It is like I don't even want to win. Yes. $1k to $6k.
Quote:
So that's where the bet is. It's a duel between gentleman at 1000 paces on my shooting ability and in a sense on market efficiency, which Brian mostly believes in I think (bar maybe edges of double SPY return or something), and I don't.
I don't think the market is efficient. If I did, I'd avoid trading. I just don't think it is so inefficient that you are a lock for a 500% gain 1 out of every 3 years.
Your stipulations are just a description of your trading preferences.
juk,
I think you believe in some form of market efficiency. What multiplier, for total number of trades < 10, would prove to you that there are large excess returns possible in the market? 6 is obviously not impressive. What is? 100? 1000?
Or is no number possible since we don't have enough trials? Is the guy who returns 20%/year over 10,000 trades statistically more likely to have higher EV than the guy who returns 10,000%/year over 10 trades?
I'm interested in your opinion on this from a risk and EV calculation perspective.
I think you believe in some form of market efficiency
I really don't - I just think 99.99% of people are easily deluded into thinking they have an edge when they actually don't.
Quote:
Or is no number possible since we don't have enough trials? Is the guy who returns 20%/year over 10,000 trades statistically more likely to have higher EV than the guy who returns 10,000%/year over 10 trades?
For the guy who returns 20%/year over 10,000 trades, it depends on whether he takes approximately equal risk per trade (eg: 9,999 trades of $1 and then a trade of $10M is very different from 10,000 trades of $1000). If his distribution of risks is approximately equal, then he almost certainly has an edge.
The guy who returns 10,000%/year over 10 trades is just gambling, and IMO is no more impressive than a gambler at a casino having a lucky streak.
---------
I urge you to think about the question I posted above as the answer might surprise you... Here is the solution hidden in a spoiler:
Spoiler:
Assuming you are betting on a European roulette table, you have ~16.2% (ie: 6/37) chance of winning when you place a 6-number combination bet.
Probability of success on a single trial : 0.162
Number of trials: 100
Number of successes (x): 20
Cumulative probability P(X > x) = ~0.123 = ~12.3%
Which is on an "impressiveness" scale, lies between watching somebody betting a 5-number combination on a roulette wheel and winning, and watching somebody betting a 4-number combination on a roulette wheel and winning!!!
PS: In case it isn't clear, the 20 value comes from the break-even solution, ie: 20*(5000+1000)-80*(1000+500)=0.
For the guy who returns 20%/year over 10,000 trades, it depends on whether he takes approximately equal risk per trade (eg: 9,999 trades of $1 and then a trade of $10M is very different from 10,000 trades of $1000). If his distribution of risks is approximately equal, then he almost certainly has an edge.
The guy who returns 10,000%/year over 10 trades is just gambling, and IMO is no more impressive than a gambler at a casino having a lucky streak.
Ok. Let's say there exists a market when there are 10 (on average) 3 baggers (on average) available per year with little risk. Someone - an advanced AI, maybe - is clever enough to spot them and bet on, returning around 100x or 10,000% in a year.
How would you distinguish this picking ability from luck, within a year? You appear to be claiming there is no possible way to do this. That vastly outsized returns do not show a likelihood of higher EV according to you, but in fact are indicative of nothing but gambling.
Do you agree that this position requires the prior assumption on your part that no such large edges exist in the market? On what basis do you make this assumption?
Ok. Let's say there exists a market when there are 10 (on average) 3 baggers (on average) available per year with little risk. Someone - an advanced AI, maybe - is clever enough to spot them and bet on, returning around 100x or 10,000% in a year.
How would you distinguish this picking ability from luck, within a year? You appear to be claiming there is no possible way to do this. That vastly outsized returns do not show a likelihood of higher EV according to you, but in fact are indicative of nothing but gambling.
Do you agree that this position requires the prior assumption on your part that no such large edges exist in the market? On what basis do you make this assumption?
Think of the "Infinite Monkey Theorem" - some analysis I read a while back (can't remember which book it was in though) showed how, using fairly conservative estimates, that the p-value of Warren Buffett being able to beat the average market return was still only about 5% (as of the year 2012-ish IIRC).
It's up to you to decide if 5% is acceptable proof or not, keeping in mind that it suggests that in 1 in 20 "parallel universes" there would have been a Warren Buffett level winner by luck alone.
So, I can't really answer your question other than to say no level of return would impress me over 10 bets; simply because it would be so easy to show an example where, with near certainty, the results are easily replicated by luck alone.
It's up to you to decide if 5% is acceptable proof or not, keeping in mind that it suggests that in 1 in 20 "parallel universes" there would have been a Warren Buffett level winner by luck alone.
Thats not actually what it suggests though. It suggests that if the markets were purely luck then in 1/20 universes there would be a Warren Buffet level winner by luck alone. An important distinction.
The relevant part of the post you quoted:
Quote:
Originally Posted by ToothSayer
Do you agree that this position requires the prior assumption on your part that no such large edges exist in the market? On what basis do you make this assumption?
Yeah. These edge case probabilities change wildly based on your priors. I'm not convinced that the view "a very very lucky person could have done it, hence he's probably lucky" is the correct view for something like the market, which is not a random spin but is sometimes moved by possibly discernible human sentiment and news.
You have to assume to very high probability that there exist no large edges in the market to take the view that juk is giving.
Last edited by ToothSayer; 01-08-2018 at 01:48 PM.
Everyone knows you can't spend money. You can only spend p-values.
It will take at least a few annual bets at least before I am convinced that it is likely that 500% gains can be had more than 1/3 of years. That will, of course, be fun.
And after such multiple annual bets, say 20 years from now, you can invoke non-stationarity: "it may have been true that 500% gains every third year were possible from 2017-2037, but it's no longer possible in 2038!"
The cleanest test of whether I am convinced is my behavior in the future when offered the same bet. Snap calling is as clear a sign as any that I am not convinced.
not sure what this is proving. you are knowledgeable...Id bet against you but you may know how to play double leveraged etfs and this is a breeze. not going to bite. love to watch tho
It's proving that Brian is willing to give me free money because he likes me so much.
How are double leveraged ETFs "a breeze"? If money was as easy/brainless as buying double leveraged ETFs, everyone would be buying them. If you think 6x is +EV at 2:1, plenty of people in this thread would be willing to give you that 2:1. Real time posted trades are foolproof as proof.