This is a followup to my previous post, and partially a reply to Brad's substack post.
Manifold's margin loans help with time discounting of long-term bets, but that is not the only thing they do. As Brad writes, they do a far from perfect job of it. Interest rates are mostly better at serving this goal. Interest rates alone don't solve the problem, and margin loans also provide several unrelated benefits in a highly cost effective way.
The core reason interest rates don't fully solve the problem is that good bettors have an opportunity cost (they expect to make money betting), and the interest rate doesn't change that. Rational bettors generally should prefer a secure return over one with risk, and so should only bet when they think they have an edge -- and it matters how long that edge takes to realize. Interest rates really only address the opportunity cost of not investing in risk-free investments (bonds). The Manifold margin loans help somewhat on top of interest rates, because they ensure the capital is only locked for a fixed portion of time (effectively 1/(margin loan speed)), regardless of the length of the prediction, in the long-running case. The interest payment system is doing the heavy lifting, but they complement each other.
Margin loans also provide an increase in the velocity of money on the platform. People bet more. It is not a priori obvious what the desired ratio is here, and I see no reason to suspect it should naturally be what you get with no margin loans. Manifold should benefit from increased activity and increased trading. Traders with a diversified portfolio should naturally be willing to take on risk at a level that corresponds to some margin usage (assuming it's cheap enough). Low-multiplier margin loans, with some element of restriction for risk management (especially on large accounts), should be a win-win.
Margin loans also provide a daily incentive, which is especially important to free to play or otherwise small accounts (anyone where the daily / weekly bonuses are relevant). Log on, click button, get liquidity. The risk of the loan to both trader and Manifold is low. And it provides a daily dose of liquidity to drive engagement, that costs less for Manifold than creating mana from the mana faucet. The cost to Manifold is simply the risk that the account is not diversified and ends up negative; risk management (such as the existing per-market cap approach) helps here. For this use case, the Manifold approach (daily loan as a small portion of position) is better than a traditional approach (as needed with risk management limits).
The daily loan system also naturally acts as a risk management system: a trader can't see a single large bet they want to make and use all their margin on it at once, they have to wait for that margin to become available over time.
For large accounts, the trader wanting margin would probably prefer a traditional approach of as-needed margin loans rather than daily, and be willing to pay interest on it.
How to do risk management well is a whole separate topic (and post), but some things worth considering:
How long does it take to reach 2x or 3x leverage?
How many markets worth making large bets in does a trader need to find to reach 2x or 3x leverage? (For a simple per-market cap, it's leverage / (per market cap), so 2x leverage and a 5% cap means spreading out across 40 markets. Hopefully that's highly diversified!
Some closing thoughts on things that might work well in this framing:
Daily loans have a leverage cap or an absolute cap, which can be removed or raised by agreeing to pay interest on the portion in actual use. "Balance" would then be "total avialable liquidity" and interest would be owed on (total investment - total account value - free loan cap).
The cap should not be at zero, because you want to encourage free to play accounts to use the system and not be scared of it.
Raising caps or otherwise loosening risk management might be a good thing to tie in with real-money spending, such as a subscription or a bonus on large purchases.
Overall, interest payments on positions should be viewed as complimentary to margin loans, not substitutes. And daily loans vs as-needed traditional margin loans should be viewed as a mix of engagement optimization and risk management.
If manifold adds interest rates, realistically what I will immediately do is open a longterm market called "resolves YES" and stick ~my entire balance in it, then use it as a piggy bank. Not sure how I'll get around the 99% cap/liquidity but I'll find a way.
I'm not sure if that's good or bad from the site's perspective, but it is true.
@EvanDaniel Could pay interest only on ranked markets. If you want to use a ranked market as a piggybank you need to pay the fees to get it in and out.
(I think it's okay to use editorial discretion to prod people into betting on certain types of things rather than others. And creating certain types of markets but not others. And you can change the criteria to suit.)
@Eliza This is a good way to discover that the boundaries on "ranked markets" get really tough to police as soon as there's a reason to want ranked markets that have a certain property. Want to gamble? OK, we've got binary futures on stock prices with arbitrarily weird structure. Want bonds? OK well we can bet on a diversified portfolio of death markets and SPX > 1000 at EOY.
@EvanDaniel I feel like it would be in Manifold's interest to make editorial decisions about what types of markets earn what type of return.
I agree with much of the above.
more broadly, I don't see how any implementation of universal interest payments make sense for Manifold. I understand @brod 's basic argument: there is a time value of mana, loans only partially fix this, whereas interest payments are a complete fix. I disagree with this framing.
loans are intended to fix a basic problem: users run out of mana, and we want them to keep "playing" on the site. interest payments address a very different problem—markets resolving far in the future are bad EV. that is a much less important issue. most people are not terribly sensitive to subtle differences in EV. the primary reason it's hard to motivate people to predict long term things is not EV, it's the lack of feedback/reward (will Manifold still be around in 2029? who knows). interest payments do little for users who are mana constrained.
in other words: interest payments address the issue where a user is deciding whether to bet on market A (resolving soon) or market B (resolving next year), and it makes both options equally appealing, in the very narrow sense of the time-value-of-mana. they can now freely choose the market they have the larger edge on without sacrificing EV. but that's not the issue—ideally we'd like them to engage with both market A & market B. loans are far better for that.
interest payments are enormously expensive (in terms of mana), whereas the only "cost" of loans is the occasional account going negative. there's plenty to tweak there (I was the main proponent of per-market net worth loan caps), but the occasional negative account prints a trivial amount of mana relative to any plausible structure of interest payments.
fwiw since markets are user generated and endlessly flexible, interest payments on investments is ~equivalent to interest rates on balances for sufficiently active users. there is no realistic plan for telling people "no only invest that spare balance in super serious long term markets, nothing silly", given that the site encourages many silly markets. so if interest payments are implemented, i'm sure soon enough someone will provide many options where it's trivial for me to invest my 2m balance risk-free and get full interest payments on all of it. so i don't buy that it will be a relevant motivator for me to go out and find many more markets to trade in.
@Ziddletwix I basically agree. They're different tools, and they solve different problems, and both are good and complimentary. But there's also some overlap, and enough so to cause confusion. That confusion is also (if I remember my history correctly) exacerbated by the fact that Manifold's loan system was launched in part to address the long term markets issue, for which it isn't actually as well suited as other options.
users run out of mana, and we want them to keep "playing" on the site.
I think this entire saga is all due to this working too well -- now they want people to run out of mana and then buy more mana.
That confusion is also (if I remember my history correctly) exacerbated by the fact that Manifold's loan system was launched in part to address the long term markets issue, for which it isn't actually as well suited as other options.
This is correct but even more so -- the primary goal of the loan system was specifically to address long term plays. If there is another mechanism for long term betting, I don't think they would actually want the current loan system, or at least it would be adjusted to be significantly less important. They would string users along with "daily bet bonus" and similar instead. Just enough mana that they're tempted to buy more.
I think this entire saga is all due to this working too well -- now they want people to run out of mana and then buy more mana.
Well yes. But can I offer some alternatives?
Charge interest on the loans for large users
Make large loan caps only available with a subscription
Provide mana sinks, like badges you can buy, a merch store, charity donations
Provide more high-quality markets. Reframe it: people don't have too much mana, they ran out of markets.
Better limit order support: markets we have could support a lot more liquidity; it's a UI/UX issue that people don't place larger bets.
They would string users along with "daily bet bonus" and similar instead. Just enough mana that they're tempted to buy more.
This would, IMO, be short sighted. You can provide the same amount of daily liquidity trickle with less mana faucet usage via combining the daily bonuses with the daily loans. They complement each other.