Hacker News Reader

Comment by filoleg | Hacker News Reader

His whole RenTech story was fascinating.

Effectively an outsider in finance who gathered a bunch of other outsiders (aka big mathematicians), and decided to start a hedge fund that takes zero interest in the actual companies and trades solely on math. Which makes sense, since none of the main people involved in its creation had any corporate or finance experience, but tons of math experience and knowledge.

This is oversimplifying it like crazy, but I recommend anyone to anyone with even a passing curiosity for this look up the details (or read “The Man Who Solved The Market”, which is documenting the beginnings and growth of RenTech, as well as that of Simons; very enjoyable read).

View on HN

“The Man Who Solved The Market” is fascinating because it spans almost the entire history quantitative finance (through the lens of RenTech) dating back to the 1970s.

Simmons was one of the first to realize the advantage of collecting and analyzing vast sums of data to identify patterns in financial markets. They were digitizing magnetic tapes and collecting more data than they could even process given technical limitations of the time.

I can second this book recommendation. Remarkable candor in such a secretive field.

Yep, not argument here at all, RenTech is a super fascinating outlier in finance.

It's kind of inspiring. I don't know a ton about finance or trading algorithms, but I know a fair bit (I think) about math and CS, and because of RenTech I've formulating my own trading strategies (just paper trades). Thus far all I've been able to do is lose all my pretend money by trying to play options, but it's still fun to try.

Will I be successful and make billions? Almost certainly not, but it's an excuse to play with different types of math that I don't play with very often, but RenTech proved that you can beat the market by taking advantage mathematics.

There are some interesting interviews around rentech. It starts to feel like they made a lot of money out of being extremely thorough, by doing a lot of reasonably simple (at least by the standards of math phds) things extremely well.

IIRC his fund averaged around 30% gains per year, every year, over 30 years. (I'm going from memory here, too lazy to look it up). That is just such an unbelievable performance number.

Aren't there some shenanigans with those numbers around their larger funds not doing as well?

It's easy to make a few high margin dollars, hard to make a lot of high margin dollars.

They limited the fund size so employees frequently got distributions from the fund instead of just rolling over their investments. However, the distributions were still in the millions of dollars.

They also got into some tax trouble with uncle sam and had to pay 7b in back taxes (https://www.wsj.com/articles/james-simons-robert-mercer-othe...)

depends on your definition of "few". Rentech made a "few" for very large "few".

lol . Only made tens of billion. What a failure.

I'd still wish to have details on this (I too heard of similar numbers for his fund before), because in my newb eyes .. such returns would mean they could absorb a huge chunk of the planet liquidity.

No, because such returns aren't scalable.

According to industry rumors, RenTech is somewhere between $10-20bn AUM (assets under management, i.e. the capital used for trading), and the profit that they make, they can't reinvest, they have to take it out as profit.

How come? Why do they have to take the profits out and can’t compound it?

I know literally zero about this stuff!

When you scale up too much it creates market impact that affects returns. You basically become too much of the market.

The simplistic explanation is, if you're doing arbitrage - i.e. "fixing market mispricing", there's only so much arbitrage you can do before you fix the price...

This is of course a completely theoretical proposition, because in reality you don't know what the "fair price" is. You don't even have probabilities, because those are also unobservable, you only see one version of "history".

In practice, what happens is that if you trade "too much", "shit goes wrong". Both of these things require empirical estimation and are easy to get wrong.

The most obvious is the market liquidity, which you can observe at e.g. BitStamp TradeView [1] - there's only so many orders at a given price, so the more you trade, the worse price you get (the average/marginal trade).

No professional of course trades like that, especially not HFTs, but similar problems happen at every scale - you're competing with other traders, they might have better information, there's limited amount of stock in the market, the edge/alpha/expected profit you can earn decays over time as the price moves, if you trade too much you move the market and inform other participants who can then trade against you, ...

[1] https://www.bitstamp.net/market/tradeview/

I know literally zero about this stuff!

I guess you will not be getting a job there

But in seriousness, when you become so big relative to the market, you become the market.

All investment strategies are limited as to the amount of money they will take. They probably couldn't have ran much more without reducing perfomance.

You'd have to sacrifice the returns if you want bigger size. For every trade you do, there will be expected return (+ve) and then some costs to pay (-ve). Commissions and similar costs are only linear so not terrible. With increasing size, the market impact cost that's non-linear will soon overwhelm all other costs. So you keep adding alpha in your forecasts (via your research pipeline), that will be eaten away by the impact cost, as you scale up. If you keep it small (-ish - still gross pfolio will be billions) - then you will get to keep high returns.

How do you 'absorb' liquidity? Are they like some kind of financial kitchen towel?

it was 62% per year for 33 years.

That is insane. Like, completely insane, shouldn't-be-possible insane.

I guess the theoretical limit to how much money you could make in the market is "the sum of all volatility", but I wonder how realistically possible it would be to even dream of beating 62% yearly.

Mathematics can only take you so far. At the end of the day, people run the exchanges. Not math.

The returns of modern HFT market makers are even higher. With their unfair “business” advantages such as PFOF, privileged dark pool and block trade access, and military internet infrastructure.

Think 60%+, per year, at least. Over 10-20 years, of course.

That doesn't surprise me; doesn't Citadel keep the entire bid-ask spread for every transaction they facilitate? Presumably between that and arbitrage opportunities that pop up from option contracts alone, I have no doubt that market makers clean up pretty well.

They wouldn't hire me either!

Citadel has plenty of competition, eg from Jane Street. The markets for market making are some of the most efficient markets on the planet.

> The returns of modern HFT market makers are even higher.

The returns of a child's lemonade stand are even higher...

Market makers and lemonade stands are mostly about paying for labour (and ideas etc, but let's call that 'labour', too). Capital requirements are rather low. So taking all the profit and attributing it to capital returns tends to give you weird numbers.

> So taking all the profit and attributing it to capital returns tends to give you weird numbers.

Why does it matter? Returns are returns. Money in, money out.

After all, people compare HYSA bank interest with TreasuryDirect bond returns with equity ETFs like VTI and QQQ. Each with vastly different capital mechanics.

Yes, but there any old schmuck can put some dollars in and get the same return.

Good luck trying that with one of those very profitable market makers and funds: they don't want your capital; or at least they don't want it at the same price (= returns) that we are quoting here. Which suggests that those returns aren't attributable to that capital at all (even though for tax reasons they might structured it so that legally these are counted as capital returns, but that just obscures the underlying economic reality).

This is very similar to observing that a particular company pays a lot of money for some very simple job; but then we notice that the job is only available for the son of the CEO. We can conclude that the extra pay isn't really for that simple job.

Or when we notice that a government contractor officially charges 5000 dollars for a hammer. Unless you and me could rock up and steal market share by offering to sell hammers for 4000 dollars, it's very likely that the 5000 dollars aren't really for the hammer at all; but just some accounting shenanigans.

Their returns worked out to something like an average of 39% per year after fees, which is the figure I've heard cited. This may be what they were thinking of. Renaissance was/is known for having higher fees than likely the entirety of their competition, which they can get away with since their returns still outstrip the rest after the higher fees.

The fund is closed off to outsiders, so the fees are don't matter in the same way they do for most funds. In the podcast episode on Rentec done by Acquired, the hosts speculated that rentec kept the high fees as a way to ensure they have enough to handsomely pay less tenured employees who don't yet have much money in the fund.

That’s a great episode and covers a lot in depth. Would recommend Acquired in general - appeals to me as interested in tech, business and finance

I'd heard that the Medallion fund was closed off, so I wasn't really sure of the reasoning behind that continuing fee structure, but that line of speculation does make some sense.

Important to keep in mind that these returns ceased to be compounding quickly: they restarted from scratch 10bn each year to score 30%.

Successful quant stratégies tend to hit capacity limits...

Their success is limited by what other party ready to lose, most of the time, these all are zero sum games.

Depends on how you look at it.

Eg selling insurance can be seen as a zero sum game, but it's a genuinely useful product for people, even when the expected value for them is negative. It works, because utility is not strictly proportional to money.

Similarly, market making delivers liquidity-on-demand for a fee.

Insurance is positive-sum because the value-generating enterprise (the buyer) gets to continue generating value after the unexpected thing happen. The alternatives is that the value creation process just stop. It is only seemingly zero-sum for the point in time when the accident happens and one side has to pay for the other.

Your argument only works for catastrophic insurance.

In practice, people take out insurance even for events that would not put them out of business.

Btw, if you are talking about 'value-generating enterprises', ie businesses as buyers of insurance, then your argument doesn't really work either, or at least not without caveats:

When a business suddenly has a large liability, and it goes bankrupt, all that happens is that the equity owners are wiped out and the creditors take over. The underlying business can and often does continue uninterrupted, and has approximately the same value as a going concern as before.

Also, being able to run as a going concern is of finite value to a business. If your business can take a 51% chance of either doubling in value or alternatively going bust, then that _might_ be a good gamble to take if your shareholders are well-diversified. For example, if index funds are your main shareholders.

Humans need considerable better odds before they consider such a gamble. But people do regularly put their life on the line in return for very finite benefits. Eg every time you leave the house, and drive a car. Or even more stark, any time people conveniently 'forget' to put their seatbelts on.

Events that merely reduce the productivity of your business has the same calculus: insurance helps you get back to speed quickly, and there are values in doing so.

I am not saying that ALL insurances provide values. Like any other kind of trades, you can lose values if you make a bad decision. That does not make insurance inherently zero- or negative-sum.

> When a business suddenly has a large liability, and it goes bankrupt, all that happens is that the equity owners are wiped out and the creditors take over. The underlying business can and often does continue uninterrupted, and has approximately the same value as a going concern as before.

This assumes the original owner brings no value to the business. Even in that case, the disruption itself is harmful, not to mention the assumption that bankrupted business can restructure rather than closing down.

You can close down the business even without a bankruptcy. And you can have a bankruptcy, without closing down.

Closing down the business doesn't necessarily mean very much: all the machines, and workers and real estate and building still exist, whether the business closes or not.

their success is limited by how much money they can move. When you are moving that kind of money through quant strats you start to move the market. It's easy to capture a triangle arb with 20k, almost impossible to do it with 10B, because by the time you enter and exit the trade the arb no longer exists or you were moving the market against yourself with your own trades.

One of the genious thing that rentech did was long out of the money bonds, and short newly issued bonds. Seems like such a simple strat, but when you crank up the leverage you can make alot of money.

When I read "The Man Who Solved The Market”, I blown away with the story of Robert Mercer who arguably paved the way for Brexit and the election of Donald Trump. I wonder how different the world would be if Simmons didn't exist, the butterfly effect can sometimes have some massive unintended consequences.

Does the book cover anything about how the fund actually works? It's understandable that they'd want to keep a tight lid on it, but I'm so curious

no. One can infer it involves analyzing data, but for obvious reasons the ingredients for possibly reconstructing it are omitted.

The book covers some earlier aspects of the strategy. And I think the "spirit" of the strategies exists today, though tangibly very different and not actionable.

I believe they were doing ML based trading. Their edge was data collection, cleaning and standardisation and the ability to trade a lot at very cheap borrowing cost. This was way before computers became a thing in trading or ML became a thing.

> Does the book cover anything about how the fund actually works?

Basically, no.

Reply on news.ycombinator.com