마법의 투자 공식은 없다 - Tom Costello

https://www.quora.com/What-should-I-do-next-I-invented-a-trading-system-which-generates-a-profit-of-50-on-average-I-developed-it-and-tested-it-for-3-years-I-can-fund-only-10-000-so-it-will-end-up-very-slow-to-quit-my-job-What-are-your

Q:

What should I do next? I invented a trading system which generates a profit of 50% on average. I developed it and tested it for 3 years. I can fund only $10,000 so it will end up very slow to quit my job. What are your suggestions?

 

A:

I worked in quant finance for 25 years, the last portion of it as Head of Equity Trading at a 1B quant fund where a substantial portion of my time was spent interviewing and hiring Portfolio Managers. So if one of the things you’re thinking of doing is going to work as a PM at an established fund in order to increase your capital, let me tell you how the story you’re currently telling will be seen.

1. You aren’t really making 50%. (by that I mean 50% annually, which is the typical metric by which strategy returns are described.) Your ‘testing’ is back testing not paper trading, so it isn’t really ‘proven’. I assume this because the major listed markets are far too efficient for anyone to make 50% annually in a reliable and reproducible way. When you try to actually trade this strategy or to properly ‘test’ it with appropriate assumptions for execution costs, (or at the very least, to trade it in any volume) you’ll find that slippage eats most if not all of your profits.
2. If I’m wrong about that and you really have paper traded it with a reasonable estimate for costs and slippage, the strategy won’t scale large enough to make it worth it for a fund to participate in. Trading in size means worrying about issues that most people don’t understand. (How do you think all those execution traders have earned their money of the last 2 decades?) There are some strategies which will deliver a relatively high return at a VERY small size, but those don’t make sense for a large fund because of other fixed management costs. As a rule, they won’t make sense for you to do on your own either.
3. If by some miracle I’m wrong about ALL of that, then your risk is too high and I’ll see that from a quick glance at your cumulative return chart. Making 50% is easy if you are sometimes down 150%. But to properly scale a strategy you’ll need to show a cumulative return history that never exceeds a 20% drawdown. That 20% is as close as the hedge fund world gets to a universal rule. The other close to universal rule is that you’ll need a sharpe ratio above 1. Of the hundreds of guys I’ve interviewed for PM jobs, all of whom had several years of professional experience working in research, I saw only 4 strategies that met both those requirements. One of them lost its entire P&L to slippage, and one other wouldn’t scale enough when we tried to run it.

I’m not trying to discourage you, but these questions are very common on Quora and they all ignore the facts of the financial markets. They’re all version of “I’ve discovered the magic formula trading program that will make me a billionaire if I can just raise the money. What do I do?” But they’re all wrong. I tend to generously think that they’re just kids trying to understand the processes in the industry and not trolls or people looking to ‘game’ it. But whether their motives for the questions are innocence and ignorance, or malice, I and the entire professional end of the industry already know that the magic formula they’re looking for (or pretending they’ve found) doesn’t actually exist.

The truth is that in large hedge funds, decision making is very strictly constrained by risk limitations. You can’t just do ‘whatever you want’. In fact, the rules for hedge funds are actually defined by the typical investors in hedge funds, and their rules are determined by the cost of capital, and the best risk management thinking available.

Within those constraints, there is NO magic bullet. There is no secret formula that will give you a 50% return because there isn’t a enough ‘information content’ in the combination of price, volume, volatility and the statistics of derived markets that hasn’t already been studied to death by tens of thousands of the sharpest minds on the planet. If it were really there, someone (probably several people) would have found it, and they’d be doing right now - and in the process eliminating the profit opportunity for everyone else.

From the outside it looks like easy money. Analyze stock prices, make a prediction, and poof… you’re driving a lambo. It doesn’t really work like that.

In reality you’re considering getting into the single most intellectually competitive field on the planet. A field where people have been systematically doing for decades what you’re hoping to do by luck. Yes you may be the smartest person you know. But so is EVERYONE else in quant finance. And they’ve been at it for a good long while.

They’ve looked at high frequency, low frequency and everything in between. They’ve done market neutral, long only, short only, long Vol, short Vol, periodic regime change, and all the possible variants you could ever imagine while looking at price, market spreads, market splits, market gaps, volume and derived price and volume. They’ve done mean reversion, mean divergence, and mean stability on every one of those combinations. They’ve tested equity to futures. Futures to options. Equity to options, options to futures and all of the above against all the ETF’s and indexes. They’ve compared all of those to every conceivable predictable variant of yield curve trajectory across every major currency and bond market.

They've compared one market to the next in every conceivable possible way. And in the process they’ve teased out every fraction of penny of predictable inefficiency in every major market in the entire world. The odds that you have actually done better than all of them on one of your very first tries, is infinitesimally small.

Collectively, professional quant traders can be thought of as a single huge machine that tests every new rational potential theory of market relationships in a lightning fast way, looking in every conceivable corner for enough predictive certainty to turn a profit, for as long as they can. (They test all the irrational ones too… but those never leave the drawing board.) Those that win today can play again tomorrow and those that don’t have to go find something else to do. To call them highly motivated is a vast understatement.

With all that discouragement stated, let me give you the flip side.

There are other ways to approach this that still work, and can still make you a living. Try looking at “Alternative Data” or getting into a less efficient market like crypto. Take a look at trading illiquid assets because the accounting rules are more generous than daily ‘mark to market’. That’s what the leaders in the space are all doing now.

If you’re looking in the right place, it might be possible to come up with a model that will give a big return by ‘luck’. It might not work for very long, but maybe by then you’ll learn enough about the market to come up with something else. But if you think you’re going to write the magic formula in the major equity, bond, currency, future, option markets, that delivers a huge return with limited risk, think again. I don’t mean to be too harsh about it but you’re wasting your time.

I’d like to add one thing to this answer that has occurred to me since. One of the biggest challenges in designing a real ‘quant’ strategy is a low correlation. Your returns need to be uncorrelated (below .5) or ideally anti-correlated (below 0) to the market you’re trading, and the overall return of the fund you’re talking to. If your correlation to the S&P is 0.9, then I don’t need to pay you to obtain that risk, I just buy the S&P. I can pay someone 50K a year to do that, and to manage P&L and settlement at the same time. I can have a tech guy do it in between writing code.

This may not sound like much, but you’d be shocked at how difficult it is. It basically means that you have to have a truly unique idea. A new approach to understanding the relationships between markets and price behavior. And when all the competition is taken from the top few percentage points of the top schools on the planet, run through the best industry training programs in the world, and set loose in direct competition with each other for the better part of 2 decades where only the very best are allowed to continue in that role, it’s no easy trick.

It can be done. It’s done by a tiny number of people every year. But the odds that you are that person are very small. The odds that you are not only that person, but that you’ve found 50% annual return in the process by using data you downloaded from google finance, are non-zero.