Trading algorithmically a folio without stops (with IB), real $$$
- Thread starter fullautotrading
- Start date
Nice one, Tom keep up the performance posting
Enjoying watching the journey
Sure mickson, as always
<b>App updates</b>
A quick update about application development.
In the weekend I have also introduced a "color code" for the games, and defined 3 main (example) games. For now they are called:
Shield, Bias- , Shield+
[ The application will initially default, for each (known) instrument, to some of these example games (based on a classification based on experience or managers' suggestions). But the manager, of course, can override everything anytime. ]
The general idea is that the first one (<b>Shield</b>) is a general purpose scalping/hedging game, with massive hedging action, for instruments which in the long term do not seem to show significant a "drift", or that in any case one wishes to trade "symmetrically" (of course, one could be wanting to trade "asymmetrically" even a perfectly mean reverting instrument, for whatever reasons, eg. strong contango, dividends, law constraints (pension funds) or whatever).
<b>Bias-</b>, which is the one I have been using for the <b>"inverse" ETFs</b> (like tza, etc.), takes care of the case of special instruments with strong negative drift (like leveraged inverse ETFs). This is of course necessary, because even if the <b>statistical drift</b> we build with the <b>player superposition</b> is pretty strong, fighting against the "structural"/"mathematical" drift of the leveraged ETFs would be really too much to take (and, indeed, pretty stupid) , and in any case you don't want to accumulate long term positions against a "structural" drift. (Note that here "inverse" is mostly to be intended wrt to mkt, and not in the sense of "bear" ETF. So for instance, in this sense, VXX will be an "inverse", as some others, which may even have the word "bull" in their denomination.)
<b>Shield+</b>. Is a similar ideas to the above, but to take care of long term upward drift ("good" stocks, mkt indexes, and so on).
In addition to the "biased" games, which simply provide some forms of asymmetry in "order sizing" and directional "entry styles", it is also possible to enforce the <b>"position constraints"</b>.
Note he that a "position constraint" does not prevent entries in both directions, as of course you can for instance remain "positive" even while massively "counter scalping" (with sell orders) against an existing long position.
[There is also, in practice, the possibility to force unidirectional entries only, but we can disregard it for now this, as, in principle, it would be obviously a blatant violation of the overall approach we are currently demonstrating in this test.]
A quick midweek update.
Investing again in the small caps moves (note the decrease of TZA, TNA pnls and some other mkt related instruments, by comparing with the previous situation), and partially hedging with the positions in metals we had, which mitigate the drawdown. I have also added a few ETFs (NUGT, GDXJ, USLV) which will possibly "replace" some of the ETFs I will probably "throw away" (ZSL, DGLD, JDST), and still watching a few stocks.
Investing again in the small caps moves (note the decrease of TZA, TNA pnls and some other mkt related instruments, by comparing with the previous situation), and partially hedging with the positions in metals we had, which mitigate the drawdown. I have also added a few ETFs (NUGT, GDXJ, USLV) which will possibly "replace" some of the ETFs I will probably "throw away" (ZSL, DGLD, JDST), and still watching a few stocks.
End of the week.
Most of the activity was with TZA and TNA (small caps) again, where we could close a few players (but still holding a lot of positions). Added a few new ETFs to watch.
Still watching some stocks, to see if there is enough "activity" to activate the s/h game. Current margin usage is around 160K, so pretty low at the moment. Drift is ok. All instruments under control. Next week we should intensify the activity.
PS
ETN have removed from portfolio margining eligibility. So our margins could rise because of that. Check out this memo (pdf) ("INELIGIBLE PRODUCT REMOVAL FROM CPM THEORETICAL OUTPUT FILE" from OCC).
Most of the activity was with TZA and TNA (small caps) again, where we could close a few players (but still holding a lot of positions). Added a few new ETFs to watch.
Still watching some stocks, to see if there is enough "activity" to activate the s/h game. Current margin usage is around 160K, so pretty low at the moment. Drift is ok. All instruments under control. Next week we should intensify the activity.
PS
ETN have removed from portfolio margining eligibility. So our margins could rise because of that. Check out this memo (pdf) ("INELIGIBLE PRODUCT REMOVAL FROM CPM THEORETICAL OUTPUT FILE" from OCC).
The week has started with another day dominated by small caps ETFs. Closed several players mostly on TZA and TNA, pushing the PNL to a new peak, exceeding 31K. Closing the session at 28.4K. Margin usage is still relatively low, about 140K.
I am still watching a few stocks too, including AAPL, which after the upcoming 7-1 split could probably be a candidate.
I am still watching a few stocks too, including AAPL, which after the upcoming 7-1 split could probably be a candidate.
Here is a detail of the <b>order cloud</b> of the current top PNL layer, which is of course TZA (small caps). (It has a "short position constraint" and currently playing a "short-biased" game.)
As usual, the top picture represents <b>all</b> orders (blue=buy, red=sell), while the bottom picture shows only the currently open players (that is, it excludes those which already closed with some profit.)
The open players, as we now well at his point, clearly represented a form of "stops" (protective players), which will (hopefully) be "recycled" sometime in the future and, therefore, will contribute to the "statistical drift".
(If you have a very sharp eye, you might note that sometimes I "overload" some move ("projected entry"), if I am near my pc. I can now afford that as the "sensitive" settings for SDX, cause an immediate automatic close of the "pyramiding" players, at the smallest sign of price reversal.)
[As usual, in case anyone wants to see the complete record of all orders, of the entire folio, for more insight and study, just drop me a message (I will email you a zipped excel sheet with the detail of all single executions, about 8,200 fills currently)].
As usual, the top picture represents <b>all</b> orders (blue=buy, red=sell), while the bottom picture shows only the currently open players (that is, it excludes those which already closed with some profit.)
The open players, as we now well at his point, clearly represented a form of "stops" (protective players), which will (hopefully) be "recycled" sometime in the future and, therefore, will contribute to the "statistical drift".
(If you have a very sharp eye, you might note that sometimes I "overload" some move ("projected entry"), if I am near my pc. I can now afford that as the "sensitive" settings for SDX, cause an immediate automatic close of the "pyramiding" players, at the smallest sign of price reversal.)
[As usual, in case anyone wants to see the complete record of all orders, of the entire folio, for more insight and study, just drop me a message (I will email you a zipped excel sheet with the detail of all single executions, about 8,200 fills currently)].
End of the week.
Nothing extraordinary happening apart the activity in small caps we have got used to in the last weeks. Scalping quietly. Margin usage is just 137K at the moment (clearly thanks to "portfolio margining" mode which eats up a lot less than RegT, especially with this kind of folio, and even though the ETNs have been made ineligible). PNL about 27.5K, executed 8.7K fills with almost 8K in commissions. "Drift" ok, around 13%. All instruments are under control.
Still watching a few stocks, but I am not convinced at all. I see too few of the kind of moves which are good for scalping/hedging in these instruments. It's possible that if we start trading them we tight up significant margins without much good scalping activity.
Nothing extraordinary happening apart the activity in small caps we have got used to in the last weeks. Scalping quietly. Margin usage is just 137K at the moment (clearly thanks to "portfolio margining" mode which eats up a lot less than RegT, especially with this kind of folio, and even though the ETNs have been made ineligible). PNL about 27.5K, executed 8.7K fills with almost 8K in commissions. "Drift" ok, around 13%. All instruments are under control.
Still watching a few stocks, but I am not convinced at all. I see too few of the kind of moves which are good for scalping/hedging in these instruments. It's possible that if we start trading them we tight up significant margins without much good scalping activity.
End of this week, which has started with an holiday. Mkt pushing up all the time and we have been loading up on metals. Some minor problems and disappointments with JDST and DGLD, which have been not shortable almost all the time (DGLD still is) and therefore were essentially "locked". So I had to compensate by overloading a bit their bull "counterparts" (NUGT, GDXJ, ...) to hedge the nasty upward moves. Still, the DD has been quite significant (PNL to 14.5K).
So they have officially entered my black list of the"to be removed" as soon as possible from the folio, which now contains: DGLD, JDST, ZSL, SLV. (DGLD practically untradable, as always not shortable. SLV too small/few fluctuations.)
Spectacular last 5 minutes with the miners <b>flash-crashing</b> shamelessly in seconds. See right end of JDST picture:
In the meantime, I have activated <b>WMT</b>, our "first" stock. The game loaded for the stock is the one called "Shield+" (so, long bias) with an additional "long only" constraint. So our stock trading will look like a sort of Buy and hold + hedging. A "miniature WB style approach") with much additional hedging during moves down.
So they have officially entered my black list of the"to be removed" as soon as possible from the folio, which now contains: DGLD, JDST, ZSL, SLV. (DGLD practically untradable, as always not shortable. SLV too small/few fluctuations.)
Spectacular last 5 minutes with the miners <b>flash-crashing</b> shamelessly in seconds. See right end of JDST picture:
In the meantime, I have activated <b>WMT</b>, our "first" stock. The game loaded for the stock is the one called "Shield+" (so, long bias) with an additional "long only" constraint. So our stock trading will look like a sort of Buy and hold + hedging. A "miniature WB style approach"
) with much additional hedging during moves down.Midweek update. Nothing special happening. Some quiet scalping/hedging ("G-L" up to 112K, PNL currently around 14K, decent drift). Mkt still pulling up.
I have been working on a complete reorganization of the scalping/hedging game rules. Now that it's much more clear what are all the mechanisms which are necessary and useful, it's a good time to reorganize the game rules in a more intuitive and flexible interface.
Probably next Monday I should be able to deliver this major update.
I have been working on a complete reorganization of the scalping/hedging game rules. Now that it's much more clear what are all the mechanisms which are necessary and useful, it's a good time to reorganize the game rules in a more intuitive and flexible interface.
Probably next Monday I should be able to deliver this major update.
Here is some anticipation on the game reorganization (mostly done, currently just testing on some test accounts of mine with real $$$)
The reorganization consists of differentiating the behaviors within 5 categories, 4 of which we already know (we have talked about them in a previous post), and the last one ("Positive move overload") is a new special "override" to take advantage of fast favorable moves of price.
The opportunity to add it, derives mainly by experience (I use a "bottom-up" process to build rules, introducing only the <b>strictly necessary</b> flexibility). Since I found often myself intervening on fast moves with temporary "overload", it seemed nice to automate this behavior, and make it a feature.
For each of the categories, the rules are the same and can be set independently.
This way we have a very good control on the order cloud formation, and several degrees of freedom to suit its development to totally different classes of instruments.
The reorganization consists of differentiating the behaviors within 5 categories, 4 of which we already know (we have talked about them in a previous post), and the last one ("Positive move overload") is a new special "override" to take advantage of fast favorable moves of price.
The opportunity to add it, derives mainly by experience (I use a "bottom-up" process to build rules, introducing only the <b>strictly necessary</b> flexibility). Since I found often myself intervening on fast moves with temporary "overload", it seemed nice to automate this behavior, and make it a feature.
For each of the categories, the rules are the same and can be set independently.
This way we have a very good control on the order cloud formation, and several degrees of freedom to suit its development to totally different classes of instruments.