trading session to the end of the year (Happy New Year, btw). The application has been running about 116 solar days now, with over 2,300 filled lmt orders: [For those interested in examining all the single orders and trading details, I am also attaching below the complete report (zipped file with the download of the html report.) ] Strategy summary As discussed, in this test we have been experimenting the following approach: For each instrument, we have been considering price levels on which to create "trading corridors". For each corridor, we have opened 3 layers (2 of which with opposite "long/short constraints", and one used to create further hedge in case the first two layers align in the same direction.) The above games would mainly either capture price fluctuations, or wait for larger local price reversions. Then, we have also integrated with "enqueued orders" (orders automatically triggered on certain prices) on ETFs, in order to capture also the drifting component of some instruments. The general crucial conceptual point has been the preservation of the "trading information" through the concept of players, which essentially allows us to recover the "stops" ("hedging players"), and the information transfer (when we needed to roll over instruments.) Note that the largest number of corridors we needed to consider in the period is just 3. In a next experiment, I would like to try some different arrangement of the layers. In fact, I feel that, for drifting instrument, having 1 or 2 layers forced against the drift, is probably something we may want to avoid.
While CL is still in an "investment" phase, we have anyway a good start with the new year, with PNL ramped up to +164K ( 171,039.08 according to the broker): I wish to make some very slight conceptual change in the higher "layer management", by separating the instrument with a "structural drift" from those which usually tend to be more "erratic". This distinction can clearly be rather subject to the the fund management preferences. Let's, just for now, simply consider instruments like ES, NKD and similar mkt indices, like instruments which may be affected by a "structural drift", while other instruments (eg. commodities) will fall more in the more "erratic" group. (Clearly, a deeper fundamental analysis or other considerations, might also suggest the fund manager to treat differently a given instrument.) For the more "erratic" instruments, we will manage the 3 layers on each corridor as follows: A. Layer1 - Long only (mild) constraint B. Layer2 - Short only (mild) constraint B. Layer3 - Activated with opposite mild (constraint) in case the above 2 layers align in the same direction. (The layer uses no constraint in case layer 1 and 2 do not have same sign position.) So, essentially, just what we have been doing up to now, but allowing, on layer 3, temporary "violation" of the constraint to close the players. For the "drifting" instrument (say, for instance, drift up), let's use this variant: A. Layer1 - Long only (mild) constraint B. Layer2 - No constraint B. Layer3 - Activated with opposite mild (constraint) in case the above 2 layers align in the same direction Let's remember that our constraints are "mild" and not "absolute", in the sense that they can be "violated" for at least 2 reasons: - A player needs to close (and doing so creates a constraint violation) - A layer needs "protection", and (opposite side) protective players may trigger, to stop the "bleeding" and this is controlled trough the "instrument settings". Through the games we control also the maximum exposure, and recall that so far we have bee using a max exposure of 2 contracts for each "price corridor". For the driting ETFs, we will proceed as we have been doing so far, by placing "enqueued orders" to capture "extreme" spikes, and thus provide further hedging action against the mkt drift, contango/backwardation. Mostly, we will use the following (but we may add some more): SCO, UVXY, SVXY, ERY, TZA,DGAZ. In essence, we will aim to capture both "price fluctuations" and long term reversions (through the scalping games), and also some of the components due to drifts, decay, contango/backwardation, wherever possible.
Two instruments which have been scalping rather well, in the given period, are ES and NKD. To take a look at the scalping/hedging action, here are their most productive layers: ES (in the "upper corridor") NKD (in a lower "corridor") I am gradually "shutting down" YM and NQ, as there seems to be no much point having these "replicas".
Today we had a significant move down with crude oil (and therefore a spike up with SCO), so our "investment" in this area has grown to significant levels. PNL has been "absorbed" (draw down) in large measure, setting back to just 20K-28K. This is, by far, the largest single-day DD we have seen so far. We have therefore a good position on oil, which we can anyway afford with our capital and previous gains. (I don't dislike being in this position actually.)
Is your account still active? How has this account been impacted by the current pullback in oil price?
Hello OldSpec. Yes, of course the show goes on . As anticipated in the previous posts, we had a large position with crude oil both with CL (long) and SCO (short), so we are currently in a phase of significant DD due to this "load". I have also a bit "overloaded" the position with SCO, as the price seems attractive, and we have enough risk capital (current maint. margin about 700K), so we are currently about -21% underwater due to this new "investment". My personal expectation is the mkt indexes to bounce back first, and, more slowly, the oil to price to reverse with a lot of fluctuations. I have also added another 3 CL layers (long/short), to scalp the lower price "corridor".
You are welcome OldSpec. During the DD or "load up" phases I tend to do naturally a slower update activity due to the reason that, in time, I have experienced that, in these phases, it's better to get a bit "detached" from the trading activity, to avoid: 1) the temptation to "mess up" with the plan (in times like these, people start freaking out, and the weirdest "predictions" start flooding the web: so just better to ignore the whole circus), 2) mood swings: despite automation, in the DD or "investment" phases the mood of the fund manager can be significantly affected, and often goes down along the PNL curve , and that is not much healthy. It also helps to have other interests. Someone in another thread suggested agricultural activities , which actually is not a bad idea in general, except that now is winter (here) and I can't take care of my garden. Seriously, one can focus on his research, and other activities (personally, I also like piano composition and gymnastics, which, I believe, do have a very good synergy with the trading/development activity).
The situation is now looking slightly better with PNL, while still underwater a 12%. CL has bounced back a bit and yesterday I have also rolled our CL layers to CLJ6. SCO has suffered a very severe DD, as I have, objectively, "overloaded" a bit too much, and too "early". If it drops some more, I will be "forgiven" and I will begin to "discharge" the position. I have done a couple mistakes in the management of the "enqueued" orders on this ETF. A first mistake was not closing it when it was in good profit, and a second mistake is having made "too close" enqueued orders, so that when it really spiked up for good, I was already out of resources. Instead, the automated part did well and according to the programmed game. Anyway, lesson for next time .
We close the week with a monster "recovery" (we made back about 500K in the last few days), mostly due to the SCO, CL, ES, YM drawdown being reabsorbed. We are still "underwater", at -8.75%. Next week, we need first of all to rollover NG, which is subject to the Physical Delivery Rules. I have missed a good opportunity on SCO, due to some mistakes, and lack of funds. Will do better next time In the meantime, I have also been doing some improvements on the software side. In particular the "clonation" of the current instrument layer settings to the "new" instrument layer on rollover ("trading information transfer"), which is an extremely useful feature, and I am still wondering how could I have missed to think about it so far .