Kudos to MMs

Quote from shortie:

1340 - the model does see something.

it still has this tendency to drift away from reality. it used to drift lower, now it drifts higher. i am guessing that if one were to use an adjusted FV that is periodically brought back to the current SPX value in some arbitrary way then the numbers would not be so confusing.

for example, today one would look at FV*, which is FV forced to be matched to SPX ~1 month ago.
More...

I have said that many times. For example, I had an instinct that this was the case back here:

http://www.elitetrader.com/vb/showthread.php?s=&postid=3004786&highlight=constants#post3004786

But I have a clearer understanding of what is going on now better than ever.

I have constants in the model that are hardwired. As the dynamics change, I need to adjust the "constants".

Notice that I said in that previous post that the coupling constants would be bad if they changed. Now I realize they are not constants at all, but scaling parameters that are dynamic. Someone that is skeptical would say this is curve fitting, but it isn't. I have to find the right equation, and the equation should be relatively easy to find because it has to map to some obvious measurement that will make complete sense that it is being scaled so. They will be data driven, not optimized in any way. I have done this by hand by messing with the values and the model comes close to SPX (1403 as opposed to 1340), but I don't know what the right scaling behavior is. It might even be a complex mixture.

I am very very close. I already know that the sign of the model is correct. Now I have to work on the level a little bit better.
 
Quote from nitro:

I have said that many times. For example, I had an instinct that this was the case back here:

http://www.elitetrader.com/vb/showthread.php?s=&postid=3004786&highlight=constants#post3004786



I have constants in the model that are hardwired. As the dynamics change, I need to adjust the "constants".



I am very very close. I already know that the sign of the model is correct. Now I have to work on the level a little bit better.
More...

Like I said its a neural net or least squares fit.
 
We reached the target of ~1342. I will be (working to) exiting position on the open. I missed almost the entire move on Friday (the hedge was not perfect I was slightly delta long). In retrospect, it was an irrational decision, but I understand myself and why I made it, right or wrong.

Overall this trade made ~80 points including the small losses I took early on. SPX is still undervalued by quite a bit, but nothing goes straight in one direction, and this is a major resistance point. Therefore there is logic in the choice.

Hope for a pullback and then try to time the reentry back long.
 
Nothing to do still. The first place we might act is 1320, and then the usual suspects of 1302 and 1272.

Brings up an interesting point. If you are trading in this manner to make a living, notice that the frequency of when you get to act is rather infrequent. That means when opportunities present themselves, you have to try to have maximum position on.

It means you have to have the right personality. If you are Pete Rose, you can't trade this system, or you will go broke unless you have an evening job. You have to be Barry Bonds. You have to go for the fences with a relatively big position on when the opportunity presents itself, because you don't know when your next opportunity will be. This is what is hard, because when the market moves against your position, managing your losses and looking to get back in tests every cell in your body. Top it off with having the discipline to ride out a trade to it's logical conclusion, and the mental makeup required to trade is...

Computing the actual account size needed to trade a particular system, and to be able to survive and grow your account is a neglected part of system trading by retail traders.
 
We may be breaking out of 1342. Next resistance is 1372. See how it opens, and if we look like we hold 1342, it is worth going long to get a few to 1372.
 
It is interesting how my growth as a modeler is on a similar path to the way artists of old understood perspective. It is worth investigating and understanding these in a historical mathematical setting:

Durer (woodcut St Jerome Institutiones Geometricae)
Barberini ( The annunciation)
Escher (impossible figure, circle limit III)
Max Weber (interior of the fourth dimension)

Understanding projective geometry and the cross-ratio is also interesting and full of ideas. It is the last piece of the puzzle of adjusting the scaling parameters aforementioned in the model of "FV".

I sometimes wonder if all the scaling parameters are is a dynamic sphere packing problem:

http://en.wikipedia.org/wiki/Sphere_packing
 
Yup as I said the top is in, when the model predicts the S&P at 400 it will bounce and go to 1450.

Great model +/- 1000


nitro, you seem like a nice guy, but I really worry, you also seem to have many delusions. Always comparing yourself to mathematicians artists etc.

Its just weird considering your model does not work.
 
You must log in or register to reply here.
Back
Top