This Forum overtrades options

[bwolinsky]

Quote from TigerBalm:

bwolinksy, you're lucky that sle is even debating you. To post so much without being able to figure out which ET user is a pro and which one is not is the mark of someone who doesn't have a clue about their context. Be careful when you trade.

Hang on, I found a Wikipedia article about your ilk:

http://en.wikipedia.org/wiki/Dunning–Kruger_effect

"The Dunning–Kruger effect is a cognitive bias in which unskilled people make poor decisions and reach erroneous conclusions, but their incompetence denies them the metacognitive ability to recognize their mistakes.[1] The unskilled therefore suffer from illusory superiority, rating their ability as above average, much higher than it actually is..."

Not if I'm trading profitably 95% of time. Like I'm not a pro? Shit, what's a guy got to do just for somebody to pretend they know and acknowledge I already have my trade secrets. It's not indictment of anybody, just that there's no point to use options without models.

 

[bwolinsky]

Here is an example of real financial research.

I take somebody's code, plug it in. See it needs to be optimized, optimize. Then see it needs some stops. Then see it needs special code snippets to function in real-time.

Then, optimize the times to trade and exit, optimize the trail stops target and loss percentage. Optimize some stops and targets by looking at profit distributions of MAE/MFE scatter plots.

Finally, re-optimize once you got all of your theory down and you're just trying to find the most optimal values.

In the case of the 5 variables that drive the core entry and exit system, not the stops, in 5 of 1000 this is 1000^5 or 1 with 15 zeroes of possibilities. Add more variables but maybe not quite so wide a range and there end up being more than google of possibilities. DAM. WTF AM I SUPPOSED TO DO NOW?

Optimize genetically for a population size of 1,000 and a maximum and minimum number of generations of 250 and 225, respectively.

Within 100-475 hours your program will have found a solution without exhaustive analysis that would take longer than you, your kids, and their kids' kids lifetimes.

So, still no reason not to do that for what you guys are doing now?

 

[atticus]

Quote from bwolinsky:

Here is an example of real financial research.

I take somebody's code, plug it in. See it needs to be optimized, optimize. Then see it needs some stops. Then see it needs special code snippets to function in real-time.

Then, optimize the times to trade and exit, optimize the trail stops target and loss percentage. Optimize some stops and targets by looking at profit distributions of MAE/MFE scatter plots.

Finally, re-optimize once you got all of your theory down and you're just trying to find the most optimal values.

In the case of the color=red]5 variables[/color] that drive the core entry and exit system, not the stops, in 5 of 1000 this is 1000^5 or 1 with 15 zeroes of possibilities. Add more variables but maybe not quite so wide a range and there end up being more than 70 zeroes of possibilities. DAM. WTF AM I SUPPOSED TO DO NOW?

Optimize genetically for a population size of 1,000 and a maximum and minimum number of generations of 250 and 225.

Within 100-475 hours your program will have found a solution without exhaustive analysis that would take longer than you, your kids, and their kids kids lifetimes.

So, still no reason not to do that for what you guys are doing now?

You can't apply delta1 backtesting to products with many orders of convexity. Px & vol; delta, gamma, theta, vega, rho; vanna, vomma, charm... the modality of delta alone would be a bitch.

Knowing what is absurdly obvious to everyone here; why not simply backtest the underlying?

The rationale for trading vol is the convexity. Coincidentally, it's also the reason that you don't backtest vol.

 

[BeatingtheSP500]

Not if I'm trading profitably 95% of time.

 

[sle]

Quote from atticus:
Coincidentally, it's also the reason that you don't backtest vol.

Just in the interest of full disclosure I do use a fair bit of historical simulaton of delta hedging for various purposes (can disclose in a PM, don't feel like sharing it here). I am not sure it qualifies as bona fide "backtesting".

 

[atticus]

Quote from sle:

Just in the interest of full disclosure I do use a fair bit of historical simulaton of delta hedging for various purposes (can disclose in a PM, don't feel like sharing it here). I am not sure it qualifies as bona fide "backtesting".

Yeah, I don't feel it's in disagreement. That's an argument for what-iffing path-dependency.

 

[BeatingtheSP500]

Not if I'm trading profitably 95% of time.

you must be selling the teenies and those 5% losers must be whoppers. or maybe you're not profitable 95% of the time.

 

[DontMissTheBus]

Actually, my boy, this is a perfect example of how NOT to do real financial research.

Quote from bwolinsky:

Here is an example of real financial research.

I take somebody's code, plug it in. See it needs to be optimized, optimize. Then see it needs some stops. Then see it needs special code snippets to function in real-time.

Then, optimize the times to trade and exit, optimize the trail stops target and loss percentage. Optimize some stops and targets by looking at profit distributions of MAE/MFE scatter plots.

Finally, re-optimize once you got all of your theory down and you're just trying to find the most optimal values.

In the case of the 5 variables that drive the core entry and exit system, not the stops, in 5 of 1000 this is 1000^5 or 1 with 15 zeroes of possibilities. Add more variables but maybe not quite so wide a range and there end up being more than google of possibilities. DAM. WTF AM I SUPPOSED TO DO NOW?

Optimize genetically for a population size of 1,000 and a maximum and minimum number of generations of 250 and 225, respectively.

Within 100-475 hours your program will have found a solution without exhaustive analysis that would take longer than you, your kids, and their kids' kids lifetimes.

So, still no reason not to do that for what you guys are doing now?

 

[1a2b3cppp]

Quote from bwolinsky:

Come up with ways to predict which way the market will move,

Haha, cuz that's so easy!

 

[sle]

Quote from bwolinsky:

Here is an example of real financial research.

I take somebody's code, plug it in. See it needs to be optimized, optimize. Then see it needs some stops. Then see it needs special code snippets to function in real-time.

Then, optimize the times to trade and exit, optimize the trail stops target and loss percentage. Optimize some stops and targets by looking at profit distributions of MAE/MFE scatter plots.

Finally, re-optimize once you got all of your theory down and you're just trying to find the most optimal values.

In the case of the 5 variables that drive the core entry and exit system, not the stops, in 5 of 1000 this is 1000^5 or 1 with 15 zeroes of possibilities. Add more variables but maybe not quite so wide a range and there end up being more than google of possibilities. DAM. WTF AM I SUPPOSED TO DO NOW?

Optimize genetically for a population size of 1,000 and a maximum and minimum number of generations of 250 and 225, respectively.

Within 100-475 hours your program will have found a solution without exhaustive analysis that would take longer than you, your kids, and their kids' kids lifetimes.

So, still no reason not to do that for what you guys are doing now?

A big computer, a complex algorithm and a long time does not equal science.
--Robert Gentleman