Best “unknown” quant funds/shops

[kuebiko]

@wpdupjuj care to name your advisor?

[ronin]

@kuebiko,



I agree. Once the signals start decaying, they are basically random. And if you don't really know why they worked in the first place, that's it.



> at low frequencies building a nontrivial allocation engine is subtle.



Low frequencies is where I live these days. Allocation is allocation - it's a case of garbage in, garbage out. When you find yourself relying on your allocation engine, your time is up.

[EspressoLover]

@u1234 and @wpdupjuj



Those are all fair points. I've updated my priors.



> My advisor has sent many students to Setauket. All of them must be using techniques very close to what we picked up in our research



I'd caution against leaning too heavily on this. RennTech is a huge operation, and its current systems are the cumulative result of more than a thousand researchers. A single person's experience, or even a group of people's experience is likely to fall short of the whole picture. It's a little like the fable of the blind men each touching a different part of the elephant.



I certainly believe that Medallion has some strategy that utilizes whatever this technique is. However it'd be difficult to say how substantial a proportion of the fund's profits are attributable to this technique. And I'd think it'd be outlandish to say that the answer is all or even most of them. Medallion does a whole bunch of things very well, and it's been doing so across a huge span of time in terms of market regimes and structure. The secret sauce definitely has more than one ingredient.

[u1234]

@EspressoLover

>> But I'm fairly confident that a substantial proportion of Medallion's profits come from intraday signals and strategies.



You definitely are more familiar with this than I am.



Had a question to ask you. Some time ago I looked at their 13F data and they seemed to have a way more holdings than their RIEF fund (back then no RIDA or RIDGE) and it looked like they had meaningful overnight positions for sure.



how do you define intraday signals and strategies? Do you mean: (i) strategies that close out by the end of the day, or (ii) strategies that have an average holding period less than a day, or (iii) strategies that are more likely to buy / sell intra-day (rather than a dumb VWAP per se) on limits, stops or other levels for trades like mean-reversion of something (i.e. intra-day move aware), or (iv) some other definition ?



Would you have any idea what percentage of their profits come from intraday strategies?

[Strange]

"most of your returns are driven by blue chip stocks, liquid futures, and major FX pairs"



My guess is that Sharpe being a rather misleading number here. For example, it's totally normal to generate Sharpe of 2-3 in fixed income by doing things like OTR or futures basis, which have pretty low turnover. And we are talking serious size. I think that once you're running an integrated firm, are also all kinds of strange synergies between lower and higher turnover strategies, where your higher turnover benefits from volatility and, in effect, hedges your lower turnover book. Does not mean that's what RenTech is doing, but it's a possibility.

[loltrading]

>> It's not my personal principle to rate people that I've worked with negatively, however I can confirm that that was not the reason for his firing. He had barely started on any strategy code at the time he was fired, and in fact had asked for a 6 month extension the day before. Happy to give more color if it's important to you. [...]



"It's not my personal principle to rate... [proceeds to do just that]."



I already caught a whiff of bull from your firm's overly aggressive PR blitz (OMG we traded a BILLION dollars--of what?), but this cements my opinion. Why so defensive, vindictive and arrogant? People who get fired sometimes write mean stuff online. Maybe they're right and maybe they're wrong. Stooping to their level only makes you look bad.



Your own posting for researchers seeks "Experience with direct responsibility in construction of alpha signals or monetization for latency-sensitive, capacity-constrained strategies" and you boast in the press about hiring away a couple ex-Virtu and GETCO employees. So you want to hire mercenaries like the NYC shops you obliquely reference, but probably don't even give a good payout structure, and then expect them to dump their code into a shared repo? lol, just lol.



Good non-siloed firms develop their own IP and train up their own employees.

[loltrading]

.

[anonq]

As far as the quantbot manager platform the idea is for an individual or very small group to plug and play and not have to deal with all the hassles of running your own group elsewhere. They do have a larger central group. It's a good set up if you just want to do research and get a cut of pnl. They are funded by Schonfeld which also has some other every successful groups under the umbrella.



A lot of places stay out of the radar mainly cause they already have access to all the capital they need and likely capacity constrained anyways due to market impact and need more alpha to increase size which is a slow process, that's my situation and I run an equity quant book. My firm only has a website cause the SEC told us we had to have one.



The returns from the funds trading large size overnight are really only impressive cause they're levered. On an unlevered basis not terribly impressive other than the sharpe which allows for the leverage.



The way I look at it is that there are say a billion dependent variables that explain the market. All of us are trading a slice of those with some of it overlapping. It's actually rather surprising how low the correlation between different quant funds can be.

[kuebiko]

@anonq, so what would you say distinguishes quantbot from the other multi-manager platforms like, say, Cubist? Are you saying that teams have to be less "entrepreneurial" at quantbot, or that they can somehow take advantage of more centralized resources? Is it not more limiting than bigger multi-manager funds?

[HitmanH]

My view - is Cubist is ultimately (/will be) focused on running client assets - and have no real short term infra. So slower, scalable strats fit better