Trading algorithmically a folio without stops (with IB), real $$$

[fullautotrading]

End of the week. The move on metals continued massively, with consequent massive drawdown (DD) and load on the "biased" instruments.
Particularly severe the DD due also to the presence of several instruments which have nearly identical behaviors: USLV, NUGT, GDXJ, DUST, DSLV, JDST.
This is not a good joint choice of instruments. Probably, 2 or 3 of them in a folio would suffice. In particular, JDST has been the greatest delusion, as, while sweeping a range of about 146%, has been not shortable most of the time and caused many troubles and losses.

Among the "inverse", DUST seemed to behave fine. The bull 3X,USLV and NUGT so far have caused no technical problem so far. Same thing for GDXJ, which seems to have the largest volume (NUGT and GDXJ are very similar, except that GDXJ has about half the volatility). USLV (a 3X) is also quite volatile, and in this case it had a very violent move up (see picture below, green lines are about 1% distant).

USLV move and DD:

It looks like next week we will have a good and instructive challenge ahead dealing with this situation to recover these running-away instruments.

Probably, I also need to rethink the "short only constraint" for "biased" instruments: considering this kind of volatility it is probably better to allow the position to freely invert if necessary to hedge more effectively (for equities, instead, I will leave the "long only" constraint as default in place, for the moment).

In the meantime we have reached the full usage of the all the available funds, and I closed almost all the instruments which were in profit, in order to focus all the resources and the "attention" on the scalping/hedging of the instrument in DD.

During the weekend I will also be implementing another scalping mechanism, which I will explain in a next post.

 

[jcl366]

I don't usually compare with simulations. The simulator to me is simply an environment which helps the fund manager educate himself and to make up his mind about the hedging/scalping rules he wants to enforce, as he can see them in action in the wildest possible ("realistic") scenarios that can be conceived.

Ok, but which method do you then use for determining if the system works as expected?

If I understand your charts right, you've lost $ 40K so far. This can be a normal drawdown, but it can also indicate a fundamental system flaw. In the former case you just have to sit it out, in the latter case you have to pull the plug for preventing further losses for your investor. So this is an important information for a funds manager.

 

[fullautotrading]

Ok, but which method do you then use for determining if the system works as expected?

If I understand your charts right, you've lost $ 40K so far. This can be a normal drawdown, but it can also indicate a fundamental system flaw. In the former case you just have to sit it out, in the latter case you have to pull the plug for preventing further losses for your investor. So this is an important information for a funds manager.

There is no such a thing as "pulling the plug" here. We scalp and hedge and it's a continuous, uninterrupted process.

The only reason when we "pull the plug" is when there are technical (technological) issues with the program itself or some bug in the algoritms. In this case, the algo is doing exactly what is meant to do.

If one does not like drawdowns and investing, he should not trade, and continue living in the dream of his curve-fitted (and never really traded) delusion models which are free of DD, or buy gov bonds.

Drawdown is actually (from a logical point of view) a necessary (not sufficient) condition for systematic profitability (I will expand this concept later on).

 

[tropicalknight]

You have missed the point. PNL is evidence of the viability of a trading system. Just like any other business, if you run a taco stand and are losing money after 6 months, then you would tinker with your formula, pricing, etc. or just shut down. PNL in itself is not the be-all and end-all as far as trading goes, but it is evidence of something. There is enough variability and randomness in the markets that it is difficult to know for sure. Your system also has extreme overnight lending costs.

At the minimum, you should settle on certain products, no duplicating double-longs and triple-longs, unless you are running some sort of NAV value engine for executions.

Also, an attempt at backtesting (with carry costs) could at-least give you some idea what would have happened in the past. Exits should be part of every system, not some manual override.

 

[jcl366]

Drawdown is actually (from a logical point of view) a necessary (not sufficient) condition for systematic profitability (I will expand this concept later on).

That's not what I meant - my question was just how do you know that a loss is caused by a (temporary) drawdown, and not by a (permanent) flaw in the trading method or a market change. Or is your trading method so perfect that a flaw is impossible and simulation / validation are not needed?

 

[fullautotrading]

That's not what I meant - my question was just how do you know that a loss is caused by a (temporary) drawdown, and not by a (permanent) flaw in the trading method or a market change. Or is your trading method so perfect that a flaw is impossible and simulation / validation are not needed?

The approach is to do the "best" actions (in a statistical sense) under the given real-world circumstances. Nobody (but scammers or idiots) can ever give an absolute guarantee of "success": you can only "guarantee" to do the best actions, conditional on the given circumstances.

 

[fullautotrading]

> PNL is evidence of the viability of a trading system. Just like any other business, if you run a taco stand and are losing money after 6 months, then you would tinker with your formula, pricing, etc. or just shut down. PNL in itself is not the be-all and end-all as far as trading goes, but it is evidence of something.

The analogy with a taco is not fit. In trading you need to continuosly reinvest in new volatility . I raised 30K ? Fine, I am now investing 70K in a new investment phase. What's the problem ?

>There is enough variability and randomness in the markets that it is difficult to know for sure. Your system also has extreme overnight lending costs.

Yes commissions and interests are costly. But I guess large funds can get better deals.
For overnight, one can always "neutralize" the position if he wants to. That's no problem.
Good capital is necessary: this is not an approach suitable for retail traders.

>At the minimum, you should settle on certain products, no duplicating double-longs and triple-longs, unless you are running some sort of NAV value engine for executions.

That's right, I already said that. But we are making a journey here and actually objectively testing and experiencing first-hand all the instruments.

>Also, an attempt at backtesting (with carry costs) could at-least give you some idea what would have happened in the past. Exits should be part of every system, not some manual override.

Right the simulation facility can give an indication.
Exits are full automatic. But the fund manager can exercise any form of control and supervision, on any degree he likes (from 0 to total).

 

[Mo06]

I cannot tell from your posts if the system is profitable or not.

What percentage return has your system made during the test period ?

 

[fullautotrading]

I cannot tell from your posts if the system is profitable or not.

What percentage return has your system made during the test period ?

The "test period" is just this thread and the equity curve can be followed in detail in the various posts.

It's still clearly too early to draw "conclusion" on the scalping/hedging games, it's just now I have been using more extensively the available funds. Most of the time I have been using much less than 200K margins, with a maximum PNL of about 30K during the first months.

We are right now in the middle of a larger "investment" phase and we need, of course, this new investment to be transformed into new profits. This is the ordinary way trading works: you have alternating phases of investment and returns. Clearly, the purpose of a folio is, in general, that of creating a useful "interference" across these phases for different instruments, in order to lower the variance. In this case, while testing the various ETFs, we were caught with a bit too many metals (and mostly 3X), thus essentially losing the "folio effects". But I would not panic. Everything is under control.
Remember the saying: Anyone can hold the helm when the sea is calm

Also, remember that this is a "journey", where, while the general philosophy and architecture is (or should) be perfectly clear ("preservation of the trading information") and the reason behind it, the actual scalping/hedging mechanisms (that is, the ways through which the information is actually used) are also object of a continuous evolution towards more mature forms, on the basis of the actual daily trading experience and the challenges encountered.

 

[Mo06]

"We are right now in the middle of a larger "investment" phase and we need, of course, this new investment to be transformed into new profits. This is the ordinary way trading works: you have alternating phases of investment and returns. Clearly, the purpose of a folio is, in general, that of creating a useful "interference" across these phases for different instruments, in order to lower the variance. In this case, while testing the various ETFs, we were caught with a bit too many metals (and mostly 3X), thus essentially losing the "folio effects". But I would not panic. Everything is under control. "

Hmmm.... I guess that is a no.