Trading with automation (with IB)

[fullautotrading]

ET is full of haters

Haters are fine, but I did not see any so far in this thread. Everyone has been making quite useful and intelligent remarks, clearly within his own experience and understanding. I accept candidatures anyway I like to be challenged and provide transparent answers. I am in full control and really I don't mind any question. If there are things to improve, I will be happy to improve them.

Think more basic - think you're trading a (possibly single) futures contract. If that contract has a huge draw down

Well, you are talking of a non existent situation, because there will be opposite players which will start almost immediately in "protection". You should know that from my previous posts. That is the analogous of the ordinary "stop" concept. The difference is that I maintain all the trading information about stops ("protective" players) and therefore I can possibly use it to create a recovery plan in most cases. In the most common automated approach people do not store any trading information, nor they can overlay long/short layers. Which means, in that case, DD = loss.

Not sure why you want to mention a situation which does not apply at all to our case (if it's an attempt to get the "hater" status you need to try better ). But yes, in the situation you mention, you would be right.

Again you cannot make a profit without a DD. You can partly control your DD according to your taste by working on the size and considering the volatility, but you cannot eliminate it. Simply because the "investment" is the source itself of possible future profits. The dream of becoming rich with no risk, no pain, no investment belongs to everyone, but quite rarely comes true.

The illusion of equity curves going steadily up is a computer artifact, obtained by curve fitting past data. It does not exist in real world trading.
Most people mentioning large profit with negligible DD, when investigated better, are people trading peanuts, with few trades and a limited timespan (or they are often unable to produce credible documentation).

You can always provide here (or in a new journal) an official broker report (as I am doing) long at least 200 days and with at least 6,400 trades and at least 1.5M capital and perhaps start making a point about profits and DD. In all other cases it's rather entertaining talk versus hard, minutely documented facts. In my experience I mostly see people blowing up their accounts, or making no real money in years of trading, let alone ever seeing their account grow to touch +415K anytime.

 

[d08]

When you buy your stock of shoes you are investing money for a possible future profit. Your shoes are in a warehouse waiting to be sold, and they may get damaged by rats, or a flood, or because a fire. Or maybe they go suddenly out of fashion, or... maybe everyone is now buying the Feiyue instead of your sport shoes because the Shaolin monks and top wushu practitioners wear them . This is analogous, in our approach, to not being able to use the preserved trading information.

Again you are mixing up a "loss" with an "investment". Instruments have drifts and of course you need to align yourself to them in order to maximize your chance to use (at least partially) your trading information. You are, however correct, that for most traders DD = "losses", because, especially the automated ones, in most cases do not have a mechanism to preserve and use information (experienced "discretionary" traders will usually compensate by using "intuition"). That is also why DD is so scary to many of them, because there is no recovery plan "built in" the strategy: money and information is gone, and that is it.

Given a decrease of risk capital, I define it as "investment" in the context of trading if: the relevant trading information has been stored, and it is practically possible to use it, at a future time, to obtain a profit; if the above condition does not hold, we have a "loss".

For instance, W.B. implicitly and intuitively is preserving and using trading information in his approach (and this costs him some DD), with the difference that he is limiting himself to the instruments he "understands" better. While, instead, I see, in an automated setting, the general concept better applied to instruments which for various reasons have a more "predictable" (or even "structural") drift.

There is a difference in the examples. If you own shoes in a warehouse, you don't have to assume they will be bitten by rats or go out of style because they can also become a fad and will double in price or there might be a global shoes shortage driving up prices.

When you buy a stock at $30 and it goes at $20 then that is it's current value. We can't just assume it should go to $30 again as it can just as easily go to $10. If the stock went down to $20, why are you sure there might not be a corporate scandal or oil report or whatever that drives down the price (the rats in your shoe example).

As I mentioned earlier and as you say, if you want less risk, you may want to use smaller sizes. Although with futures, even the starting size of 1 contract can be too large for many undercapitalized traders. Leverage is not however the only a factor. Volatility is crucial too, 1 contract of a bond futures behaves completely different than 1 contract of silver. (Similarly the ETFs we have been using have monster volatility.)

Absolutely, volatility is an important factor. You said risk capital is there to be used, it's true but it's not as linear as you say it is because assuming you invest a percentage of your portfolio in a position (whether volatility adjusted or not), coming out of a huge DD requires a lot bigger returns than coming out from a minor DD. So theoretically you could have a 70% DD but at that point the account is almost done because to reach HWM you need incredible returns that could be unrealistic at that point. Again, this is just an illustration, not specifically about your DD or returns.

 

[fullautotrading]

When you buy a stock at $30 and it goes at $20 then that is it's current value. We can't just assume it should go to $30 again

I am not sure what's the point of that. None is assuming anything about price behavior. All what I am saying is that as soon as the stored trading information can be used (to make a practical example, if you are constantly going long on UVXY this condition would be violated, as you start fighting a "structural" drift, thus systematically stranding players, and their info) I can consider, in my algorithmic approach, the decline of the PNL as an "investment". Otherwise it's an unrecoverable loss.

Note that the absence of information (or the practical impossibility to use it), given a DD, will make impossible to systematically recover that specific DD, by "undoing" the respective "causes". In practice, this means that if you don't have a systematic recovery plan in place, within an automated system, once you see a 20% decline, there is absolutely no reason not to make another 20% and thus blow up. This is the reason why people fear DD.

Instead, as shown in practice, I can even go down to about 40% and quite easily come back by systematically undoing all my stops, and accumulating further scalps and in addition shoot up to almost +30%. So what kind of "return" is that ?

it's true but it's not as linear as you say

I don't think I said that. In reality, in my approach, unless you are not fighting "structural" drifts, once in DD, it becomes even more likely for the PNL to bounce back. What kind of return is necessary and how it is scaling does not even matter, because you have being accumulating a larger "potential" which is intuitively embedded in your position and it is balancing that aspect. The G-L curve (dotted green in my screenshots) in fact is expressing exactly that potential (which of course, as the name suggests, is a "potential" and not something necessarily occurring in the future).

 

[d08]

But if you know that for sure, it would make sense to simply begin with a small position and once in a significant DD, add a much bigger position, martingaling essentially.
The problem with that is no trading logic works forever and eventually you might have a 40% DD when the likelihood of a massive rally in your equity is statistically likely, your edge stops producing and you go into a further drawdown. Markets can and will change in behavior.

Anyway, we seem to have radically different views on DD so I'll leave it at that. Being in a big DD for significant amount of time is mentally draining for me and I don't consider it a positive.

 

[sysdevel99]

Again you cannot make a profit without a DD.
The illusion of equity curves going steadily up is a computer artifact, obtained by curve fitting past data. It does not exist in real world trading.
You can always provide here (or in a new journal) an official broker report (as I am doing) long at least 200 days and with at least 6,400 trades and at least 1.5M capital and perhaps start making a point about profits and DD. In all other cases it's rather entertaining talk versus hard, minutely documented facts. In my experience I mostly see people blowing up their accounts, or making no real money in years of trading, let alone ever seeing their account grow to touch +415K anytime.

You've obviously figured it all out - so who am I to tell you you're wrong.
You remind me of an options trader who had a thread

http://www.elitetrader.com/et/index...options-credit-spread-trading-journal.212817/

Initially he thought he was smarter but then reality caught up with him.

Not sure you'll have the capacity to take advice - several folks now told you that the way you look at draw downs is wrong and the only reason why you can handle a 60% decline is cause it's a paper trading account - hence I'm no match for you and your +415k fictitious paper trading profits (buy a candy bar with that - oh wait it doesn't even pay for that ).
Good luck - I'm happy trading 1/3 of your paper trading account size (I'm just not as big as you) in real money with much smaller draw downs and a non curve fitted steadily growing equity curve.

 

[fullautotrading]

why you can handle a 60% decline

We never had a 60% decline. You may have problem with trading, if you have issues computing basic percentages.

The criticism by d08 was that I took a bit more than 30% DD, while he thinks that 30% is fine for him.

Well I think that, in general, it depends on the circumstances. In my case, we had the price of the Crude Oil pretty low, and I liked to take some extra risk, by a "manual" override on the (high volatile) SCO (see the posts of the event).

You would not have done it? So fine. You would have had less DD (less than 30%). What's the big deal ?
Further, with the occasion, we have also shown we are fully equipped to easily recover, because of the recovery mechanisms built in the algorithm engine.

If you are a shoe trader and see a convenient (low price) stock of shoes, you may consider buying a bit more of them for a (potential) greater profit. Do you have to stick to some rigid "rule", made-up by some other losing chicken ? You don't.

The risk capital is there to be used. As I said, if there is a part of your capital you don't want to risk, just don't risk it. It's as simple as that.

Not sure you'll have the capacity to take advice

I may be able to take intelligent advice better than you can compute %. And that is of course an hypothesis.

Reality can catch up with me, as there is always a component of risk involved in what we do, and blowing up an account is not a rare event. Nevertheless, I can see what are the underlying principles, which are valid even in case of blow up, and that I can keep, and what are the habits and personal shortcomings I might want to give up.

Instead, you are failing to see the general principle I am indicating ("preservation and use of trading info") and you are seem content with that (it must be because of the candy bars you can buy with your profits ).

 

[fullautotrading]

Last week we continued our "investment" on CL, mostly, and SI. We are also slowly recovering NG (not really a good instrument for our scalping/hedging games, which needs more "fluctuations" than it does), while ES is also starting a new "load phase".
Since we have a good position on CL (riding the contango), I am also shorting a good number of PUTs for which we have a "cover". Same for SI. (This yields about $15-20K a month.)
AG stuff (ZM, ZL, ...) is definitely not worth the capital allocated, and I am looking forward to the possibility to get rid of that stuff.

As expected NUGT also spiked up and we did not miss the chance to (cautiosly) invest some $ on it (up to a notional value of about 80K). Currently it is still $7K "underwater", but these are instruments with monster "decay" (just as UVXY and the likes).

 

[fullautotrading]

begin with a small position and once in a significant DD, add a much bigger position, martingaling essentially

"Martingaling" is not viable. Actually, it's a sure way to quickly blow up an account. What we do can actually be described as almost the opposite of that.

The application is continuously "hedging" placing "protective" players opposite to the direction where there is the greater load. This normally means that a good proportion of DD that one would have with some avg up/down action on either side is recovered (thus increasing the chances to hold up on large moves). The proportion depends on the user parameters and must be watched carefully. Currently, this is controlled by making orders which are a (user defined) "fraction" of the opposite side. In addition, the continuous scalping action on price fluctuations also provide another big source of edge/protection (clearly the instruments must be chosen "suitably" for that).

Extensive simulations show that essentially you cannot engage in a systematic "stop and reverse", as this would be unprofitable (=avg of sells <= avg of buys) on the long run, but in essence you must let one side grow (even alternatively) for future reversals. On the long term, this works most of the time, provided you do not go (=let the "load" grow) against some long-term structural "drift", in which case the continuous scalping action, is no more sufficient to counterbalance the "losses" (in such a case, we can use this term, because trading information, although stored, cannot be used in practice) of the permanently "stranded" players , and the running trading expenses.

 

[globalarbtrader]

The one thing that strikes me about this thread is that the process being followed sounds incredibly (over-) complicated. However it might just be the language that the OP uses; they seem to have a knack of describing things in high falutin' terms that might actually be very simple.

GAT

 

[fullautotrading]

it might just be the language that the OP uses; ... that might actually be very simple

LOL. You are definitely right globalarbtrader. I have one mild justification: that I am very far from being a native speaker (actually I learned some of this language in adult age), and the process of "translating" my own terminology and concepts into a non native language may likely play tricks to the native's ear.

The basic ideas are indeed relatively simple once explained. The implementation does require effort though, but I guess it's the same for any application designed to stay up without ever failing and taking into account an endless number of small details.