Back to Community
Quantopian's First Discretionary Capital Allocations

Our community reached a watershed moment in Q4 of 2015: the first community authored algorithms were chosen and deployed with capital. We analyzed tens of thousands of algorithms, and then we made six-figure allocations to 3 of them. We are writing to the whole community to say thank you — this is something we accomplished together. We also want to share some details, and to encourage and inspire you to keep writing great algorithms.

Allocation and Compensation

The allocation process worked out just as we had hoped that it would. We approached each of the chosen authors with a contract. The terms of the contract promise that a portion of the returns be paid to the author as compensation. The contract also covers the other things you'd expect - length of the contract, maintaining the intellectual property as the author's, etc.

We then did some additional diligence with the authors and worked with them to make the algos more robust for live trading. For instance, one of our author's algorithms had some brittle rebalancing logic that failed during stress testing, and we asked him to correct that before we deployed it with real money. Once the contract was signed and diligence was complete, we deployed the algorithm using our capital.

Algorithm authors were paid for their 2015 performance, and we look forward to writing checks again.

These allocations are in addition to, and separate from, the real-money trading we have done for Quantopian Open winners. These new allocations were made on the basis of analysis by our quant research team, using tools like pyfolio tearsheets, and not the contest rules.

More Allocations Coming

We are making additional allocations this quarter and going forward. We expect our allocations to increase in size over time.

All algorithms with at least 6 months of out-of-sample data are being considered. We evaluate the algorithms by looking at their in-sample and out-of-sample performance; we never look at the code itself. As we have outlined, we are looking for well-hedged, low-beta, low-volatility algorithms. Obviously, they also have to be profitable. Finally, they must have low correlation with other algorithms we have allocated to.

Many algorithms we have considered were promising but fell short in one vital area or another. We've been contacting these community members and have been working with them on improving their algorithms. After additional out-of-sample time passes we hope to provide them with allocations.


These allocations are all being made from Quantopian's own balance sheet - we believe in this model enough to risk our own capital. This is proprietary trading, or "prop trading." We are using prop trading to refine our algorithm selection process, our risk management, and our algorithm portfolio construction. We are very pleased with the progress in all of these areas.

Our business growth depends on making much more capital available to your algorithms. More capital also means you can earn larger royalties by licensing to Quantopian. We are working to ramp both the number and the size of capital allocations to your algorithms by transitioning from prop trading to a hedge fund.

Keep Writing Algorithms!

These allocations are the latest step in the progress of our community and in our business, but it is far from the last step. The platform is improving and expanding. There is much more capital still to allocate.

Most importantly, there are more algorithms for you, our community, to write. These allocations are how we can convert your hard work and insight into your profit. The great algorithms you're writing today are the allocations we will be making tomorrow.

Note: Past performance is not an indicator of future returns. This note does not constitute an offer or solicitation to invest in any securities.


The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by Quantopian. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. No information contained herein should be regarded as a suggestion to engage in or refrain from any investment-related course of action as none of Quantopian nor any of its affiliates is undertaking to provide investment advice, act as an adviser to any plan or entity subject to the Employee Retirement Income Security Act of 1974, as amended, individual retirement account or individual retirement annuity, or give advice in a fiduciary capacity with respect to the materials presented herein. If you are an individual retirement or other investor, contact your financial advisor or other fiduciary unrelated to Quantopian about whether any given investment idea, strategy, product or service described herein may be appropriate for your circumstances. All investments involve risk, including loss of principal. Quantopian makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances.

46 responses

Congatulations! 59,997 allocations to go (assuming 1 per registered Quantopian user).

This is great news!

One question, given that the algo writers have knowledge of what's going to be traded before pretty much any of the trades are made, what safeguards are in place to prevent front-running orders that are generated for Q's fund?
Realistically speaking, unless the quant who wrote the algo is really wealthy, his/her front running probably won't impact execution quality of the fund's trades. However from a regulatory stand point, any indication of front running/insider trading/whatever you want to call it, will spell disaster for the fund.
I'm not sure what the contents of the contract are, and if makes the quant have a fiduciary responsibility (I'm guessing it doesn't since I think that requires licensing that most quants don't have). At any rate, my point is that there is nothing preventing the quant from using knowledge of the trades to be executed for his own personal gain.


@Jack: good point, but on the other hand if what he does negatively impact the algo, it will eventually be eliminated due to poor performance

@Anh: True, but odds are, performance won't be affected in any significant manner unless the quant has very large sums of capital at his disposal. My point is that, regardless of whether the quant earns money and "cheats" the fund, Quantopian still HAS to address the fact that the trades queued are no longer truly confidential.

Additionally, the quant can run an identical algo to trade a couple seconds before the one running in the fund does, skimming a few bps each time. You add that up on a daily basis and the cash gained is no longer trivial. Doing this won't affect the performance numbers of the fund either (maybe a fraction of a basis point but nothing serious to disqualify it)

Thank you all!

Jack, that's a good question, it's one of the things we've been thinking about. The contract we sign with the algorithm author prohibits front running like that. During the diligence process, we talk to the quant about that, and we take additional steps like running a background check on them. By the time we're done the quant is no longer "someone out there on the internet" and becomes "someone we know better and have a legal relationship with."

It is a risk we're going to have to be very attentive to, now and going forward.

@Jack, if it doesnt affect the algo, it doesnt affect Quantopian. Associated legal issues are the author's problem. So they can do it at their own risks :)

It would be quite interesting to see some backtest equity curves and numbers for selected algos, is there any plan to release some of the data so the community could see what kind of algos are selected? Ballpark sharpe, drawdown, alpha/beta etc.

I am surprised it took 8 hours for someone to ask this question =). We are going to share some information about the investment performance. We weren't ready to do it today, but I expect it in the near future.

As for the kind of algos selected, it really goes back to the points I highlighed above: well-hedged, low-beta, low volatility. One of the better examples of that is the long-short equity example. That algo itself isn't going to get into the fund, but it is the foundation to build a great one.

EDIT: fixed link.

Did any of the managers voluntarily share their code with Quantopian?

Also, I understand why, at this stage of the game, you'd check out managers and have them sign documents, etc. However, it doesn't seem like it'll scale. What if you want to bring on 100, or 1000, or 10,000 managers? I know it sounds like crazy talk, but that's the whole crowd-sourcing idea, to get to $10B, right? At $1 M each, you'd need 10,000 managers and you'll need a scalable system for on-/off-boarding.

@Anh are you serious? Of course it's quantopian'so problem. They're have a fiduciary responsibility to their clients!

I couldn't give a rat's ass about the quant but the firm is still liable and HAS to be concerned about front running REGARDLESS of performance.

Great release.

This is motivating.


Does Quantopian allows any investor to invest money say a Bridge between quants and investors or only investors with connection can use these algorithms?

It would be great if it was a bridge from any Quant to any Investor. Quants publish their algorithm and set a commission, investors without reading the code can back-test it play, check then decide to invest on it.

Congrats! Inspiring indeed.

@Lucas, cool idea - sort of how the crowd-sourced loan model currently works...

Wow, that's amazing news. Congratulations on that achievement!

Dan, are you able to share minor details regarding the three current contracts? Specifically, what is the payment arrangement so far assuming a profitable algorithm? Is it monthly paid such as the mechanism in the contest just before the $1M change, or more like let it run for 1 year and then let's see if there's any profit share to be paid out to the quant?

@Dan Dunn your "long-short equity" link fails for me FYI.

@Dan Dunn it would be great if we can see some comparative results on the strategies' performances.
Are there any plans to improve Quantopian's execution capabilities? Without an efficient way of executing large orders, it could reduce the profitability/scalability of the strategies. Current "market order" approach may not be optimal. Improving execution helps all strategies too.

All, thank you again. I'm particularly gratified to hear "motivating" and "inspiring." We want to create as much investment capacity as we can.

Anthony - link fixed, sorry!

Grant - Yes, there are things that need scale there. One way to scale is to increase the capital allocations. When we get to an average allocation of $20M, the number of algos needed for a $10B fund drops to 500. The other way to scale is through software. We've already built some tools where the quants who are doing diligence/deploying have special interfaces to work with us, and we have a matching backend system to manage the process. Software helps us scale yet again =)

Lucas - We gave that business model a lot of thought. I think of it as creating a two-sided market with investors on one side and quants on the other. One of the challenges there is scale. The investments need to get pretty big before the aggregated revenue is worth it. It also has huge regulatory challenges. Either the quant, or some intermediary, has to become a registered financial advisor, or the investors have to be accredited - and at that point you're looking at a hedge fund. In the really long run I think we can satisfy the interest that you (and others) have expressed, but it's not on the road map at this time.

Frederik - There are a couple of payment arrangements in play, and we're debating the model going forward. In one model we pay the quant annually. That's friendly to investors, and it has the benefit to the quant that their profits stay invested for longer. A competing model is to pay the quant quarterly, with a holdback that is paid annually. That model has the benefit of faster payments to the quant, and the negatives of resulting in an overall lower payment to the quant, plus it is less desirable for the investor. Monthly payments like the contest are not being contemplated.

Charles - I don't anticipate that we are going to share individual algorithm results. I do expect that we will share aggregate results. We completely agree we need to get better at order execution, too. We do have VWAPBestEffort available, but we don't have a way to backtest the effects of it. We need to improve the backtester, and we need to further improve order execution.

@Dan Dunn. it's a pity that we won't know the individual results. It would be fun to see how orthogonal the strategy performances are. and I am sure it will satisfy the sense of competitive curiosity in the few chosen ones. I think it would be hard to improve the backtester without higher-definition data. To simulate execution algorithm, one would normally need tick data and a simulated book with appropriate matching algorithms. Not sure how much you can improve execution if you are planning to stay with IB, But switching away opens up a whole new mess. Anyways, I am sure you guys are well aware of all these obstacles when it comes to scaling up.

Congrats and hope quantopian can discover and on-board many more quality algorithms in the near future. What you guys have done is truly revolutionary.

Did any of the managers voluntarily share their code with Quantopian?

@ Dan - Why did you decide not to answer my question? I know you saw it, since you responded to my comments immediately below it. I know one of your selling points is that you won't look at the code, but when the rubber hits the road and big money is at stake (you talk about $20M allocations), it kinda doesn't make sense to have a black box that only one quant fully understands. And if you ever go after outside (e.g. institutional) money, the black-box nature of your fund may be a sticking point (and have legal/regulatory implications). So, I'm curious if any of the first-round managers recognized this, and just provided their code to short-circuit the problem from the get-go?

How did you get access to the managers' algos to evaluate them? Were they contest submissions? Submitted directly? Gleaned from surveillance, without their submitting them (I assume that your terms of service allow this sort of thing, so long as you don't look at the code)?

This is exciting news! But I'm left with a question related to Grant's...

Does Quantopian allocate company funds to algos created in-house by Q employees? I'm curious because whether or not you do might affect how forthcoming Q members are in terms of revealing their code during the due-diligence process. If I had an algorithm being considered for an official allocation, I would certainly enjoy being able to have it looked at by people as qualified as the staff at Quantopian, and would love hearing their thoughts/recommendations. However, I'd be nervous to do that if those same people (or their close colleagues) were also allowed to write their own algos and compete for an allocation.

Sorry if this is specifically addressed somewhere else, as I haven't been following things as closely the last few months.

Either way, it's really exciting that the allocations have officially started! Certainly gives all of us something to work towards.

Grant - None of the managers shared their algorithm with us. They did share functions and/or code snippets that they were seeking help with.

The contract language with the author includes permission for review of the code as necessary for fulfillment of fiduciary duties. We haven't exercised that right, and we hope that we don't need to. If for some reason we do need to exercise that right, we would consider several options including review with the author or contracting the review out to a 3rd party, like a big-4 auditor. We will cross that bridge if we get to it - and we may simply never need to get there.

The phrasing of your questions about which algorithms were evaluated is pretty loaded. Rather than try to answer it in that tone, I'll say that we do testing on any algorithm on the Quantopian platform that we think might be useful - contest entries, live algos, real money algos, backtests, etc. I know you know, but not everyone reading this does: we never look at anyone's code stored on Quantopian without the author's permission. When we evaluate an algorithm, what we're doing is looking at simulation exhaust of that algorithm and seeing how it performed, using pyfolio tearsheets. Was the beta low? Did it survive the '08 crash? Were returns consistent? Were positions too concentrated? Was it well hedged? Was it exposed to particular Fama-French factors? If you can see the test in pyfolio, then it's something our quant research team has found useful at some point or another during the evaluation.

Graham - we do not allocate company funds to algos created by Q employees. We do have Q employees who write algorithms, but they're the things you see here in the community. They are the lectures, the content pieces, the demonstrations of data usage, the demonstration of new features, best practices, stuff like that. They're not alpha-creating algorithms.

This set of questions all comes to the core question of trust. We know that algorithm authors are trusting us with their incredibly valuable intellectual property (IP). We take that responsibility very seriously. We work very hard to protect the IP from external threats, and we work to make sure that we treat it properly with our internal processes. Furthermore, we know that trust is earned over time, not simply given, and we strive to earn that trust every day.

Thanks Dan,

I'm not sure what you mean by "The phrasing of your questions about which algorithms were evaluated is pretty loaded." I suppose my using the word "surveillance" could be misinterpreted, but it may not be clear to users that, as you say "we do testing on any algorithm on the Quantopian platform that we think might be useful." In fact, I'd always assumed that for consideration in the hedge fund, one would either have to enter the contest or send an e-mail to Quantopian requesting evaluation of a given strategy. If I'm reading you correctly, it sounds like that's not necessary. You are independently evaluating users' work on an on-going basis, and reaching out to them if it is applicable to the hedge fund. I guess this is captured in your statement above "We analyzed tens of thousands of algorithms." You haven't had tens of thousands of contest entries, so unless you are getting lots of extra-contest submissions, the algorithms have to come from somewhere.

So, how did you end up identifying (i.e. finding) the new managers' strategies? This was the question I'd hoped you'd answer, since I already have a feel for how algos are evaluated, once they are identified as candidates for the fund.

And did you select them solely on the basis of a minimum of six months of simulated trading results, combined with backtest results? Or had they been running with real money?

Interesting that the "contract language with the author includes permission for review of the code as necessary for fulfillment of fiduciary duties." I suppose that until you manage other peoples' money, there will be no fiduciary duty. But if you move beyond proprietary trading, some or all of the black boxes could be opened, but the managers would still maintain IP rights? The contract presumably allows managers to take their code to another hedge fund, for example, if it is no longer being run by Quantopian?

Why did you decide not to post the contracts? Or is that in the works?

Any new information about this? I would still be very interested to see some numbers and curves about selected algos so it would be possible to evaluate my personal algos against the selected ones.

They may not have even started trading the algos yet. I imagine we will see something in 6-12mos as they need some time to bake.

Plus, I think they will only release aggregate performance

Backtested performance should be fine though. Although I imagine that youll see nothing more than algo that makes or loses about 5k day

Financial Times article:

April 5, 2016 8:56 am
Fund using freelance programmers beats US stock market
Robin Wigglesworth, US markets editor

A “crowdsourced” hedge fund that uses the trading algorithms of a handful of freelance
programmers notched up a 1.93 per cent gain net of its theoretical fees in its first full quarter of
trading — beating the US stock market.
Boston-based Quantopian runs a platform for programmers, mathematicians and data scientists
that allows them to design and test stock trading algorithms, offering a $100,000 trading kitty to
the best-performing strategy every month, with the winner keeping any profit they make over the
next six months.
Its community of registered “quants”, or quantitative traders, swelledto nearly 60,000 last year,
when it also started an internal hedge fund seeded with its own capital.
This harnesses the best set of algorithms on its platform, sharing 10 per cent of the proceeds with
the authors.
Quantitative investing — using complex mathematical models, supercomputers and even artificial
intelligence techniques to identify profitable patterns in the swelling sea of “Big Data” — has been
around for several decades. But it has become increasingly popular in recent years, thanks to the
enviably strong and steady returns by the industry’s heavyweights, such as Renaissance
Technologies, DE Shaw and Two Sigma.
But there is a shortage of talent, with many computer scientists preferring to work in Silicon Valley,
so hedge funds have had to offer various competitions or establish alliances with universities to
ensure a pipeline of talent. Quantopian offers members access to data and coding tools so they can
trade as a sideline to their main jobs, or even from home.
Quantopian’s unaudited 1.93 per cent first full quarter returns are net after the fees paid to the
authors, an assumed 2 per cent management fee and the firm’s own 10 per cent performance fee —
a structure that mimics the common “two and 20” hedge fund fee structure. The S&P 500 rose 0.8
per cent during the first quarter.
John Fawcett, Quantopian’s founder and chief executive, stressed it was important to “stay sober”
on the early, unaudited results on trading the initial slug of $500,000 of capital, but said the initial
results were encouraging.
“The number of algos and the size needs to be ramped up in the coming months to achieve our objectives.”
The plan is to open up the hedge fund to outside investors once its record has been established.
Quantopian’s initial foray used computer-powered strategies for the US stock market designed by
eight members — who range from a mechanical engineer in Australia to a US software developer at
an internet search company — but the company plans to eventually use 20-30 algorithms at any
given time.
It plans to increase the firepower of the nascent hedge fund to $1m of capital this year, and
increase the leverage of the hedge fund from close to one time its capital to three times. In time it
intends to ramp up the leverage to six times capital to enhance the returns.
While there is growing enthusiasm over the potential for “DIY quants” to go head-to-head against
industry heavyweights, thanks to widely-available quick internet connections and increasingly
powerful personal computers, some sceptics warn that these freelance algo traders will over time
not do much better than their more primitive day trader predecessors.

Fact check - contest prize is actually:

One winning algorithm will win $5000, 2nd gets $1000, 3rd gets $500. Plus 100 limited-edition tshirts.


Nice article and glad to see Q generating profits. I wish Quantopian every success.

Can you share a some algos that didnt make the list but were noteworthy.

@ Dan -

Any updates from Quantopian? Above you say:

We are making additional allocations this quarter and going forward. We expect our allocations to increase in size over time.

How's that going?

Also, you alluded to a requirement that has been mentioned before:

Finally, they must have low correlation with other algorithms we have allocated to.

Any guidance at this point? Is your portfolio in need of specific "flavors" of algorithms, for diversification? This "low correlation with other algorithms" requirement is nebulous--how can algo writers determine if a given algo meets it?


The lack of guidance makes me think that the idea of a crowd sourced hedgefund is dead.

Miles -

I doubt that it is dead. They are just getting started in earnest, would be my interpretation. Based on, my sense is that there may be a bit of a pickle that most of the black-box algos are over-fit, to the point where backtests don't yield that much information. But then I guess the model is that too few master chefs would reveal their secret sauce, if this were a requirement for funding. Presumably one could share such details, and in fact it is implicitly encouraged on

Strategic Intent
We are are looking for algorithms that are driven by underlying economic reasoning. Examples of economic reasoning could be a supply and demand imbalance, a mixture of technical and fundamental drivers of stock price or a structural mispricing between an index ETF and its constituent stocks.

Q hasn't said whether they'd look at code of prospective managers, if it were provided to them. It would be odd not to, but then maybe there is a sound rationale for not learning the details?

I hate to say it, but the vision for low volatility, low beta, high returns, is basically an economic impossibility without the use of excessive (and potentially dangerous) leverage, that is.

I just coded an algorithm that over the last 16 years made about 1,600%, levered 2:1 (not so high) but the volatility is huge (40%). So, here's a question: Would you rather make 4% in a year with almost no volatility or 70-80% with some volatility?

Would you rather make 80% with low beta, low volatility but with 100:1 leverage?

I can guarantee than in the long haul, the Quantopian winning algos will get beaten by the higher volatility, higher risk S&P 500.

These are inescapable economic realities. You cannot make high returns without higher risk, and if you feel you have accomplished it, you are merely underestimating, or not correctly measuring, your portfolio risk.

I gather that the global crowd-sourcedness should be an advantage in diversification, which will beat down the volatility, so that leverage is safer.

Jeffrey is right. Nothing is ever free. If you want low volatility and high returns, you gotta leverage it up. It's just exchanging one risk for another (volatility vs leverage).

My simple-minded view is that this business is all about volatility, and not returns-chasing, which is a pitfall. Imagine creating a strategy with returns that follow (1+r)^n, with r high enough to cover operational costs and high enough to cover the cost of borrowing. It becomes a money-printing machine, up to a limit of total capital that it'll support. Low volatility comes from the fundamentals of the strategy, which in the case of Q's conceived crowd-sourced fund is to combine lots of diverse strategies, to beat down the volatility with statistics. Once low volatility (risk) is proven, then gross leverage can be applied up to a targeted volatility. The approach is captured in the article shared above:

It plans to increase the firepower of the nascent hedge fund to $1m of capital this year, and
increase the leverage of the hedge fund from close to one time its capital to three times. In time it
intends to ramp up the leverage to six times capital to enhance the returns.

I wonder what assumptions go into the 6X leverage number? Sounds kinda arbitrary. Maybe it is just the max they could get from the broker, and thus marketing spin?

If Q has access to 6X leverage, then low-volatility strategies with consistent but not stellar returns should be preferred (in fact, since they are dealing with black-box strategies, extraordinary backtest returns might be suggestive of over-fitting, etc.).

In my first algo which rebalances on a monthly basis 45% in SPY, 45% in TLT and keeps 10% in cash, and uses no leverage, I achieved almost the same returns at half the volatility and drawdown as SPY itself. This "algo" should never win an allocation from Quantopian, but the question is if I leave it as "black box" will they consider it for their "uncorrelated" sets of algos for the fund?

Searching (crowd sourcing) for low vol, low beta, low correlation, and high return algo is a fool's errand, and Quantopian should know this.

Hi there. I think Quantopian have more info that you think. They can look at all the trades and return stream. This is plenty of information to assess correlation to other algos.

Your algo would likely be disqualified for a number of reasons:
- concentration risk (2 etfs)
- not trading enough (not statistically significant)
- too big a drawdown

"Your algo would likely be disqualified"

You are missing the WILL be disqualified but it should NOT be...that is the main point....if you think about it.

" concentration risk (2 etfs)"

Do you know what SPY and TLT represent?

"not trading enough (not statistically significant)"

Rebalancing once per month for 14 years is more than significant. Some would call this active trading or swing trading.

"too big a drawdown "

Not when compared to the benchmark.

I don't really get your black box point. The asset managers of Q can see your algo's trades into SPY and TLT and therefore could see its exposure to stocks and bonds directly. Even if they could only see the return stream of your black box it would be a quick linear regression exercise to confirm the factor exposure to stocks and (treasury) bonds, and that this is comprehensive (no other factor exposures). They would then conclude the algo adds no diversification benefit to a traditional stocks and bond portfolio, and discard it, even though it passes the initial screen of having a relatively low market beta.

There's a good description of the fund criteria here:

On the rebalancing point, I don't think what you're proposing is statistically significant. The only decision is the initial one to hold equal dollar amounts of stocks and bonds. That's a great decision, but I count that as only two trades: one to buy and hold SPY and one to bug and hold TLT. I'd want an algo to make a new decision like this every month at least.

From the above link:

Actively Trading Algorithms
Your algorithm has to actively trade. We measure that by looking at how often your portfolio turns over per year. This is important because we need your algorithm to be predictable. Our algorithm selection model only works if we have sufficient data to analyze about your algorithm. Algorithms that rebalance or trade very infrequently are difficult to model with limited historical data. We are looking for algorithms that turn over their portfolio between 12 and 500 times per year (e.g between once per month and twice a day).

I've read that other thread now. I agree with you, that 1/n asset allocation is a good starting point for most retail investors, as long as the assets have roughly equal volatility (as SPY and TLT nearly do... EDV is better).

But the point of any hedge fund is to allow large institutions (or HNW), who already are already long plenty of stocks and bonds, to diversify. It's not to "beat the market" (I.e large cap US stocks). They are already doing this by diversifying into bonds, large caps, small caps (private equity), non US assets, property, etc.

"But the point of any hedge fund is to allow large institutions (or HNW), who already are already long plenty of stocks and bonds, to diversify."

The point of any asset manager (including hedge funds) is certainly not to offer diversification, but to offer a return greater than X on a risk adjusted basis, where X is a minimum return expected by the asset holder/investor. If you are a hedge fund manager and your client tells you they are already 100% in SPY, then you need to convince them to turn over the entire account, so you can "manage" it by adding other assets to it. If you think that your client needs to be 60% in SPY, then you are not going to ask to manage just the other 40% - you will ask to manage the the entire account, and offer a superior return on the 60% which are in SPY. Diversification might work as a sales pitch, but it is not the ultimate goal of any fund - the ultimate goal is gathering assets.

So, taking diversification out as a goal, and since Quantopian is about trading stocks (soon futures) then the only thing you can call X so far is the return of SPY. If Quantopian could be used to trade bonds, then X could be the performance of TLT, and so on. Therefore, beating the SPY on a risk adjusted basis is the only identifiable goal on Quantopian, and not beating or diversifying into some nebulous undefined alternative "market".

Ok, what you describe is not my understanding. What I understand is closer to this:

Behavioral Trader -

Presently, Q is putting up their own money. As I understand, they are working to position themselves to take on outside investors (presumably, is not about investing in Q but in their future financial products). They need to offer something unique as a product, to be able to justify something like a 2% management fee plus 20% of the profits. You are probably correct that if investors can't tell the difference between a Q product and a 50/50 portfolio of SPY/TLT, it is a non-starter. There is a lot of money sloshing around out there. My sense is that if Q can put up a bit of a real-money track record, they'll start to attract outside money. They have the "unique" part nailed with their platform and approach, so investors (speculators) just need a warm fuzzy that the whole thing might work. In 50 years, it might turn out that the investors should have just gone the SPY/TLT route, but what's the fun in that.

Hedge funds as a group have done pretty terribly over the last year or so, so you have a lot of money willing to move and the hurdle is being lowered. And investors are realizing a lot of hedge funds are just levered beta and there's no reason to pay 2 and 20 for many of them. Quantopian would actually offer something truly unique and diversifying for their portfolio. I think now is actually the perfect opening for Quantopian.

Another article -

Quantopian hires investment chief ahead of investor push

Hedge fund that crowdsources trading algorithms appoints former Millennium executive to woo market

"The company’s initial trading only used eight trading algorithms but plans to ramp that up to 20-30 over time, and is in the process of setting itself up and registering to be able to raise outside capital from institutional investors by the end of the year. "

Perhaps there have been 5 more allocations, after the 3 mentioned above by Dan?