Manifolio is a calculator I have built which optimises the size of your bets accoriding to the Kelly criterion (see the calculator itself or the announcement post for more info). It has just been approved as a Chrome extension, woohoo!
You can install the extension here.
Having it as an extension rather than a separate website makes it a million times more convenient to use for non-power users. I didn't expect it to get that much usage as a separate website tbh because you have to bookmark/remember a whole separate url so it's hard to get into the habit of using it.
Will there be at least 40 NET installs by 1st September (i.e. installs minus uninstalls, according to the chrome developer dashboard)?
Attemping to fix this market by installing the extension is not only allowed, but encouraged! Although I've set the target slightly high to account for people e.g. double installing it in both Chrome and Brave, stuff like this does count towards the total because I have no way to determine truly unique installs.
Resolution criteria notes:
The market closes before the cutoff time, this is not a mistake it's to avoid last minute shenanigans
The cutoff time for counting installs is midnight on 1st September UTC (i.e. the end of the day)
This market will resolve early if the high watermark of net installs is above 40 at any time before the cutoff date
I've never published a chrome extension before so I'm not sure how "live" the numbers are. I will go with whatever the dashboard says, I assume it will be no more than a day out of date though
I may or may not give running updates in the comments
If I make a firefox/other extension these installs also count
1,000
3.00@WilliamHoward I think right now manifolio treats my own limit orders as normal limit orders I can buy into to make a bet. They should probably be excluded (maybe with a warning that you'd have to cancel them).
@WilliamHoward Btw in case you were not aware: I suggested the warning because manifold will in fact let you buy into your own limit orders. In this case it would just mean you place a smaller bet, but it's still confusing UX. I guess ideally you'd have it ask for confirmation to cancel any opposing limit orders you'd run into if you bet through manifolio.
/
@8 I'll check the official numbers in the dashboard in about an hour, but yeah it's looking good
Edit: Nevermind.... I re-read the description...
@WilliamHoward if the data is 'delayed' by about 48 hours, will the user count be looked at on september 3rd instead?
@Dennis5a87 No I'll take whatever the number is on 1st September, even if there is reason to think waiting another day might get it over the line
Finally got around to checking this out. Looks really cool! Well done for making it. Unfortunately I had the same issue as brubsby below when using the web app version, which puts me off installing the extension.
Obviously my first question is "how do I turn this off", but my second, more interesting question is: is it right? Am I being unwise letting my total loans get bigger than my portfolio? Should I wait until that's not the case before betting again? (How would I even do that? I guess I have to just wait for some markets to resolve in my favour?) But it really feels like there are some extremely low-risk opportunities on Manifold right now (AI extinction by 2030 is probably the most clear-cut one, but LK99 is in my opinion almost as certain and shorter timescale), so it's very counterintuitive that I would make more money by foregoing those opportunities.
It's right under Kelly assumptions, if the goal is the assumed Kelly goal (maximize expected log wealth). I think the goal is probably a fairly reasonable assumption (but that's debatable), but the assumption that ruin (reaching 0 wealth) actually ruins you is clearly wrong on Manifold, since you can obtain mana through means other than bets. Not sure what the best way would be to model and fix this, but I assume there's existing literature on it.
@Fion also like, the consequences of "ruin" on a website like this are: "oh okay i guess i'll log off now" or make a new account. maybe the message can still be there as it's interesting and informative, but as it is now i can't even engage with the app.
@Fion also, I thought about it more, and it's using the market odds to predict event occurrence to do the P(ruin), when I believe certain events that I'm greatly overleveraged on have a 0 or 100% chance of happening (or very close), which, even with strict kelly betting should reduce the ruin probability enough (versus using the market probability). of course there is the chance that I'm wrong about my probabilities, but that doesn't factor in.
even when betting on the overleveraged markets, setting probability estimate to 0 or 1 doesn't sub in that probability in the P(doom) calculation (even with no/max deferral factor), so I really feel like this P(doom) lockout is behaving wrongly and in an overly paternalistic manner.
@Fion See my reply to brubsby below. TL;DR I'm planning to add an overdraft setting to make it work at all for people who have a lot of loans while keeping the essence of the Kelly betting logic intact, hopefully this will make it useful for both of you.
Am I being unwise letting my total loans get bigger than my portfolio?
I'm not really sure about this. The calculator is quite conservative in some ways, so your actual risk may be a lot lower:
In real life it can be possible to avoid complete ruin by selling out of other positions when things start to go badly (whereas the calculator assumes you just wait for all positions to resolve)
But it's also quite un-conservative in other ways:
It assumes markets are uncorrelated
It assumes the market probability is correct
Personally I just try to avoid accumulating a lot of leverage by not holding positions for very long, but I haven't worked through the maths completely carefully to make sure this is actually the best strategy. It may be better to go with leverage + a lot of diversification
Should I wait until that's not the case before betting again? (How would I even do that? I guess I have to just wait for some markets to resolve in my favour?)
I would recommend selling out of some old positions that don't have much remaining ROI for you (bc they have long time until resolution or the market has already converged to your probability estimate), you repay the loan when you sell the position