Do you need to understand where the edge comes from?

[tommcginnis]

Tom, you only make money when 2 things come to fruition.
1) You have a strategy that has a positive expected value.
2) You manage your risk so you don't blow up before you realize that positive expected value.

Look at Roulette, If the casino allows you to bet WHATEVER you want. Then a billionaire can come in and clean them out. -> Poor risk management + positive expected value = Loser.
BUT No matter what risk management the player does. He will ALWAYS lose in the long run. In fact his best bet is to walk in their and bet his whole stack on red or black and then walk out.

saying "Risk management is the only edge" is so lunacy. Risk management is not even an edge. It is an equation LOL. Any monkey can look up the Kelly Criterion for optimal bet sizing.

I agree with so much of what you say here. But the thing is, (*I* think,) that B1S2 does (to a degree) too. "Risk Mgt" is an edge only when the other side is sloppy. "How often does that occur?!?!" Well, maybe often enough, IF WE DO IT. "Any monkey can look up Kelly..." TRUE TRUE TRUE -- but with enough beer in you, you'll admit that almost NOBODY YOU KNOW actually knows the expectancy of their trades, unless they're trading OPM and have reporting requirements. Right? RIGHT?!?

(That last part? That was the arm twist right there. Tell me you felt that. Tell me. LIE to me.)

I crack me up. TIME TO GO OUT!
https://trapindy.com/musician-directory/a-i/hyryder/

 

[TheBigShort]

Yes I agree that long term expected value is very hard to measure. But as a trader you have to try and measure it and trade based on your assumption Manage your risk accordingly. I have nothing more to say on this topic. Its pretty straight forward.
1) Find a trade that you believe will have a positive expectancy
2) Don't listen to what @Buy1Sell2 is saying
3) Manage your risk based on already documented and public theories.

Ps. Arbitrageurs/Index holders/Debt holders and many more all have a good idea of what their expectancy will be.

 

[sle]

I need to understand where Ive come from

"Мама анархия, папа стакан портвейна" (c)

 

[dtrader98]

People often refer to ML systems as opaque black boxes. I’d be curious to see a successful ML application whereby one could not possibly understand why it is successful, or how it is processing the data, given all of the information it utilizes as well as the underlying architecture. Any examples?

 

[panzerman]

Love the creativity here, but some issues abide.
First, is there a T/A indicator that wouldn't qualify as a lowpass/highpass/bandpass filter?

Sure, lots of them, RSI, pattern recognition, ADI, etc. Digital filters come from the physical world of digital signals processing (DSP), and are based on solid scientific and engineering principals. None of the other indicators are based on anything but pseudo-science at best.

Of course using DSP to model the market as a signal plus noise is wrong, but then so are all other finance models to one degree or another, even the ones that get guys Nobel Prizes. Pick your poison, right?

Interpretation of what negative group delay means can be a bit subtle, but I don't think it means that it has any predictive power, if that's what you are going for by the term "leading indicator." I believe it just means that it does not lag.

 

[fordewind]

"Мама анархия, папа стакан портвейна" (c)

ну так ёптааа..

 

[Buy1Sell2]

So throw a dart to pick an instrument, risk no more than 2%, close position once 2R is reached. Rinse and repeat? You must have an amazing risk management system no one else was able to figure out.

You don't understand PRM

 

[aquarian1]

For people who develop and test their strategies, do you try to understand why it works? Fo example, your strategy shows great results and when you dig down you realize it was buying dips. So you say to yourself - great I have a good strategy to deploy in a strong bull market, but keep it on a shelf otherwise.
What I don't understand is how people use ML etc to get seemingly random strategies. How will they trade it without understanding how/why it works? How will they know when to stop? Are they just trading a basket of uncorrelated strategies? Would be interested in how people on ET approach this.

You have many questions here.

If your strategy is developed by yourself (without ML) then I would say that in some circumstances you need to know why it is working so you know when it has stopped (or when to apply it or not).

In ML if you are a big house and can dump money in then you need only understand when the parameters such as volatility have changed and turn off the algo. (They are, to my understanding only taking tiny pieces very rapidly)

 

[ElectricSavant]

You get stuck on Candlesticks...when you should read my post...high low close is the point. How we look at it bars...candles....does not matter. but you just gave me an idea...VOLUME Bars or candles...

Thanks Tom. Also Thanks for taking me seriouisly so I did not have to put you on ignore and blocking like some idiots in this thread.

Es

Yes. Even, "Of course."

But you're follow-up mixes apples and oranges. (Still fruit, but should only be mixed under the supervision of a sweetened creamy sauce, and watch for too much exposure to the Sun. )

You bring up candlesticks, which are agglomerations of what could be *thousands* of prices into just four: Open-High-Low-Close. If it *is* thousands, then the likelihood that we might miss something important by some slip of how the candle (timing) is defined -- that likelihood is small. But if we're looking at a 1-minute candle, and there were only half-a-dozen trades, then the story which the candle 'tells us' is highly subject to being victimized by when the trades occurred. Takeaway? The fewer the trades in the candle, the less use the 4-priced summary of the candle is. Apples and oranges -- still fruit, but...... yowie.

Second is that "The Patterns" of Japanese Candlesticks so overblown (and so often) are very different animals than the patterns of, let's say, a "MACD crossover" or a "Stochastic crashing through 80" or etc etc. (That's more like Crab-apples and Macintosh.) The "patterns" of use in T/A were coded long ago, put into trading algos, and their creators done very well for themselves. Testable, repeatable, reliable results, tuneble for robust outcomes over time. (And now, you've just gone from fruits to orchards....)

 

[Sprout]

If you allow time to be squishable or squeezable, you've wiped out an essential part of all of market studies (from finance to econ to sociology) being, "I am first! I got here first! I thought of this great idea first! I staked my claim first!" Uh-ohhhh.

At any rate, what is termed a 'leading indicator' needs an a priori interest in some ensuing event. Look at any of the Index Of Leading Economic Indicators. An increase in Building Permits this month does not signal an increase of Building Permits next month, it signals (signals! indicates!) an increase in Construction. No matter how many MAs we take of price, it lags what price will do next. It's math; it's 'by definition.'

It's not my fault.

Time is a useful illusion.

As for price, of course it lags, it’s the dependent variable of volume. Volume leads price.
If one hasn’t done their DD of the truth and usefulness of this foundational concept then you have the thoughts and experiences that you are attached to and will continue to defend them no matter the cost.

And for that price you get to be right in the limitations of your own mind and will seek out evidence as proof for the ‘righteousness’ of your world view.

When decoupled from time, the market operates in a sequence of events.

Just as you say, it’s the math.