Optimizing or curvefitting

[OddTrader]

Quote from hypostomus:

...re your Stridsman quote, I would have to disagree in light of my understanding of Katz and McCormick, who found little applicability of any one system across all markets. But I only trade NQ, which is confusing enough for me. My definition of robustness is "the system doesn't make me feel like I was a fool when I developed it". Right now I am a twenty-fold fool.

LOL. We'd better keep our minds open (Stridsman is not nobody) before we making more than enough money.

Now look at this (from the same book):

Q

Over the next pages, I indulge in a lot of tweaking and modifications to the original systems, which some of you might call curve fitting after the fact.

I do not call it that, however, and my arguments for that are several.

First, I am not optimizing the systems, but rather trying to find the most robust solution that should work on average equally as well on as many markets as possible, without really taking the actual profitability into question until the very last moment.

Second, the little "optimizing" I do (if you still call it that), is very coarse and based more on my own reasoning and understanding of the systems' underlying logic than on actually stepping through each and every input combination.

Finally, the systems are tested on as many as 65 different markets, and I don't care which markets the systems are profitable on or not, or on which markets the systems will be traded in the future

UQ



Comments from anyone, and Hypo?

 

[hypostomus]

...the sucessful modelling example you gave of 24h energy markets IMO was highly deterministic and modelled a physical process. IMO the only market I trade (NQ) is a manipulated market subject to external influences only when the news is unforeseeable by the manipulators (e.g.,Saddam's capture). All else is known in advance and is used to make fools of public traders. WE have to be wrong so THEY can be right, and THEY have ample tools to assure that outcome. We do not have a financial market, but rather a financial parasitism. Therefore what I seek to model is the structure of manipulation witnessed only by price and volume evolving over time (actually, I don't even use volume, as it doesn't show up on my brokerage statement).

All my stuff triggers on one bar. Simple stuff, really, except for a couple of accidental discoveries. One is a simple price breakout, one is a simple range breakout, and one is a reverse Tri-Pack (I short the retrace he's waiting to buy).

 

[OddTrader]

Quote from hypostomus:

...I have a home-grown definition of robustness. I am not smart enough to understand even the one instrument I trade, much less the entire market. I approach system design statistically, and attempt to understand why a system might work only after it is developed. I find that most "logical" system ideas fail ignominiously.

My definition is:

- the equity curve is monotonically increasing

- the equity curve doesn't change slope much when volatility changes

- the optimal parameters don't change much over time

- the maximum number of losses in a row doesn't change much over time

- flat spots in the equity curve don't last more than about ten trades

- profitability doesn't depend on one big trade out of many losers or scratches

Obviously this definition is highly idiosyncratic to my anxiety-prone psychology, but it works for me.

Do you have a price tag for the above system, please?

 

[hypostomus]

...the guy who goes to the trouble to publish gets to define the terms. IMO anyone who thinks there might be a system which works over all markets and instruments over all holding periods is a perfectionist more interested in the PROCESS of system development than in actually making lunch money.

As a schmuck, my job is to look for the fingerprints of manipulation and parasitize the parasites, not worry about the quasi-mathematical niceties of system development. But I will say that an element of "robustness" to me is equal profitability long or short. Henceforth if you want to know how I measure the goodness of the chimerical systems I trade, please refer to its " argentiquite' ". Harry no doubt has a literature reference to it, or invented the term himself.

 

[TriPack]

Quote from hypostomus:

All my stuff triggers on one bar. Simple stuff, really, except for a couple of accidental discoveries. One is a simple price breakout, one is a simple range breakout, and one is a reverse Tri-Pack (I short the retrace he's waiting to buy).

Aha - I knew it wouldn't take long for people to fade me!

No serioiusly, I've thought about that too - trading towards the swing system's entry points rather than the other way around. I would guess that it isn't working out as too well right now because of the increased volatility and more straight line moves - just as my swing system is missing a lot of fills. Have you noticed that the market trends down with smaller reactions when it really gets ripping than when it is trending up?

 

[hypostomus]

...the market is in freefall and you're posting?

I have not read the Stridsman because I have read many of his ActiveTrader articles. They are of a piece with the Bulkowski and Hartle articles. They should all stop contributing because Niederhoffer has lowered the bar with his quasi-academic prattle.

I think I should shut up because you are inducing me make ever-more-sweepingly-idiotic pronouncements on a subject of which I am abysmally ignorant.

My systems are not for sale because they are so stupid nobody would buy them, and after seeing the rules, nobody would trade them.

 

[OddTrader]

Obviously we now would have at least two schools of thought. We just simply choose the one that works (making profits consistently).

Another issue is: Would you be very keen on finding and building some (statistical) edges into individual system to make it better than other people's?

 

[OddTrader]

Quote from hypostomus:

...the market is in freefall and you're posting?

I swing trade multi-day based on mostly EOD decisions. I've been shorting USD/JPY since yesterday morning.

Perhaps one day you should think about currencies, when a single market is too small to accommodate you order volume.

 

[hypostomus]

...perhaps you are right about changing to a more liquid market. NQ already responds to my one car sells by rallying and to my one car buys by collapsing.

At the risk of overgeneralizing, I think that all systems ARE statistical, whether their developers realize it or not, because they are based on multiple Boolean logic conditions, which is effectively utilizing binary decision theory in a multivariate setting. I think that explicit realization of that can lead to some interesting wrinkles, like applying two Boolean conditions to ONE variable.

 

[OddTrader]

Quote from hypostomus:

...IMO the only market I trade (NQ) is a manipulated market subject to external influences only when the news is unforeseeable by the manipulators (e.g.,Saddam's capture). All else is known in advance and is used to make fools of public traders. WE have to be wrong so THEY can be right, and THEY have ample tools to assure that outcome. We do not have a financial market, but rather a financial parasitism. Therefore what I seek to model is the structure of manipulation witnessed only by price and volume evolving over time (actually, I don't even use volume, as it doesn't show up on my brokerage statement).

Q

From a probabilities perspective, this means that instead of being the person playing the slot machine, as a trader, you can be the casino. ... Then, like the casinos, you will own the game and be a consistent winner.

--- Trading in the Zone (Mark Douglas)

UQ

Some say the highly liquid markets would have better odds to replicate the historical price patterns.