https://outsidetheasylum.blog/manifold-search/
Please be honest.
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ89 | |
| 2 | Ṁ73 | |
| 3 | Ṁ12 | |
| 4 | Ṁ8 | |
| 5 | Ṁ0 |
People are also trading
@firstuserhere Could you give me an example of search criteria that are showing you no results, that you believe should have had results?
@IsaacKing I did try using it to find markets that were about to close, but often the market probabilities were out of date, so I didn’t get too much use out of it. I was trying to find markets closing soon and probabilities that weren’t so decisive, that is for me being able to specify a range of probs rather than scrolling many markets and having the correct probabilities shown I think I would find useful but I couldn’t use it much for this purpose. I only used it a handful of times though.
@parhizj I've updated it to fetch markets 100x faster than it was originally, so market probabilities shouldn't be more than 10 minutes out of date now. (Starting 10 minutes from now.) Is that good enough, or do you need it even faster?
@IsaacKing Here is an idea without using more complicated queries... rather than clicking on the link going to a market have it add to a "list" that pops up a frame on the side of the window (those selected markets have that probabilities are updated every 30s or so); or if you want to keep the links just add a button/checkbox to the left side. This way I can scroll through many markets and find ones that have interesting titles and acceptable probabilities to create a list. And then once I spend time building up a large enough list I can look at the up to date probabilities in the list and click on them one by one to examine them in detail. This will reduce task switching tremendously. As it is now, clicking on a link to open a new tab, reading the probability (and hopefully the description/research and doing the bet) and then closing the tab and then trying to resume where I left off is too much task switching.
@parhizj Hmm, interesting idea! I think it'll probably be easier for me to just fetch more up-to-date info and keep the current UI; would that also do what you want, or is there some other reason you'd like what you described?
I've added a custom sort option, which would satisfy your other request. Here's a link that will show markets that are closing soon and have probabilities close to 50%.
@IsaacKing The weights idea take some tuning and guesswork but it looks like it might be much more useful than before. I will try it out for the next week and report back.
@IsaacKing Can find examples where probabilities are definitely more than 10 minutes out of date.... (eg this one is at 22% 2 days ago)
[Ṁ200 subsidy] Will Tropical Storm Philippe and Tropical Storm Rina merge into one system? (37%, Closes in 1 week, M$310 liquidity)
For anyone with some extra mana lying around, feel free to manipulate /IsaacKing/if-i-were-to-make-a-market-search-p
@RobertCousineau You had expressed interest in such a thing a few months ago, just letting you know it exists now.
@IsaacKing It's super fast and when I tested it I found what I was searching. So yes.
Let's see if I will use it in the future for some actual searches. I'll report back here.
Looks nice, though haven't needed to use it yet.
The search is kinda laggy.
You could precompute/cache the normalized strings for the markets as just a new field on each market. Or maybe a separate object original => normalized since you update market data every 5s (do you really need that often, once a minute would be more than enough for typical cases I think of?)
Creating all the intermediate arrays via filter is a bit eh, though might be fine.
The render function could manually construct the elements so you don't force the browser to parse the html when setting the innerHTML.
May be able to get some benefit by creating the searchResults element in render and then appending each market entry, so then the browser only has to update once. Though maybe browsers handle that just fine.
@Aleph Thanks for the suggestions! The biggest problem was actually that it was filtering every field for the empty string, since "".split(" ") doesn't return an empty array. Fixing that brought search times from ~800ms down to ~150ms, and then pre-normalizing down to 50ms, which I'll call good enough for now.
If you notice any more issues, please let me know. :)
@IsaacKing I was thinking sorting by size and close date. Is there something else like total bets or total mana that’s easier?
@IsaacKing Yea, that’s seems pretty good.
I am not sure I’ve found it useful yet. Not obviously better to me than the one in the app though
@IsaacKing What are the problems with the existing search that I should expect this to work better for?
Mainly just annoying to leave the app to do a search, and lately haven’t had as many issues with it
@Tyler31 I mean if you somehow find the native search to be good for your use-case, then I doubt mine would be useful to you.
I find it terrible as it's super slow and uses extreme fuzzy matching, so it often returns markets that seem to have nothing to do with the text I typed in the search box.