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

fullautotrading said:
I am also not sure from where you infer the approach is not making money. My experience is just the opposite, and in particular this session we just started is just doing as expected.
More...

I infer the approach is not making money because the liquidation PNL is negative. Didn't the PNL on the prior live trial also go negative because of big drawdowns before you pulled the plug and reset the chart?

fullautotrading said:
We are after is long term systematic profitability. The "G-L" must be built up gradually to pull up the PNL, while at the same time hedging and recovering the hedging orders, and that we are doing, systematically. (If you are looking for "get-rich-quick" trading methods, unfortunately I don't have them, and actually I believe they do not exist.)
More...

I am making no claim about "get-rich-quick" or otherwise, and I am really not attempting an unfair criticism. I am trying to be constructive and to help if I can.

The "G-L" that you are referring to, does it represent an existing long position? How exactly is the potential gain determined? My understanding is that this potential gain is either a measurement of how much markets have moved against your position, or a measurement of how much money you could gain if the price returns to previous levels. In either case it is only potential gain if the markets do in fact go back to prior levels. If they never go back, the gain does not exist. This is why I say the model is built on an implicit assumption of mean reversion.

This part is very critical. You claim that the cash PNL does not represent the true performance, and only the potential PNL via this "G-L" accounting is the real measure. The problem is, realizing this "G-L" accounting seems to require markets moving in your favor. Unless you have invested some part of your model in predicting that, which I have not seen, I don't think you can count on those future gains as being guaranteed to eventually occur. Every other investment strategy is evaluated based on liquidation value... I don't think it is a fair or realistic evaluation to evaluate yours on a hypothetical future gain.
 
@monkeyjoe

<b>>I infer the approach is not making money because the liquidation PNL is negative</b>

I am not aware of legal trading methods whose PNL is systematically only positive.

<b>> Didn't the PNL on the prior live trial also go negative because of big drawdowns before you pulled the plug and reset the chart?</b>

Nope. I interrupted that session at around 5K PNL, after going over 20K, because of technical issues with the game and the order sizing in relation to hitting the IB initial margins. In fact, all the DD was caused practically by 1 instrument only (UGAZ), which had big technical sizing issues, and could not hedge.

<b>>I am trying to be constructive and to help if I can</b>

To be constructive on an idea, you should first understand the idea. Then you are able to provide contributions to it.

<b>> The "G-L" that you are referring to, does it represent an existing long position?</b>

Nope.

<b>> How exactly is the potential gain determined?</b>

The gain is defined here as the Gain component in the PNL decomposition:

PNL = G + L + Unr

It's an actual quantity (computed realtime), not a potential concept. I have devoted several posts before to explain that in detail.

<b>> if the markets do in fact go back to prior levels</b>

You should never mention any concept related to "reversion" or alike, as we do no use it, nor anything which has to do with prediction.

<b>> You claim that the cash PNL does not represent the true performance, and only the potential PNL via this "G-L" accounting is the real measure</b>

This something you are either making up, or you have problems understanding my posts.

<b>> Every other investment strategy is evaluated based on liquidation value... I don't think it is a fair or realistic evaluation to evaluate yours on a hypothetical future gain</b>

Sure. Evaluation is made of 2 parts. <b>1. Understanding of the idea</b> (which must be sound and not crackpotical). And also <b>2. proving the concept in practice</b>. As I said, you need to allow it some time to work. Nobody will ever get rich in a short period (nor with small capital or fear).
As for all investments, one needs to allow the right (capital and the right) time for it to work.

The problem of 90% of the "strategies" you see around, is not even the second part: often they don't even have a reasonable basic reason for edge in the first place. (I may need some technical adjustments with my large code and the unbelievable complexity of a realtime folio trading system of this level, but at least the idea is there, and I am going to make it work smoothly in the real world too.)
 
I seem to have offended you and for that I am sorry. I am not able to right this minute, but I will try to give very specific examples of my understanding of your method from past posts.
 
monkeyjoe said:
I seem to have offended you and for that I am sorry. I am not able to right this minute, but I will try to give very specific examples of my understanding of your method from past posts.
More...

Nope, no offense taken at all. (Incidentally, we just started a 10M fund based on this, so I don't need much more professional gratifications, or a unanimity of people supporting my views. Instead,) having the chance to clarify and address any misunderstandings and possible criticisms is just what I am looking for, as that is the way to strengthen and improve further, and bring some real novelty in the field. :-)
 
Let me post a quick "flashcard", summarizing my view on strategy evaluation.

<img src="http://www.elitetrader.com/vb/attachment.php?attachmentid=143925&stc=1&d=1394185556" />

I am pretty sure that many people looking at it would say: So what? this is obvious.

However, my personal experience, discussing with many people in several forums (in particular, those more generalist than ET) suggests that it isn't.
In particular, point 1) is almost invariably neglected. And, in several case, some people replace the lack of meaningful ideas with the adoption of math sounding techniques which make no sense in this context, or some obscure pseudo mathematical sounding argument (some even arrived to invoke quantum mechanics :-) as now it's very stylish, because as Feynman said, nobody understands it).
Many of course do that in good faith (as result of lack of scientific experience), and often denial does not really allow to apply scientific reason. Others, more pragmatic and shrew, are just interested in the fund <b>management fee</b>, as they already know their trading results will, at best, fluctuate around zero, mostly depending on luck (they would often integrate with more traditional investments to hedge), so the math sounding argument becomes simply a marketing tool to grow the amount under management. Clearly, not all investors are fooled, but the world is large and diversified ...

Clearly, let me know your different views, or things to be added to the flashcard.
 
Mo06 said:
Trading is not a science.

When scientists/mathematicians trade, the results can get 'interesting'.

LTCM, for example.

http://en.wikipedia.org/wiki/LTCM
More...

Yep, Mo06

it depends on who is doing the trading and how. Personally, to me it is a scientific endeavor.

Of course, I have no problem conceiving that to other people it may be something else, gambling included.

(Not to mention many other activities related which are not legal, like scamming, money laundering and so on. When you see the apparently miraculous results of "closed" funds (not referring to your example), you know that has nothing to do with what we are talking about in this thread.)
 
This is a basic volatility arbitrage strategy even though you've created your own terms for how you personally trade it. You're just trading and hedging synthetic volatility positions.
 
jb514 said:
This is a basic volatility arbitrage strategy even though you've created your own terms for how you personally trade it. You're just trading and hedging synthetic volatility positions.
More...

Hi jb514,

There a something interesting in what you say, as the "loss recovery mechanism" (what I call "the conservation of trading information", which essentially is responsible for the upward drifting of the "G-L") could also be intuitively seen as an artificial surrogate of options (without the time decay part). But I would, probably, not fully characterize this approach as a volatility arbitrage, as that would probably miss much of (what I consider) the main point. Not sure, it's an interesting argument, surely deserving to be discussed more...
 
End of week. Here is our update. Everything proceeding fine so far. We have been steadily growing the "G-L" figure, doing some a decent job with hedging the unrealized (max margins usage has been around 30-40K for this instance).
Probably, there is still room for improvement of the sizing dynamics (protecting players, and so on), especially for the "shield" game, which is probably the most difficult to "tune", so that it behave in an intuitively appealing way. In hindsight, 4 layers of TNA have not been a particularly brilliant idea. The ETF is a bull 3X (DIREXION DLY SM CAP BULL 3X) and we know the effects of the FED influence on the mkt. Just a couple of UPRO would have been probably a safer choice. Anyway, nothing really worrying, probably just postponing a bit the time when the PNL will go "above the water".

After all the ultrashort ETFs, played with a "short position constraint" and the "Bias" game, seem generally among the most comfortable to deal with. Clearly, interests do bother a lot, but they seem worthwhile anyway. A general impressions (but probably too early to say), is that the wildest volatilities in ushort ETFs are associated with the most bearable drawdown (probably due to stronger downward drift). This seems to be case of DGAZ, JDST (very large vol), for instance (but the matter could be investigated better on a longer timeframe.).

<img src="http://www.elitetrader.com/vb/attachment.php?attachmentid=143961&stc=1&d=1394229975" />

Maybe in a next post I will show the order clouds of the "worst" and "best" PNL so far, so that you can get an idea of the action so far.
If you are also curious about the reason while the G-L curve appears always a bit "jagged", in a next post I will also explain what causes this characteristic of the curve.

During the weekend I will be looking for improvements in the order sizing and, in case, on Monday update the games.
 
You must log in or register to reply here.
Back
Top