It looks like it was mostly college kids and analyst/associate level employees. I'm sure most traders could get it right, but I wouldn't be surprised if when scaled up in sophistication more advanced people would make similar mistakes.

The answer is that these were young college men who didn't give a **** about the EV of the $25 offered, and wanted to gamble it up for fun.

Extrapolating this out to anything other than "bored young college men offered $25 on a gambling game are going to donk around" is a big fat lol.

The abstract:
LOL @ the bolded. The morons who did this study should take some philosophy of science classes, or maybe just go for a walk and get some fresh air. This is a classic cuckademia, a worthless study with meaningless results extrapolated out to the stratosphere when it hasn't even left the ground.

While I agree with you, I also think it's a pretty good interview question for trader/quant roles. But they asked a bunch of dumb economics/finance majors, probably in a context where they didn't give a ****, so of course their results are going to be stupid

my understanding of it was that they attacked spreads that they deemed inefficient, and made tons of money by being the biggest player and unlimited martingaling. their big problem was that everyone else started doing the same stuff too; and instead of adjusting and scaling down and finding a new opportunity, they continued applying their gigantic size in spreads that either had less of an edge, or spreads where the strategy of correlation maybe doesnt even apply (like shell vs dutch shell).

i'm sure there are more things that went into it that i forgot about, it's been like 4 years since i read the book.

I think this is something we've talked about and agreed upon dozens of times. When you see a good set up, you should essentially have a default bet size. Betting small because something feels off makes little sense. I've seen guys take tiny size or even sit out through some of the best trades because it "felt like a trap".

On your second point, imo it's not that they are taking larger size periodically, they are frequently trading too small and rarely getting to full size.

Yes, mathematically variance doesn't matter. But I think in reality it definitely does, at least psychologically. I've seen guys take a strategy that is consistently profitable every year and have losing years with it because of irregular sizing. People frequently will reverse martingale a strategy and trade it bigger and bigger until they give back all the profit. I think the guys who give up some EV to reduce variance probably generate more EV over their entire careers. I think lower variance leads to better work ethic and a more disciplined mental game.

Didn't read the article but I'm pretty sure the rules of that game were whoever ends with the most money wins or something in which case betting contrarian at some point even if -Ev could be optimal. RoR is so high because of a fixed number of iterations and fixed money supply significantly less then their net worth.

If they didn't do that I'm sure they put a cap on how much you could win.... 2^10 is 1024 so they could lose up to $1,500,000 in 10 iterations. Pretty risky experiment for ****ty data.

I highly doubt this experiment was run like actual investing or a casino where your earnings were fully realized and betting unrestricted.

yeah this is just very wrong, problem is that trade environments are not solvable in terms of interpreting EV so there's nothing i can throw behind this beyond "you are wrong". not all opportunities are created equal, this goes for many things in life. there are plenty of poker analogies you could make. fact is, certain trades are juicier than others, and worth risking more. to deny that fact is to deny that traders can even see relative edge. if you feel that strongly about it, i question your ability to see edge at all.

I'm not denying that traders can determine whether one opportunity is higher ev compared to another. Some opportunities are definitely way better than others. I was just saying that when someone makes only like 5-10 outsized bets in their career, it's hard to judge them off of a minuscule sample size.

that's fair. frankly anyone with a long trading career and only 5-10 giant trades is prolly doing it wrong, idk. there are tons of diff bankroll/edge strats and sometimes a lower variance / fixed wager one may be better for certain personalities. i'm a nitty grinder at heart so there are times where i may be quicker to book the winner even tho higher variance / "stretching it out" may be higher EV. i still maintain that the psychotic nits are generally the ones doing best in trading, grind grind grind and then risk 100x their normal betsize on something that looks really good etc.

Or you could have taken 3 minutes to read and see what they actually did instead of writing that all out and thinking people care what you think they might have done in the experiment

That probably also describes the worst traders. Err, ex-traders.

Yeah, optimal bet sizing given known odds and outcomes is something you should know off the top off your head. How are you sizing correctly in somewhat unknown situations if you can't get known situations correct?

Right, but imagine you could just "get over it" and trade appropriately while taking larger size? You would crush the people who trade appropriately taking smaller size. Is this really such a hard thing to do?

Your position here is that doubling or more your EV is -EV. That's an extraodrinary leak...if you can work around even a part of the source of that leak, you're gonna do a lot better. Compounded it leaves the other strategies in the dust.

Imagine you bet 5% of your stack and double every trade.
Now compare with someone who has the same entries and exits and bets 10% of their stack and doubles every trade.

After 100 trades:

- Regular sizer has 131x his starting money
- Double size has 13780x his starting money

Basically, double size is 104 times richer than regular sizer.

For ****s and giggles, I calculated the (in sample statistics used from 1/1/2005 to 3/3/2017) Kelly for the strategy my son found over Thanksgiving Break. Since this was a ridiculous exercise, I didn't take into account trading costs or interest rates. Kelly says 397% is the optimal bet size.

In the situation they described (30 mins, start with $25, max payout $250, 60/40 coin paying even money) the correct answer is actually far too complicated to solve quickly, but Kelly is a good approximation for early strategy. The most important thing is getting rolls in as fast as you can with a reasonable bet size. Anyone who spends more than a couple minutes trying to solve it before starting to flip automatically fails.

If you can get 300 rolls in with an average EV of $1 you basically can't lose, so I'm pretty sure betting something as nitty as 10% of your roll every time is essentially a lock.

Why is this hypothetical trader not using a huge amount of leverage? 10% of bankroll is way too low for a 100% win-rate.

all a function of how bankroll is managed. very analogous to poker. difference between the best and the worst is how they interpret the edge (the worst may be overconfident) and how they manage their downside (the best risk a fixed % and stick to it, the worst could risk it all under duress)

I can't imagine that I could find an actual financial trade that I was confident enough to put 100x my normal bet-size in, even with some high-grade pharmaceutical aids custom-designed to eliminate the relationship between my confidence and reality.

100x on Mayweather would have been good

It was probably the best bet of a lifetime, but there was some small chance that Mayweather pulled a hammy climbing into the ring.

I'm not really talking about smaller or larger, more so consistency. My main problem is with people who are varying size wildly based on untested theories. Trading big is fine and probably optimal if you do so consistently. But if you trade big consistently, then you're just trading your normal size.

Yes, kelly is just absolute potato when you plug in stats from most any trading system

Some guy who's in the early stages of filing for a patent in a field you're intimately familiar with has a weak understanding of how to value it and is terrible at communicating his idea. It could happen. It just will basically never happen via reading annual reports and articles on the internet.


If someone had been trading for a long time and consistently took small position presumably due to lack of confidence in their edge, if they one day went balls deep in a company, i would be inclined to believe that there was some substance to it - and expect it to be pretty unlikely that the position carried the potential for financial ruin.

The worst traders are probably going to be throwing hail marys from day 1.

You mean untested theories about the size of their edge, or theories about hedging for risk?

Betting the same size for large edges as for small is a pretty clear mistake.

abbadabadu, he means ev divided by expected variance.

Do you consider it possible that risk is somewhat discernible, and can vary by trade?

Do you bet the same on your merger arbs as you do on your short pump and dump nano cap bio?