Trading with automation (with IB)

[fullautotrading]

So we have reached a new weekend. Still up +230K, but with some "investment" going on with NG, CL and ZL.

As to folio selection, here is a list of instruments that do generally ok, imho as far as scalping/hedging is concerned. The reason for suitability is a combination of abundance of rapid price "fluctuations", "good but not extreme" volatility (say, suitable price microstructure), liquidity, presence of "usable" options:

CL FUT 201602 NYMEX 1000 Light Sweet Crude Oil
ES FUT 201603 GLOBEX 50 E-mini S&P 500
NKD FUT 201603 GLOBEX 5 Dollar Denominated Nikkei 225 Index
SI FUT 201603 NYMEX 5000 NYMEX Silver Index

(I am not liking NG much, for the excessive contango and the "problematic" options.)

On the ETFs front (to be played in the "semi-automated" mode previously described), I feel comfortable with the following ones:

UVXY STK SMART PROSHARES ULTRA VIX ST FUTUR
TZA STK SMART DIREXION DLY SM CAP BEAR 3X
SVXY STK SMART PROSHARES SHORT VIX ST FUTUR
DGAZ STK SMART VELOCITYSHARES 3X INVERSE NA
NUGT STK SMART DIREXION GOLD MINERS BULL 3X
ERY STK SMART_USD DIREXION DLY ENERGY BEAR 3X
UCO STK SMART PROSHRE ULT DJ-UBS CRUDE OIL

(these are all surfed short, except svxy which is ridden long. vxx would be similar to uvxy, except with lower (half) volatility)

 

[fullautotrading]

Not much happening on Monday and today: PNL remains fluctuating around +230K (196 days so far and over 5,500 lmt fills). I have enabled 3 layers for CL in order to start loading up again, since the maint. margin usage is quite low (just 225K). Most of the investment is now on a few layers of NG, and we need to wait for most players to close, to turn them profitable.

In the meantime, I have been making some small refinements to the scalping/hedging game, the sizing logic, and also risen a bit the order frequency, to gather some more action.

 

[fullautotrading]

Tuesday/Wednesday has brought some more gains and we currently (after 197 solar days) are up to +302K (also the G-L curve has risen considerably). Maint. margin usage below 200K. The situation with NG is pretty stationary. SI (and NUGT) has been moving, as expected. According I have almost completely "unloaded" NUGT. I have also been shorting a few put CL options.

[In hindsight, what we have been missing in the past was just a careful consideration of the instruments' long term "drifts" (and therefore suitable layer constraints definitions and layer overlay), more careful consideration of specific instrument volatility, and suitable risk capital. Once that is fixed, the scalping/hedging game (which includes the various stop and rollover "recovery mechanisms") suffices, if well defined by the fund manager, to provide a tangible edge.]

 

[fullautotrading]

First day, after the weekend (and the Easter holiday). Not much happening apart "choppiness", which of course is beneficial in our approach.
We remain near +300K (according to IB a 19.10% up, in about 202 days). We have some "investment" going on on CL, NG, SI and ZL, but it's not causing much DD because we are using actually few layers and little resources (maint. margin is just 293K):

 

[Alex27]

What software did you use to automate?

 

[fullautotrading]

What software did you use to automate?

Hi Alex27,

I currently take care of all the automation and related functionalities, writing all the code lines I need (this is a .NET application), and "talking" to IB APIs for the order execution.

[ As the approach is original, based on "superposed players" and "overlaid layers" concepts (in addition to several other innovations I have been introducing in time), it could not be implemented on different existing architectures. ]

 

[fullautotrading]

Today we have officially broken the +20% threshold, with IB API reporting:

Δ NetLiquidation 310,088.12 USD +20.33%

solar days elapsed are 202, with 5,725 lmt fills, over 20K comms, and over 12K interest.

Apart, and with, the SCO "blunder", this is illustration has clarified several conceptual points, and in particular the importance to take into careful consideration the various instrument "drifts" (in a broad sense). In essence, each instrument has to be considered carefully, with adjustments taking into considerations long-term factors such as "drifts", contango, backwardation, decay, daily rebalancing, as well its volatility and the "codirection" relationships with other instruments in the folio, and so on.

This appears to be fundamental, because no matter how much we scalp away and trading information we "preserve", if then we cannot use it (for instance because a drift has cast away most players), it's just the same as "throwing" it away (as it normally happens with the "ordinary" stop concept, and the common rollover procedures, when single, "nonoverlapping" trades are used).

At this point, we might start another new journal from scratch, to confirm what we have learned and realized to far, and also to introduce some variations on the way to overlay the layers and the scalping/hedging game definition (that is the way to "superpose" the players on a single layer).

 

[fullautotrading]

If interested on all the trades in details and all the charges, here is a zip file with the official broker report, to latest available date (attached).

[ For some reason, just like it happened the last time, the initial date selectable for reporting is "shifted ahead" again. Unless I am missing something obvious, this looks like some sort of bug (?) (or at least limitation) of the IB reporting system. It does not seem right not having the possibility to select always the same start date. Maybe you guys have more info about that. ]

 

[radex78]

Very complicated in my view, so far I am never use automatic tading software and only trade manually, too many code to created robot, and my brain limited to understand, so my mindset is making trade being simple

 

[fullautotrading]

Very complicated in my view, so far I am never use automatic trading software and only trade manually, too many code to created robot, and my brain limited to understand, so my mindset is making trade being simple

Hi radex78,

the concept of "complexity" is a very relative one. For instance, when proposed special relativity also seemed complex, but now any child apparently can explain it. Similarly, if you look at the Menichelli (

) exercise in Tokio, it was obviously something for its times (Franco got the gold medal). Now, in my gym, every 10 years old can do that, and way much beyond.

So, simple/complex are very relative concepts. In my view, this is the minimal effort required to meaningfully extract $$$ from the mkt, in the context where I operate. In fact, personally I always strive for maximum (relative) simplicity, by continuously remaking or removing what looks redundant or unnecessary (as Albert put it : "everything should be made as simple as possible, but not simpler" ).

Anyway, when limited to few instruments, it's pretty fine to trade "manually", as you suggest, if one has the time to devote to it. Clearly, it also depends a lot on the approach one takes.

My perspective is that one of perpetual fund management with good capital, with massive scalping/hedging action (which is essentially producing the required edge), and in this context a good degree of automation helps a lot. Also, even though it may look "arcane" on an html trade report, visually, when followed "on screen", it is quite easy to understand the action and what is going on. Anyway, even within this framework, I still to have a lot of tasks left to discretion (although obviously anything can be automated. I just try to have a "reasonable balance" between the various needs and priorities.)