Long Bets by Confidence Level |
December 8th, 2019 |
bets [html] |
- Both parties put up the same amount, at least $200/each.
- Long Bets effectively runs a donor-advised fund (DAF).
- When the bet concludes the winner chooses a charity to receive the money.
- The charity gets the initial stakes, plus half the investment income.
Let's say I claim we'll have talking horses ten years from now, and you're skeptical. You consider betting $1000 against my $1000 via Long Bets. If you win you'll get your $1000 back, my $1000, and half the investment income which (figuring the stock market returns a nominal 7%) will be ~$967, for a total of ~$2967. On the other hand, if you had just put your $1000 in a DAF you'd have ~$1967. Is this a good deal?
Provided putting the money in a DAF for at least that long would otherwise be your best option, if you're 100% confident that (a) you'll win and (b) Long Bets will still be around, then it's a solid deal. You're up about 50%. On the other hand, the less confident you are the worse the deal looks:
(sheet
you can copy and play with)
For example, at 60% confidence you're neutral at 6 years, and negative after that. At 75% you're down to neutral at 16 years. At 90%, 32 years. At 99%, 75 years. For an organization trying to promote long-term thinking, it's surprising they would choose a fee structure that penalizes long-term bets so heavily.
Comment via: facebook, lesswrong, hacker news
Binary options meets long bets, interesting idea, but illiquid Markets make lousy predictions
Without variable odds, there is no information here to be discovered, since there will always be as much money for & against the bets. The liquidity doesn't come into it.
Unmatched offers could still provide some information.
is 7% a reasonable 90 year return? I would guess closer to 1-2% would be reasonable over that time frame, which puts you in the black at year 90 even at 90% confidence.
These are nominal returns, but when you say 1-2% it sounds like you're talking about real returns?
I use nominal returns in the post because Long Bets keeps half of the investment income without any accounting for inflation.
The variance over 90 years must be huge.
The humany truth is that public bets, particularly long bets, are less about making profit and more about stating your view boldly and publicly. Within this context opportunity costs barely matter.
Just to underscore the point I understand you to be making:
Good name in man and woman, dear my lord,
Is the immediate jewel of their souls:
Who steals my purse steals trash; ’tis something, nothing;
’twas mine, ’tis his, and has been slave to thousands;
But he that filches from me my good name
Robs me of that which not enriches him,
And makes me poor indeed.
https://www.enotes.com/shakespeare-quotes/who-steals-my-purs...
It seems a bit limiting that they only deal with equal stakes/odds, i.e. we both bet $1000 for/against an event. Allowing for variable odds would make the deal better for events which you had less confidence in, solving the problem highlighted.
Variable odds don't solve the core problem which is that "half the investment income" is a high enough fee that you need a very large edge before you do better than just putting the money in a donor-advised fund.
Changing the odds is the only way to increase your edge... but admittedly, that 'half the income' will become a very large fee, such that the odds you need to offset this would make it unlikely that someone would take the other side of your bet.
Hi! I wrote the original software for Long Bets some years ago. I can't speak officially about the project, as I was just a coder. But maybe my take here is useful.
I enjoyed the article, and think the analysis is correct as far as it goes. If your goal is to donate maximum dollars in the future, a Long Bet may not be your best choice. I just think that's rarely the goal.
The main issue here is that nobody does a Long Bet so that their charity gets more money eventually. The Long Now's mission is to promote long-term thinking. Long Bets in specific promotes clarity and accountability via carefully-stated long-term predictions and bets.[1] Take, for example, Bet id 362: http://longbets.org/362/
St. Warren of Omaha was not trying to maximize money to his charity. He has approximately infinite dollars; this was a rounding error. He was trying to make a point very publicly. He thought that hedge funds were generally horseshit, and that most investors would do far better making simple, straightforward choices rather than giving their money to people who promised to do wonders: https://money.cnn.com/2018/02/24/investing/warren-buffett-an...
What he was buying was not a specific dollar gain, but a clear, visible test of contrasting views. And it worked wonderfully; this bet was in the news off and on for more than 10 years.
I believe another goal of most people doing a Long Bet is to support the idea of long bets and long-term thinking, so I suspect that they look at the Long Now's rake as another donation to a charity they like. This analysis values that at zero.
I'd note also that for truly long-term bets, this analysis discounts the fact that the bettors will be dead. I think our current longest bet runs from 2002-2150. For a mere $1000, I doubt there's a donor-advised fund in the world that will agree to act as trustee for 150 years, selecting the most appropriate charity at the time of resolution. And of course no DAF will provide the clear record and eventual judging of a bet.
And really, if one's goal is giving money to charity, I suspect the best thing to do is give the money now. Whatever need they're addressing is presumably urgent enough that they'd rather spend the money today than put it in a long-term fund. And if it isn't (as is perhaps the case with the B-612 Foundation [2], which is saving up for a satellite), the charity in question is probably a better judge of exactly how to invest the money until they're ready for it. At the very worse, they money could be given with the restriction that it be used as an endowment.
So I look at this analysis as a good start, but definitely not complete given who actually uses the service.
[1] It's very much in the spirit of the Erlich wager: https://en.wikipedia.org/wiki/Simon%E2%80%93Ehrlich_wager
[2] https://b612foundation.org/
> The Long Now's mission is to promote long-term thinking. Long Bets in specific promotes clarity and accountability via carefully-stated long-term predictions and bets.
Unfortunately, they completely undermine this when they allow bets/predictions to remain on the website even after a challenger has agreed to take the other side of the bet but the original poster declines to participate. Long Now does't even make a note of this online.
> Take, for example, Bet id 362:
Note for other: this is a highly unusual example, since it's the most high profile bet on the site.
I did something similar last year with https://gwern.net/Long-Bets . The even-odds restriction is a major one, and I think sabotages all of LB's goals. Why do prediction markets like PredictIt do more in a week or so than LB has done in almost two decades? It's hard to say that LB has been more of a success in getting people to put their money where their mouth is or bringing more rigor to public discourse (compare how often PredictIt or other betting markets are cited in political analysis to how often LB is cited... ever).
I agree with your complaints and indeed years ago I independently wrote an email to LB complaining about the fact that predictions could remain on the site even after challenges had been rejected. However, directly comparing LB to PredictIt seems unfair. LB is just not a big priority for the Long Now Foundation, and I expect it took vastly less investment to launch or for continued operating expenses. It's not surprising it accomplished less.
Edit (in response to your edit, I think): Sure, public exposure has not been commensurate with contribution. Not shocking, but frustrating.
> LB is just not a big priority for the Long Now Foundation, and I expect it took vastly less investment to launch or for continued operating expenses. It's not surprising it accomplished less.
Sure, it's not a priority - now . Because it failed. But it had quite a splashy launch and they had high hopes and really pulled in their board of luminaries for the initial batch of predictions. It launched so long ago it could have dominated any emerging online prediction markets. It had the opportunity, a useful goal, a nonprofit backing, tremendous goodwill, and celebrity. But instead, even something like Augur, a weirdo cryptocurrency project which formally launched hardly a year ago, will turn over more in a month or two than LB over 2 decades.
> But instead, even something like Augur, a weirdo cryptocurrency project which formally launched hardly a year ago, will turn over more in a month or two than LB over 2 decades.
Maybe I'm just missing something, but Augur seems pretty dead to me? I only see three bets currently open (one about the UK general election, two about NFL games).
That being said, the Long Bets project was always about high-profile individual bets, not the larger markets that Augur seemed to be trying for.
What are you looking at, https://predictions.global/ ? An estimate of $2.6m at stake sounds pretty good to me, compared to LB which has like $40k at stake right now.
It's also worth noting that as part of the migration to Augur 2, Augur 1 is supposed to be gradually shutting down: https://www.augur.net/blog/v2-transition-update/ https://www.augur.net/blog/v1-cutoff-update/ So that may be reducing activity.
I was looking at the site that's linked from the augur.net home page, which now only lists two markets:
https://cloudflare-ipfs.com/ipfs/QmaM5kLjo21i2eCSiaDQh4S1dG4...
I don't know enough about Augur to understand what the difference between the two sites is. What does seem strange to me, though, is that predictions.global claims that $2.6M is at stake, but the top 10 markets sorted by "money at stake" only sum up to roughly $288k at stake. (The last one in the list is under $1k, too, so I have a hard time believing that the remaining $2.3M is all in the long tail.)
Thanks, that's useful background
It should be possible to turn an odds bet into an even-money bet with some financial engineering. Like "I predict that either we will have talking horses, or the last two digits in the Dow will be between 0 and 45." Probably there's a way to do it with cryptography too.
This only solves one of the issues you identify. But I've always seen LB as a publicity gimmick rather than a serious marketplace.
I'm not quite following what you mean. Sure, you can change the probability and make one side arbitrarily likely to win by appropriate wording. (Heck, just take one of the existing predictions which has ~0% probability - there's plenty of them.) That doesn't change the problem that you can only win at most twice the money you put up despite locking it up arbitrarily long.
I expect the people who make these bets to be overconfident.
"Amusingly, even though predictors presumably are at least 50% confident when they make binary predictions at 1:1 odds, by my count, over 3x as many expired predictions were false rather than true, even judging generously and skipping the hardest. So, the predictors as a whole are not just badly calibrated but systematically overconfident, and one could’ve made a considerable profit just betting against everyone blindly."
-- https://gwern.net/Long-Bets
Gwern also notes that Long Bets invests much more conservatively than I had been modeling (they don't just put the money in a standard index fund)
Not very wise to offer a bet that the LHC will destroy the world. How do you collect?
@Frederic you collect acausually on mother everett branch? Or something.
In your horse analysis, the $1967 is in today's dollars, but the $2967 is in future dollars. The longer the bet time horizon, the more the time value of funds will impact the comparison. Do you just consider that separately, not at all or rolled into the expected value payout?
Bob the way I handle this is by seeing how betting compares to putting your money in a donor advised find for the same length of time. Both the $1967 and $2967 are in the same units (money going to charity in 2030 in 2030-dollars).
Jeff Kaufman thanks for the explanation.
Hmmm. Depending on the charity, I am not sure the donation would be invested for monetary gain. I guess when I think of charities I donate to, it is for more immediate need.
Yes, if you wouldn't normally put your money in a donor-advised fund (DAF) then a Long Bet is even less attractive.
The "half the investment income" bit seems like a pretty clear mistake to me were this to scale, for this reason. Good point!
Wait, does that mean Long Bets keeps half the investment income? That seems like an outrageously high mediation fee. For a 10 year bet, that’s roughly 5% carry per year for doing almost nothing.
Yes, Long Bets keeps half the investment income. This is very high.
Though there are even worse options: https://www.charityvest.org/ keeps *all* of the investment income!
My read on their interpretation is that they expect the investment income on the losing bet (which is what they keep) to be thought of as a donation to the Long Now Foundation, rather than as a mediation fee.
1. The calculation here seems to consider only
but (1) if you don't place the bet then presumably your counterparty will do something with that money (which might have value or disvalue from your perspective) and (2) if you place the bet and lose then all the money goes to your counterparty's chosen charity (which almost certainly will have value or disvalue from your perspective).
Unless your expectation is that your counterparty's chosen charity has negligible effectiveness (for good or bad) relative to yours, it seems to me that this calculation is unlikely to be the one you actually want to do.
2. My impression (which could be very wrong) is that making a Long Bet is usually at least as much about raising publicity as it is about directing money where you'd like it to go. It's a thing pairs of people do when they want to get everyone talking about the issue about which they disagree. If I'm right about this, then in many cases the final disbursement of money is likely to matter less than the consciousness-raising.
Great point! I think this is often the case for bets between people who do or don't consider effectiveness in choosing charities, or where people have sufficiently strong value disagreements that each thinks of the other's charity as neutral, but you're right that this is unusual.