Trading with automation (with IB)

Finally back after a busy day, at least I find the application has been working fine, hedging well the moves.

The "newcomer" NQZ5 has been doing some profits too with the initial 2 layers, while ESZ5 is leading along with NKDZ5 and SIZ5 (it was started first anyway) due to a large number of nice fluctuations. The G-L curve has reached $42K and is "pulling up" well the PNL curve (currently around $19K).

PNL_15.png
 
Last edited:
To gain some more insight in the current scalping/hedging action, we might take a look at the orders placed on the (currently) most favorable layer, which is one of the layers of CLX5.

CL_5.png


This layer is showing a total of 17 fills (made in 11 days). So, 8 players completed their "life cycle" and 1 (short player) is currently open (in the green circle, on the right). As you can see (horizontal green lines represent a 1% grid of prices), the fills are decently spaced up (this can be set by the user) and the "game" is not very high frequency in this case.
[Overall, we made so far a total of 168 lmt orders ($410.16 commissions) in about 21 days.]
 
Last edited:
Another fine looking morning. It appears that the application has been busy during the night (we have now 177 fills in total, and paid $431.16 comms):

PNL_16.png


So far, scalping and "bounding" the unrealized (that would be the "gap" between the G-L and the PNL in the bottom picture) through superposed hedging players, overlying layers with long-short position constraints, and limiting the max position on a layer, seems to have been working fine for these 21 days. The relatively extensive scalping action has been "pulling up" the G-L while, the "gap" wrt the PNL has been successfully kept within acceptable values (currently about 18K). Let's see if there are some larger moves, to explore how it behaves, and if we need to make adjustments somewhere.
 
Last edited:
I am adding another 2 (long/short) layers of a new instrument

Code: 
ZM FUT 201512 ECBOT 100 Soybean Meal Futures  [ ZM   DEC 15 ]

ZM_1.png


The first ("long") layer almost immediately shot a long player. The other one ("short" layer) is still just quietly collecting and crunching tickdata.
 
Today we had pretty wild swings of the PNL (I wasn't here to watch them), with fluctuations from $30K to $23K. These were mostly caused by NG, ES, NQ and the agricultural stuff. Apparently, we did not get hurt too much after all: we remain with $25K PNL, and the G-L has raised to $47K. Current maint. margin is around $43K, with 9 instruments active. Fills have gone from 177 (this morning situation) to 206 with $498.03 comms.

PNL_17.png
 
Last edited:
This morning the situation is looking particularly good, with PNL pulled up to $30K. During the night apparently the program has been working, and the fills have gone from 206 (last night) to 216 (right now.) Current maint. margin usage is about $45K. Almost 23 solar days of trading, comms: $525.60. The return on margin is actually momentarily tremendous. We will see how it develops and if it can actually hold on.

PNL_18.png
 
Today we had again some DD, but nothing worrying. It's mostly NGZ5 which is pulling down and the agricultural stuff. On the other hand CL has been hedging with some gain.
This time, however, the DD resulted in a marked improvement of the G-L which is shooting at level $53K (this is due to the fact that the losses on "one side" generally result in some gains on the "other side"). The fills are now 240, so 24 more since this morning, comms: $583.72, maint. margin around $26K.

PNL_19.png
 
Not much happening tonight. Just 3 fills and no significant changes in our situation.
To entertain ourselves, we can take a look at the orders executed on the (currently) most profitable layer ESZ5 (before we have seen CLX5), to see the result of the automated scalping action:

ES_10.png


Here we have 21 fills (comms $46.46). 10 players have completed their "life cycle", thus giving 10 "scalps", while one long player is still open at level 1,944.00.
 
Last edited:
We have been using a very small portion of the $$$ in the account so far (current maint. margin is just $36K).
To use a bit more, I am going to increase the "packet size" of CLX5 to 2 contracts, and also add new layers for YM (so, in practice nearly "replicating" ES, due to the very high correlation). [Gradually we will proceed to increase the other packet sizes too.]

Note that is is possible to create several instances (clients) of the application working on the same account (10 clients per user was an IB limit last time I checked, unless they changed the policy). In such a case, then it would be advisable not to exaggerate with the number of layers for each application instance because it is easier to follow and supervise. And also, it is probably convenient not to use layers for the same instrument on different instances, because it is a bit easier to make the rollovers and check the synchronization with the underlying account. A fund manager with large capital, could actually use a single instance for all the layers of the same instruments (those could be started in long/short pairs at different price levels).
 
Last edited:
Wow. I am going to leave for several hours, but I can't help sharing the spectacular fall on the new YM layers I am looking at, just after 3 hours running:

YM_1.png


also CL and SI have spiked crazily:

SI_4.png
 
You must log in or register to reply here.
Back
Top