Robot Trading on futures - ES et al

[Kedwards]

Thanks for the chart ASusilovic

But yeah I think being long premium would help offset the unrealized loss from the robot on the initial leg downward (circled in the chart; also, strong move downward should increase volatility so that will also boost straddle gain), and then on the reverse move higher the robot would get back to normal and hopefully have a large gain.

 

[iuykcif]

Hello mockney.

I am aware of this point. I have discussed this partly in a previous post in this thread and I take
the chance of your question to also partially rectify something.

If you are:

Acct Long 10
Acct Short -5

this is equivalent to:

Acct Long 5
Acct Short 0

This holds true, at any given instant, for 1 trade.

From this one would imply that any algorithm - which is running **independently ** on 2 accounts - can be "projected" into 1 account.

For a sequence of trades, if there is *dependency*, the equivalence holds no more.

The fact is that here the 2 scalpers are *not* running independently, but they have strong dependence. They are, say, "tied".
This may make the algorithm not "projectable" on 1 account.

[At least I could not do it, yet]

** If I could project this algo on 1 account, I would have solved all the margin problems and by now <b>I would probably be richer than a sheick </b> , ahaha )) **


In other words, chances there are that, using 1 account only, one will never be able to match this level of performance.


imho, it is also possible that this might hold in general, i.e. 2 accounts strategies "dominate" 1 account strategies. Which means
that for any strategy S1 run on 1 account, you can find a dominating 2-acct strategy S2. Perhaps a pair of S1 itself + dependence constraints.
(just a little conjecture

I did not embark (yet) in this kind of formal research. Most concern is now in creating the money machine. While a good cash flow is coming in, we can
relax and write some (seemingly) useless academic paper )


Tom

Quote from mockney:

I'm not clear why you need to maintain an open long and short position concurrently when this is the same delta as being flat.

when you take one side off or add one, its just the same as opening a position from flat, is it not? so why not do that and reduce transaction costs for un-necessary trades?

 

[trend2009]

in your chart, what do square and circle stand for?

 

[iuykcif]

Nothing esoteric :

Square = Entry (position before order = 0)
Circle = Exit (position after order = 0)

You guess the pyramid!

Quote from trend2009:

in your chart, what do square and circle stand for?

 

[trend2009]

i explored this idea for a while. it seems the strategy is similar to two accounts that adopt average down strategy separately. in this chart OP provided
(http://www.datatime.eu/public/TradingDiary/YM-2009-09-04-1Hour.gif), there is a moment that OP averaged down on short side while the constraints between the two account is much more than 2 contracts. I guess at the highest point, the drawdown is a lot. luckily, the market moves down, so OP made lots of profits. in case the market keeps up, OP will lose a lot.

thus, profits and hedging are contradictory to each other here as OP pointed out at the beginning of the thread. the profits really come from the moment that the accounts are not hedged, i.e., the profits come from risk, which OP also stated. in this sense, I can not see the benefit of the strategy comparing with regular averaging down approach.

 

[iuykcif]

> the constraints between the two account is much more than 2 contracts

Of course it is. You missed perhaps to see that I stated I was doing an experiment without constraints (see the post):

"Got curious to try <b>without constraints</b>.
I have been running for 1 hour on ES, YM "


Not only we can force the max difference to be constrained.

But also, we could also force, a <b>max number of contracts per account</b>, jointly with a straddle or a strangle (depending on our preferences).

Putting together these elements:

- Robot Scalping
- constraints
- options (even on multiple strangles)

you are able, for instance, to scalp within one or more strangles, and if the price moves too much you "scalp" the option, close all and restart.

Plus, when an "instance" of the Robot gets locked within it's constraints, you can start another instance.

<b>There is no limit to the number of concurrent (and independent) instances of the robot you can launch at different price levels.</b>

Profit will have hard time to escape.

Trading is a war, and an art. And you must use all the resources you have to defeat the enemy.


Currently I am adding the feature that any instruments will control automatically an arbitrary set of attached strangles/straddles (even cross instruments: for instance ES/SPY options) to get advantage of rallies too.

(Further, scalping could take place directly on options.)


Tom

Quote from trend2009:

i explored this idea for a while. it seems the strategy is similar to two accounts that adopt average down strategy separately. in this chart OP provided
(http://www.datatime.eu/public/TradingDiary/YM-2009-09-04-1Hour.gif), there is a moment that OP averaged down on short side while the constraints between the two account is much more than 2 contracts. I guess at the highest point, the drawdown is a lot. luckily, the market moves down, so OP made lots of profits. in case the market keeps up, OP will lose a lot.

thus, profits and hedging are contradictory to each other here as OP pointed out at the beginning of the thread. the profits really come from the moment that the accounts are not hedged, i.e., the profits come from risk, which OP also stated. in this sense, I can not see the benefit of the strategy comparing with regular averaging down approach.

 

[iuykcif]

Many friends at ET are asking an example on ETFs

Here is some BackTest/tuning on the pair SPY/SH:

<a target="_blank" href="http://www.datatime.eu/public/gbot/RobustnessAssessmentExample_SPYSH.htm"> SPY/SH tuning </a>


(this is unconstrained, but it can be, clearly)


Tom

 

[trend2009]

if you long and short at the same time,clearly the two accounts will cancel each other, and you will lose from commissions; in other words, if the system is profitable, it comes from the time delay you enter the long or shor positions. thus, I can not see any value of the system because if the system can profit from the time delay of entry, you can do it in one account.

 

[PocketChange]

Quote from trend2009:

if you long and short at the same time,clearly the two accounts will cancel each other, and you will lose from commissions; in other words, if the system is profitable, it comes from the time delay you enter the long or shor positions. thus, I can not see any value of the system because if the system can profit from the time delay of entry, you can do it in one account.

Only at instances when your flat ...
net = 0. Obviously the system trades adjusting weighted side long or short and through a sequence of trades scalps the curve.

Been discussed in many posts. The challenge is managing the inflection points where the accumulating trades can be offset realizing profits and managing unrealized positions so they adequately contain risk.

The purpose of holding contracts open both long and short are to average up and average down so both sides exit with profits (Inflection Point). When you try to trade these types of strategies in a single account the positions immediately offset and close.

Suggest you take one of his graphs and plot the sequence of trades in excel and see how the net posiitions project to a single account. Hopefully you'll recognize the difference.

 

[iuykcif]

Good advice <b>trend2009</b>

If you succeed projecting the trades equivalently to a single account, you can trade without margin worries.

Try with today's scalps:

<a target="_blank" href="http://www.datatime.eu/public/TradingDiary/ES2009-09-08.gif"> ES2009-09-08.gif </a>

[This was "constrained" as follows:

Max diff = 5
Max contracts = 15 ]

I am going now to delve into options to make the strategy financially viable also for small funds.

Tom

Quote from PocketChange:


Suggest you take one of his graphs and plot the sequence of trades in excel and see how the net positions project to a single account. Hopefully you'll recognize the difference. [/B]