conformist

Yeah and this is a much more intuitive way of generalising from the n = 2 case. Weights are proportional to inverse variance even for n > 2. Importantly this assumes independence so it doesn’t translate to portfolio optimisation very easily.

In Germany, cheese with stinging nettle is a thing and I remember eating it a few times as a child (Brennnesselkäse), typically Gouda-like types.

It’s fine to be non-rigorous as long as you’re not a jerk and treat your own conclusions on relatively little data as preliminary. Sometimes it’s good to have some information with relatively large error bars and be willing to update quickly rather than ignore information because the error bars are large. From the article it doesn’t become apparent that the author wouldn’t update her views quickly as new information comes in (eg by talking to someone). So yes the observations in the post are not scientific, but there’s nothing preventing them from being the starting point of rational reasoning and behaviour.

Conditional on “the study being published and getting attention” the real effect is likely smaller and not larger.

Eg if you assume there is a real effect plus a lot of noise, given the study has been published etc the noise will have more likely acted in the favourable direction.

IMHO given the relatively large size of the effect it seems quite likely that the noise part is in fact potentially large (this is much more subjective) which makes is less clear that there is measurable signal at all here. I’d have to see a lot of replication or a very strong explanation of the underlying mechanism to believe the magnitude of the effect, but will very easily believe the sign (with a small magnitude).

This is very common in finance. Knowing when finance research that made right predictions with good justifications falls into the "Gettier category" or not is extremely hard.

Yes, but it's potentially more subtle - there's a competition between professional players and casinos, in particular online. There's an interesting bloomberg story on this: https://www.bloomberg.com/news/newsletters/2023-04-06/meet-n...

There are many people in group 1 in academia eg in physics and maths that are comfortable with latex and scripting languages but mostly use email to share files. Anything that helps them organise their collaborative work better without having to deal with git helps (eg see eg success of overleaf).

I’m a trader

This is bizarre over engineering:

https://github.com/goldmansachs/gs-quant/blob/51a7ff1afb722c...

In the mathematical sense this is clearly not true - it’s easy to come up with a smooth single mode distribution that doesn’t look like a bell.

Published in an MDPI journal so the prior should be that it’s not serious research, unfortunately.

No, eg (Corr(A, B) = 0) and (Corr(A, B) > 0 | f(A)) can say something nontrivial about f.

While this might well be correct, the argument is missing that academic qualifications conditional on growing up in the US (as supposed to in Afghanistan) are still a proxy for potentially useful skills. In other words, some of the enemies might have had comparable academic qualifications had they grown up in the US.

How about 220 Hz music and 880 Hz?

(Obviously this study won’t replicate and categorising music by the frequency of the note A is not a reasonable thing to do.)

Yes, it's easier for mathematicians, because a lot of background knowledge and intuition is encoded in mathematical conventions (eg "C(R)" for continuous functions on the reals etc...). Note that this is probably a book for mathematicians.

By pure IQ: The smartest people I know very likely work in pure maths at universities.

Unfortunately the stock market is not (reflective of) the economy? It seems reasonable to force them to diversify nonetheless.

The two graphs have visibly quite different volatilities. A number like 20% is not meaningful unless you also somehow state the risk, eg in terms of volatility or beta or other measures.

Yeah, of course, if the name is clearly absurd you’d use the insights from that and update your posterior, and if you have more information about the person you can include it in the prior.

The point was that 1/n over all possible names is far from a reasonable prior, and that the Bayesian update barely depends on the name itself (within reason), but on other information gained during the introduction.

This is a nice insight. I slightly disagree with the first example in a gambling context, although it makes sense in the context of information theory. I don’t think the choice of prior is right when giving 20:1 posterior odds you’d be willing to bet on. A better prior should be related to the probability of a random person lying about their name when introducing themselves. And the posterior then doesn’t really depend a lot on what name he says.

Thanks, this is a nice reference! Interesting that the abstract and introduction mention the lack of existing evidence - seems like an under-studied question?

One aspect that's quite easy to criticise about this study is that it uses existing groups of students with different levels of maths training. This means that there is possibly self-selection etc, and one may argue there might also be a causal effect in the opposite direction (e.g. folks that are good at reasoning like to do maths.)

I've encountered this argument before, but also applied to other subject such as eg Latin:

> 3) math has an indirect usefulness which is a way of thinking that is transferable to a myriad of disciplines and solutions to problems in everyday life. And it is this third reason that makes math relevant for most people.

Are there actually any good studies that show this in a counterfactual setting? Like, do we actually know that spending time teaching maths (or, less plausibly, Latin) helps students aquire these abstract skills more than other subjects? Is this "transferability" of meta-skills a testable outcome?

What if the museum were to put up a sign that said "This is not art"?

They probably found many small things and combined them extremely well. In quant trading, if you've got a lot of excellent researchers, chances are that you make better choices in lots of "micro" decisions that add up to a superior outcome.

Yes, like, all uniquely shaped ones?

I think this is a relatively weak argument, because in colloquial English “the shape” sort of implies uniqueness in the question?

Ah, yes, and they do indeed cite Barnes-Hut, too. So the (claimed) novelty seems to be how they connect it to Monte Carlo steps.

Arxiv: https://arxiv.org/abs/2207.14670

If I understand correctly after skimming, one of the fundamental ideas behind this appears to be similar to the well-known Fast Multiple Method [1]. It’s also a tree-based approach where far away points are aggregated into larger chunks?

[1] https://en.m.wikipedia.org/wiki/Fast_multipole_method

This is clearly meant to be humorous etc, but just to explain a part of what’s happening here: Regressing price levels (integrated series) against each other does not lead to meaningful R^2 values and is considered a “chart crime”. See [1] for a detailed model.

[1] https://www.stat.rice.edu/~pe6/AOS1509.pdf

Can confirm empirically that using a VPN from a European country is not sufficient to be blocked.