4 ms·

If they're "outliers", their success can be explained by chance. With many people tossing coins, some will do very well by the laws of probability. The other p

If they're "outliers", their success can be explained by chance. With many people tossing coins, some will do very well by the laws of probability. The other problem is you're measuring success in hindsight, not from the beginning. (You could restate it as if from the beginning if you had the full list of initial hedge funds - including those that no longer exist). The point is that you can't predict which funds will be successful. Buffett beats the market - but he does it by understanding the businesses behind their stocks, their value and prospects, compared with their present price. He buys value for money, and generally holds it for the long term. He isn't an active trader, which to him is speculation.

dsacco 9y ago
> If they're "outliers", their success can be explained by chance. With many people tossing coins, some will do very well by the laws of probability. No, they cannot be. I meant outlier in the sense of a probability distribution with e.g. standard deviations, not to imply randomness. I’ll start by quoting what I wrote the last time this topic came up on HN: “Virtu only lost money trading one day out of 1278 trading days between 2009 and 2014. In the most uncharitable analysis (1278/2; or the lost day happened in the middle), they had a 0.5^639 chance of doing that. Maybe you disagree with 0.5 per day. Let's make it 0.9! ...But that's still 5.7 x 10^-30. How many firms do we need to exist for this to emerge by chance?” Draw up a probability space for me and actually quantify what you’re asserting. What is an event? Is it a single trade? A trading day? A trading year? How many of these events are there, total, in that timespan? Of those, how many win? Can you map each of these events to a boolean function, such as a coin toss, or are the chances more nuanced? This is before we even get to the issue of quantifying risk as a metric. How about we make the event a trading year; how many funds would need to exist to explain hedge funds like RenTech or Baupost? Moreover, can you explain three decades of extraordinary returns in a single fund as an outcome consistent with a uniform distribution across all eligible hedge funds? This armchair probability analysis turns up every so often on HN, but to be blunt, the onus is on the analysis to actually make it empirical. You can’t just say “well, they’re doing this by chance, we can expect some of these to eventually emerge because statistics and coin tosses” without formally proving that there are actually sufficiently many hedge funds and criteria to satisfy that claim. > Buffett beats the market - but he does it by understanding the businesses behind their stocks, their value and prospects, compared with their present price. He buys value for money, and generally holds it for the long term. He isn't an active trader, which to him is speculation. This is false. Buffett is, by his own admission, an active investor. He picks stocks - what is your definition of an “active investor” if not someone who makes active decisions about how to invest? His letter, “The Superinvestors of Graham and Doddsville”, specifically outlines his investing and describes it as an active strategy.[1] In fact, the way you’ve described his work is an investment thesis consistent with modern trading strategies. At their core, statistical arbitrage and high frequency methods seek to identify price inefficiencies just as Buffett does, but on much smaller timescales. It’s a difference of degree, not category. ______________________ 1. https://www8.gsb.columbia.edu/rtfiles/cbs/hermes/Buffett1984.pdf https://www8.gsb.columbia.edu/rtfiles/cbs/hermes/Buffett1984...
- hyperpallium 9y ago
  You misquoted me with "active investor"; I said he wasn't an "active trader". As I reinforced by saying he was long term, he is an investor rather than a trader. Select your set of traders today. Make predictions on how they will perform. Compare with what happened. Of course we should expect them to be better than pure chance. This article itself is about insider trading, which should give an edge. And yet, the hedge fund average is indeed worse than chance investment in the market (which approaches the index). The point is: we don't know who will do well. We only know in hindsight. Hedge funds also have management fees (higher than index funds), and pay more in brokerage for their active trading. > statistical arbitrage and high frequency methods seek to identify price inefficiencies just as Buffett does, but on much smaller timescales. It’s a difference of degree, not category. A difference is the sustainability of the inefficiencies identified. Perhaps if Buffett was purely an old-school value investor, picking up discarded cigars for a few puffs, there's an argument for similarity. But (as I carefully stated in my comment), he also looks at prospects, not just value. A business with a long-term sustainable economic moat makes money at a compounding rate. Now, I think we could have a largely semantic argument about whether buying a business with those prospects is just identifying price inefficiencies. To be honest, I can't really forsee how that argument would turn out, because of course you can view it as mispricing. Perhaps I can submit: more of the work is in evaluating the prospects, than the price (maybe you can argue the same for trading); simply because of the time-scale of long-term investing, you get to keep it longer, so it's more stable (but you can argue that that's covered by difference in timescales). PS. Also, I do wonder that Buffett also has better information. His dad was a senator, and he knows lots of influential people. I don't think it explains away his success, but it would be some help.
mrchicity 9y ago
You're both right. There are traders who beat the market in a way that's clearly statistically significant and doesn't involve selling tail risk or other cheats. However, they can stop being successful at any time. To give a concrete example, GETCO circa 2009 was the premiere HFT firm trading 19% of US equities market volume and earned high 9 figures in trading profits on a tiny capital base. They were doing millions of little trades daily and made profits almost every day, if not every day. The odds of someone doing that due to chance are infinitesimally small. They had been doing this since the early 2000s, quietly building an empire, and the people there were incredibly bright. They had their pick of the litter from every top school and one of the toughest interview processes. Retention was impressive, too, as the firm made many employees into millionaires in their early 20s and gave them plenty of autonomy. You'd look at this operation and expect it'd never end. But it did, and quite quickly. Competitors eventually learned what they were doing and saturated the market. Their market share plummeted, as did profits, while the cost base grew. High speed data and microwave lines became a new cost of doing business. Layoffs happened. Once loyal employees defected and with them went even more of the firm's ideas. Eventually they got a soft landing by buying out KCG, and the entire operation was later acquired by Virtu, but from what I understand, few of the original GETCO trades still exist or make any profits. So I believe they won due to skill at the time, not luck, but more skillful competition came along to replace them. Some small group will always beat the market, but I'd need more evidence to believe one group would do it for a long time. Brain power and guarding IP are necessary but not sufficient.

dsacco 9y ago

> If they're "outliers", their success can be explained by chance. With many people tossing coins, some will do very well by the laws of probability. No, they cannot be. I meant outlier in the sense of a probability distribution with e.g. standard deviations, not to imply randomness. I’ll start by quoting what I wrote the last time this topic came up on HN: “Virtu only lost money trading one day out of 1278 trading days between 2009 and 2014. In the most uncharitable analysis (1278/2; or the lost day happened in the middle), they had a 0.5^639 chance of doing that. Maybe you disagree with 0.5 per day. Let's make it 0.9! ...But that's still 5.7 x 10^-30. How many firms do we need to exist for this to emerge by chance?” Draw up a probability space for me and actually quantify what you’re asserting. What is an event? Is it a single trade? A trading day? A trading year? How many of these events are there, total, in that timespan? Of those, how many win? Can you map each of these events to a boolean function, such as a coin toss, or are the chances more nuanced? This is before we even get to the issue of quantifying risk as a metric. How about we make the event a trading year; how many funds would need to exist to explain hedge funds like RenTech or Baupost? Moreover, can you explain three decades of extraordinary returns in a single fund as an outcome consistent with a uniform distribution across all eligible hedge funds? This armchair probability analysis turns up every so often on HN, but to be blunt, the onus is on the analysis to actually make it empirical. You can’t just say “well, they’re doing this by chance, we can expect some of these to eventually emerge because statistics and coin tosses” without formally proving that there are actually sufficiently many hedge funds and criteria to satisfy that claim. > Buffett beats the market - but he does it by understanding the businesses behind their stocks, their value and prospects, compared with their present price. He buys value for money, and generally holds it for the long term. He isn't an active trader, which to him is speculation. This is false. Buffett is, by his own admission, an active investor. He picks stocks - what is your definition of an “active investor” if not someone who makes active decisions about how to invest? His letter, “The Superinvestors of Graham and Doddsville”, specifically outlines his investing and describes it as an active strategy.[1] In fact, the way you’ve described his work is an investment thesis consistent with modern trading strategies. At their core, statistical arbitrage and high frequency methods seek to identify price inefficiencies just as Buffett does, but on much smaller timescales. It’s a difference of degree, not category. ______________________ 1. https://www8.gsb.columbia.edu/rtfiles/cbs/hermes/Buffett1984.pdf https://www8.gsb.columbia.edu/rtfiles/cbs/hermes/Buffett1984...

hyperpallium 9y ago
You misquoted me with "active investor"; I said he wasn't an "active trader". As I reinforced by saying he was long term, he is an investor rather than a trader. Select your set of traders today. Make predictions on how they will perform. Compare with what happened. Of course we should expect them to be better than pure chance. This article itself is about insider trading, which should give an edge. And yet, the hedge fund average is indeed worse than chance investment in the market (which approaches the index). The point is: we don't know who will do well. We only know in hindsight. Hedge funds also have management fees (higher than index funds), and pay more in brokerage for their active trading. > statistical arbitrage and high frequency methods seek to identify price inefficiencies just as Buffett does, but on much smaller timescales. It’s a difference of degree, not category. A difference is the sustainability of the inefficiencies identified. Perhaps if Buffett was purely an old-school value investor, picking up discarded cigars for a few puffs, there's an argument for similarity. But (as I carefully stated in my comment), he also looks at prospects, not just value. A business with a long-term sustainable economic moat makes money at a compounding rate. Now, I think we could have a largely semantic argument about whether buying a business with those prospects is just identifying price inefficiencies. To be honest, I can't really forsee how that argument would turn out, because of course you can view it as mispricing. Perhaps I can submit: more of the work is in evaluating the prospects, than the price (maybe you can argue the same for trading); simply because of the time-scale of long-term investing, you get to keep it longer, so it's more stable (but you can argue that that's covered by difference in timescales). PS. Also, I do wonder that Buffett also has better information. His dad was a senator, and he knows lots of influential people. I don't think it explains away his success, but it would be some help.

mrchicity 9y ago

You're both right. There are traders who beat the market in a way that's clearly statistically significant and doesn't involve selling tail risk or other cheats. However, they can stop being successful at any time. To give a concrete example, GETCO circa 2009 was the premiere HFT firm trading 19% of US equities market volume and earned high 9 figures in trading profits on a tiny capital base. They were doing millions of little trades daily and made profits almost every day, if not every day. The odds of someone doing that due to chance are infinitesimally small. They had been doing this since the early 2000s, quietly building an empire, and the people there were incredibly bright. They had their pick of the litter from every top school and one of the toughest interview processes. Retention was impressive, too, as the firm made many employees into millionaires in their early 20s and gave them plenty of autonomy. You'd look at this operation and expect it'd never end. But it did, and quite quickly. Competitors eventually learned what they were doing and saturated the market. Their market share plummeted, as did profits, while the cost base grew. High speed data and microwave lines became a new cost of doing business. Layoffs happened. Once loyal employees defected and with them went even more of the firm's ideas. Eventually they got a soft landing by buying out KCG, and the entire operation was later acquired by Virtu, but from what I understand, few of the original GETCO trades still exist or make any profits. So I believe they won due to skill at the time, not luck, but more skillful competition came along to replace them. Some small group will always beat the market, but I'd need more evidence to believe one group would do it for a long time. Brain power and guarding IP are necessary but not sufficient.