A truly riskless system?

[MathAndLogic]

Quote from sambian:

Thank you very much. There are 5023 heads and 4977 tails. I give Martinghoul advantage and choose tails for me. My starting bankroll is $0.01. I "tie up" half of it on each flip and win 2*"tied up capital" when the coin is tails, and lose "tied up capital"/2 when the coin is heads.
On the 521 flip my bankroll is $74763066129 and then it becomes too much for me
I let Martinghoul choose how much he owes me

http://docs.google.com/fileview?id=...jMtYjk2ZC00ZTMwLWE0YzYtZmM1NzM2MDc5NDhl&hl=en

Guys, betting on a fair coin is not how the real market works. As I was trying to point out in my previous post, taking profit or loss on the outcome of each flip is bypassing a random walk. This is a critical flaw. What you need to do is this:

set X=1; then flip a coin; if the outcome is head, set X=1+0.0001 (gain one pip); and if the outcome is tail, set X=1-0.0001; do this until X=2 to take profit, or until X=0.5 to take loss.

Now you will realize the probability of X hitting your stop loss is far, far greater than X ever hitting your profit target.

Now think about this: What is the probability of EURUSD moving 100 pips in one day? 500 pips in one day? 5000? 10000? I do not dispute the possibility of any such moves. Another problem with Sambian's system is that the rebalancing is done when EURUSD moves from 1 to 2 or to 0.5. How long is this going to take? Till next lifetime? Since the inception of the Euro, how much has the exchange rate fluctuated? Without taking into consideration of the actual time factor for price moves makes Sambian's system impractical. By taking the time frame smaller, the corresponding magnitude of price move is also smaller, but the level of noise and whipsaw is greater.

 

[Martinghoul]

There's a thread on NucePhy discussing this at length. As I said, this turned out to be a gem of a subject, for which I am very grateful to sambian.

http://www.nuclearphynance.com/Show Post.aspx?PostIDKey=67428

 

[sambian]

Quote from NasiWarrior:

Apologies if this has already been suggested...

Sambian: Would this work with going long on both an ETF & its Inverse twin at the same time and averaging daily?

Anyone care to backtest it?

I backtested it with FAZ and FAS and with SPXU and UPRO since inception. In both cases there were huge trends and buy-and-hold performed better than my system. I used adjusted closing prices for FAZ and FAS since there were stock splits, but I doubt whether one would achieve such returns (over 5000%) if he just held those two stocks.

http://docs.google.com/fileview?id=...jUtNDExNy00YjU5LWI0MWMtMWRkMzk3ZWQ1Mjcy&hl=en
http://docs.google.com/fileview?id=...DctZWU4ZS00NDZmLTlhODEtZDIzMDJiZmUzNGVj&hl=en

 

[asiaprop]

you guys still discuss this, wow. It belongs to the class of martingale strategy and the problem is that no tradable asset follows the payoffs of a coin toss, not currencies, not stocks, not options, not futures.

Yet there still is an issue with the random walk that was assumed and the incorrect numeraires but I also stand corrected that this is not the underlying problem why this strategy does not work. The main issue is that assets dont follow such payoff profile. Theory is beautiful until we reach reality...

Quote from Martinghoul:

There's a thread on NucePhy discussing this at length. As I said, this turned out to be a gem of a subject, for which I am very grateful to sambian.

http://www.nuclearphynance.com/Show Post.aspx?PostIDKey=67428

 

[MathAndLogic]

Quote from Martinghoul:

There's a thread on NucePhy discussing this at length. As I said, this turned out to be a gem of a subject, for which I am very grateful to sambian.

http://www.nuclearphynance.com/Show Post.aspx?PostIDKey=67428

Precisely:

(A) In random JUMP of extreme volatility (NOT random WALK), (sambian system) = (buy and hold with stop loss) = (huge easy profit)

(B) In real market without repetitive random JUMP (random walk), (sambian system) = (negative expectancy).

QED

 

[Martinghoul]

Nah, it's actually more subtle than that, because, for any given counterexample you provide, sambian will be able to tweak the parameters of his strategy that will still make it outperform. Problem, of course, is that it requires the same forward insight that is the stumbling block for all Markowitz-like portfolios.

 

[Martinghoul]

Quote from asiaprop:

you guys still discuss this, wow. It belongs to the class of martingale strategy and the problem is that no tradable asset follows the payoffs of a coin toss, not currencies, not stocks, not options, not futures.

Yet there still is an issue with the random walk that was assumed and the incorrect numeraires but I also stand corrected that this is not the underlying problem why this strategy does not work. The main issue is that assets dont follow such payoff profile. Theory is beautiful until we reach reality...

Seriously, asiaprop, I think this actually turned out to be a pretty interesting subject. It's got the geek in me all hot 'n bovvered, that's for sure.

sambian, are you going to answer my question? If not, can you just tell me how much money you're willing to put up to play the game? Will you be happy to invest 100% of your net worth? What about leveraging up?

 

[asiaprop]

I beg to disagree at this point: The strategy will not outperform if the trend is persistent. It all comes down to the matter of how mean reverting a process is. Even in the coin toss if there is a series of one sided outcomes long enough that strategy goes broke no matter what.
Isnt that what most traders on wall street actually do, including volatility traders. Some sell the wings just long enough until they go broke. Then you get your ass fired, wipe yourself clean and start with a clean slate. I dont see where the magic is. Nobody lets you get to a drawdown of 70, let alone 90-100%, and I would claim even those with the biggest balls of steel dont have the stomach to manage such strategy once you are in such trend against you.

Quote from Martinghoul:

Nah, it's actually more subtle than that, because, for any given counterexample you provide, sambian will be able to tweak the parameters of his strategy that will still make it outperform. Problem, of course, is that it requires the same forward insight that is the stumbling block for all Markowitz-like portfolios.

 

[asiaprop]

i agree with you thats why I was happy to admit this is more intriguing than I first thought. Still the math on the assumed random walk and choice of numeraire is wrong on the initial fx example.

But you are asking (again) the key question: How much of your wealth are you willing to employ. Because it comes down to the willingness and ability to sustain an almost complete blowout. It could be done on self-invested funds, however, every professional environment will never let you put this strategy to work because the risk is just too high.

Quote from Martinghoul:

Seriously, asiaprop, I think this actually turned out to be a pretty interesting subject. It's got the geek in me all hot 'n bovvered, that's for sure.

sambian, are you going to answer my question? If not, can you just tell me how much money you're willing to put up to play the game? Will you be happy to invest 100% of your net worth? What about leveraging up?

 

[sambian]

Quote from Martinghoul:

sambian, are you going to answer my question? If not, can you just tell me how much money you're willing to put up to play the game? Will you be happy to invest 100% of your net worth? What about leveraging up?

I'm willing to put up $0.01 to play the game. You already owe me the net worth of Bill Gates + Warren Buffet, so I suppose you are the one who is leveraged