I'm going to test myself as a trading bot developer (here on Manifold) and would like to learn more on this topic. Post your tips as options, I'll choose ones which were most helpful for me.
Feel free to write just a gist as an option and elaborate more in comments :)
(I didn't find information anywhere on how do these type of questions resolve, but let's say I'll choose my top three.)
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ155 | |
| 2 | Ṁ40 | |
| 3 | Ṁ12 | |
| 4 | Ṁ10 | |
| 5 | Ṁ4 |
There are some Manifold accounts that are known for making absurdity irrational bets, like @Spindle, @ButtocksCocktoasten, and @TheWiggleMan, as this very market demonstrates (though I think the Wiggle Man might be permanently inactive). I know someone (I think it was Conflux) suggested making an "anti-Spindle" bot at one point, but as far as I know, no bot has ever actually used that strategy. You could probably make a good deal of profit by just hard-coding some of these users into the bot and telling it to bet against them whenever they make a trade that moves a market significantly. I'm pretty sure @acc already does something similar (it bets against users with a history of bad trades, and it bets in the same direction as users with a history of good but underconfident trades), but it doesn't take even close to all of the profits that could be made by betting against joke users.
@JosephNoonan botlab is meant to do that but I haven't been keeping the manual troll labelling up-to-date.
@MrLuke255 I wiggle so you profit and I take the L. Some people didn't like this, it violated their worldview of what's rational. So I wiggled more and more.
I know Michael Wheatley just told you not to do this, but including arbitrage in your bot couldn't hurt, could it? I still see tons of markets that have probabilities out of line with each other that arbitrage bots haven't touched. You could also look into arbitraging other than just "These two markets are guaranteed to resolve the same way." For example, markets on mutually exclusive things, like who will win the presidential election, should never add up to more than 100%, but often do.
I'm not exactly sure how most arbitrage bots work (do they use AI to determine which markets are equivalent, or the creators manually tell them which markets to arbitrage?). But either way, if you can find a better method of finding arbitragable markets, or make your bot more versatile with the types of arbitrage it can do, maybe your bot can do well despite the competition.
@JosephNoonan they probably add up to more than a 100 as it locks up too much mana to keep the prices correct.
Well, I only have one piece of general advice: test every strategy manually on a few markets before you automate it to save your own time. But I can give you some tips on what to do to get started. Here are two simple strategies that are pretty easy to implement.
First, you can try reversing big bets. On Manifold, when you buy a share with a lot of mana, it sometimes moves the probability past the "real" probability if you aren't careful. You can capitalize on this mistake by detecting all big bets and reversing them a bit. Examples: @acc, @Botlab. It's simple to implement, but there's plenty of competition, so you won't make big profits unless you're faster (or better) than the other bots.
Another simple strategy is arbitrage. You pick two markets with equivalent questions but different probabilities (there are plenty of these) and bet YES/NO until they are the same. You will always be profitable if you ensure you have the same payoff in both markets but in opposite directions. @Bot does that AFAIK.
@JosephNoonan Yep it might be, but:
1) I haven't realized it until Ponzi schemes came out recently 😅
2) I don't have mana to give out 😐