My algos are up 57% YTD. I want to sell them.
I’m an engineer trading prop using algorithms I built. I’m currently the top trader using these algorithms, and looking to sell them to a hedge fund to move on to my real life passions.
But, I do not know where to start. Has anyone sold algorithms to a HF here or purchased algorithms for a HF? Should I just knock on HF doors? Do HFs search for outside algorithms?
Any advice would be appreciated.
The algos: +57% YTD, Max DD 6% FX based, built on 14yrs of data WF on 6 yrs, live performance 2yrs. Fully automated.
I know versions of this questions have been asked before
- I’m ready for negative responses and deflections. So I'm going to answer common questions here.
- Q: Why don’t you trade it yourself if it's so good? A: I am, but 1% of 10,000 is much smaller than 1% of 1,000,000.
- Q: If the algorithms work why sell them? A: I made the algorithms to be sold to fund my engineering ventures (aka my real dream)
- Q: What about strategy decay? A: I trade a portfolio of strategies. All strategies decay overtime, so it is expected to experience strategy decay. But no one knows when that would occur ahead of time.
- Q: What about scalability of the portfolio? A: It's hard to determine if this portfolio can scale to X, without knowing what X is. I believe this will be apart of the due diligence of the potential buyer.
First question should be scalability. A strategy that returns 57% on 10K might be eaten away by transaction costs when you’re trading a few hundred million, especially if it’s making trades pretty often.
This is 57% after my transaction fees. And my setup isn’t optimized for tight spreads and low commissions so maybe a firm with better execution could benefit.
A TCA would have to be done to know for certain the costs of the potential buyer. But that can be done down the road
Not exactly what I meant - transaction fees grow exponentially with trading size.
Scalability is the first concern (as already mentioned)
You say you have been trading for 2 yrs but only give the 57% YTD number, how has it done throughout your live trading (and at what scale) and what is the backtested performance?
Over the two years live 73%. Backtested results gave ~35%/yr. I only mentioned the live results because most people discount backtested performance.
I’ve only been able to scale up to 600k at the prop firm. And I recently got to 600k after a few scale ups in capital by the firm.
I believe this is scalable because the strategy doesn’t rely on short targets (5-10pips) and it doesn’t rely on speed. So it shouldn’t be affected by latency, or broker spreads.
Of course a TCA would have to be done for each individual case of account size and broker costs.
My transaction costs over the two years was less than ~1%. But that’s specific to my firm’s broker setup.
Maybe I’m misunderstanding, but if live PnL over 2 years is 70% and you are 57% these 4 months, you were up 13% over 20 months?
Best move it to get a seat at a first loss shop or an arcade style shop w/ low commissions if you really want to put your money where your mouth is. We dont "buy" algo's and we sure as shit wont run something where the creator doesn't have active skin in the game / isn't actively monitoring and tuning it on a daily basis.
There is no "get a lump sum payment and move on" option - if its good, then you'll be glad this isn't even an option. That is assuming its good (not trying to throw shade, but 99.9% of these kinds of scenarios dont work out / scale at all / decay in a matter of weeks)
Hi, are you providing advice or commentary? I have no issue supporting the algorithms that is a deal that can be worked out. I'm not trying to run, leaving a firm holding the bag. I would just like to know options in this scenario. My money is where my mouth is. Development cost was not cheap. And I've been using the algorithms to support myself through the avenues open to me.
Strategy decay is inevitable. No one can create a strategy that doesn't decay, not even Jim Simons.
What’s he’s telling you is that no reputable shop will want anything to do with your algo unless A. you’re going to post up everything you’ve got because you’re that confident in it or B. you have a track record with capital of scale. Most funds get sales pitches like yours all the time, and have zero interest in buying a strat that the developer is not actively involved in. It’s not like selling an app, if you want to sell your strategy you’re going to be sitting in a seat trading it/maintaining it, and some of it will be with your money
Im not saying track record - and this is advice. If you really want to see how far you can take this, this is what you do:
1. Run the algo on some brokerage that can provide an audited track record. Even if you're running only a 10k book, this is farther than most people get. You may have already done this.
2. Contact first loss shops that deal in the asset class you are looking at. Build a brief pitch deck on what your strat looks like and what you would like from them. Offer to start at the same scale you currently run at and build as you go.
2b. Alternatively, try reaching out to arcades to do the same thing. Terms and infra wont be as good.
You created this algorithm at a prop firm, who helped pay for the strategy and initial scale ups? Are you fully sure you can even go and market this strategy elsewhere? Truly will not lie what you thinking of is not how it really goes, more sensible is to set out a goal to leave the industry in 5 years and use this past success as an ability to jump to a firm that gives you greater access to capital. What is the holdback from the firm to further scale up, are you getting 20% or greater of the book?
I created the algorithms before joining the prop firm. I funded the development and own the algorithms outright. So I can go where I please. The max capital they allocate to their traders is 600k not sure why.
Okay I have no problem being with a firm that would pay me to use my algorithms. I just would want to control more than 600k how do I do this?
Okay that was a key bit of information, the firm's agreement was to offer you a "X amount" of capital but the models/algortihms are yours and they get % of book I take it. Now when you mention 600k, are we talking the outright exchange initial margin they are allocating or is this the cost for entire book/strategy. If you are up 50%, is the firm allowing your initial margin to now grow to 700k, 800k etc...Or do they refuse to reinvest back into the firm and keep your capital limited. What is holding period for this capital, and how much is typically utilized on a weekly basis, since you mentioned this is not a HFT style strategy.
.
-Lever it with a PG and scale as much as possible. Once you are big enough, move off the PG.
-Arcade/prop shop?
-HF unlikely due to small scale.
What is a PG? And what is a arcade?
this problem has been solved.....go checkout FundSeeder.com and Collective2.com setup your account with your real name attached in public (so you can't create multiple and just cherry pick the best)
these create a live track record that hedge funds can use to evaluate the performance....lesson #1 nobody will trust you without this public track record
Hi I have an account with fundseeder and collective2. But this only solves the problem of having a live track record. It does not solve the problem of me getting funding. Unless I hope and pray that someone browses the >1000 accounts to find mine and fund me. Also collective 2 boosts different strategies every week based on their preferences. No really a fair and balanced platform.
Fundseeder has no guarantee of funding, or guarantee that anyone will even see your account.
I can either be one in 1000 or I can directly reach out to hedge funds or potential buyers. I prefer the later.
FNYS (First New York Securities) will hire prop traders who prove to have successful strategies....and other prop firms / platform hedge funds will also...however...nobody will "buy" your algo....the best they will do is hire you to trade the strategy. if you do well for 1-2 years, you'll make enough to fund your other ventures
Hi I have an account with fundseeder and collective2. But this only solves the problem of having a live track record. It does not solve the problem of me getting funding. Unless I hope and pray that someone browses the >1000 accounts to find mine and fund me. Also collective 2 boosts different strategies every week based on their preferences. No really a fair and balanced platform.
Fundseeder has no guarantee of funding, or guarantee that anyone will even see your account.
I can either be one in 1000 or I can directly reach out to hedge funds or potential buyers. I prefer the later.
I’m not even asking for someone to trust me yet, I’m asking for the process of even getting someone to the table.
contact First New York Securities http://firstny.com/ ...also DV Trading https://www.dvtrading.co/
both of these firms have capital and will hire a trader with a profitable strategy with a track record...they don't care about prestige
A few key parameters could help get a sense of scalability, could you attach a link to your collective2 account? I may be interested in investing in it myself, esp if capacity constrained and market neutral.
And generally few funds will outright buy an algo, its generally more lucrative to join a firm as a PM and trade it yourself.
Why not just trade it for Republic credits and then gamble it on a podrace?
Galactic inflation is too high.
This year has been so weird that 57% YTD returns could easily be a fluke. You said your backtest was 35%/year, and you didn't specify how long that backtest was, but let's say it was a good multi-year period.
Consistently doing 35% a year with only 6% drawdown is a FABULOUS result, maybe not the Greatest Of All Time, but way way up there. Most funds can't do nearly as well and they have dozens of PhD quants and invest tens or hundreds of millions of dollars in market data and compute resources. You're claiming you beat all those funds single-handedly, with no formal training, with far fewer resources, in basically a part-time hobby project?
The question every fund will ask is -- you know how that sounds, right? Do you really believe you're a genius who in your part-time discovered strategies that all the full-time quants on Wall Street overlooked? Are you *SURE* this isn't just another "90% chance I make millions but 10% chance I lose billions" strategy that will eventually give back all its gains and then some? Make sure you have an answer for that question.
Returning 35% with 6% drawdown is by no means easy at any scale, but considerably easier on 600k than 600m, so it’s probably possible. Those dozens of PhDs aren’t hired to collectively produce a few hundred thousand PnL lol.
Hi thanks for your comment. I mentioned in the post that I built the strategy on 14 years of data and did a WF (Walk Forward Analysis) on 6 years of data. This means this is a 20 year backtest. After doing 10,000 monte carlo simulations of the portfolio in a backtest my risk of ruin is still 0%. Also from the live data my risk of ruin is also 0%. Only time will tell, but this portfolio survived 2020 so I am hopeful.
I'm a PhD Engineering student, learning the same things those quants learned. Except, I read hundreds of trading books, listened to hours of podcasts, interviews courses online and spent a lot of money on magazine subscriptions... all before I made my algorithms. I spent hundreds of dollars on market data and software to build my algorithms "the right way". I have a deep background in statistics, robust optimization and machine learning... I'm publishing new research in the field as we speak.
I graduated valedictorian from my engineering school at the age of 20 and was making six figures by 21 in a fortune 500 company. I may not be a genius in some people's eyes, but I'm no chimp either. I'm a true definition of a self starter, and yes I was able to do this on my own and apparently beat those funds.
You should remember those funds with PhDs still have to teach them the basics of trading and essentially rewire a stat/math major for the markets. Those PhDs don't come with any idea or experience around trading and they then would get biased by the firms style of trading and reproduce whatever the firm is doing. However, in my situation because I had no guidance I explored to find my own style of trading algorithms and developed something with an edge.
Umm. Walk forward is good, yes, but just based on those three sentences alone, I can already tell your approach to backtesting and risk management is...um...let's say "different than how my firm does it".
But hey, that's the beauty of trading -- as long as you're willing to put your own money at risk then it's possible that I'm the one who's wrong; so if you really believe in your strategy, then you should ignore what I think.
My only advice then -- did you say you work for a prop shop but developed this algo before your joined and you therefore have the right to sell it? That may well be true, but are you 100% sure your company agrees? Even if you do manage to find a buyer for your algo, they will require proof this is not stolen IP. If the buyer contacts your old company asking whether you own the strategy, your old company could scuttle the deal (and maybe even threaten to sue you too) if they want. Do you think your old company would allow it?
something difficult to believe is the backtest thru 4th quarter 2008...many major markets had HUGE intraday swings with VERY wide bid/ask spreads when Lehman blew up which caused forced exits of positions at uneconomic prices...any strategy that did well before or after the Lehman blowup should have been harmed DURING the Lehman blow up...and vice versa. Most of these dislocations eventually mean reverted...but intraday almost everybody with tight stops would have been stopped out multiple times.
Very interesting, keep us updated
On an another note, the same way small funds are outsourcing sales/fundraising, trading and IT/support/compliance, I think you might actually have something of interest with the right kind of marketing. But targets would be startup fund/boutique AM and as stated before, they might want you in the PM seat to handle the strat for a few years first.
Awesome thanks for the insight. I will update if change positions hopefully.
Damn this is sick. I hope everything works out well for you man
Thanks! I will update if anything big happens (fingers crossed)
Well it's a year later now. I always wondered what happened to all these posts about people who had amazing returns in their backtest. Curious if it's still working for you now a year later...
Would love to see it as well...
Hey see above.
Hi just seeing comment, I've been busy finishing my PhD. But I did update near the end of last year here. The prop firm I was trading for turned out to be a scam and they ran off with most traders funds (including mine lol). So I'm currently running my algos on my personal funds and documenting that on my public blog. The algorithms are still doing "good" this year but still too early to tell. :)
Really stupid question: what programming language do people use to build these algorithms? Python?
Also, Im assuming you start with a strategy that backtests well and tweak it from there? Anyone have any good recommendations for more reading material on this stuff?
Yeah, python is by far the most common language for individuals working alone to write trading algos. If you work with others in a hedge fund, then python is still the most common, but sometimes you'll see C++ or Java, or maybe a quant will develop an algo in something like matlab and then give it to the developers to translate.
As for how to get started...try reading the headlines on https://quantocracy.com/ each morning and clicking whichever links seem interesting. They run the gamut from easy "how to set up python for your trading environment" to medium "here's a sample trading algo" to hard maths.
I have developed my algorithms in python and Metatrader language. Metatrader makes it easy for me to deploy, backtest, build and host in a cloud. Which is all important as a solo-operator.
As someone who’s interested in learning how to code to eventually build an algo, where did you start as far as finding which macro variables to focus on/include in the algo?
Iure corporis aspernatur a eum vel. Consectetur omnis ullam et est.
Totam occaecati ad ut ut ex commodi deleniti. Rem odit sit ut id optio quas. Optio consequuntur voluptas aut ut sed rerum repudiandae. Quidem voluptate et voluptatibus quae non modi. Consequatur velit quia ratione aut non autem. Odit harum dignissimos aut vero autem molestiae magni.
Sed ipsa quia asperiores eum doloremque explicabo porro. Est dolor consequatur exercitationem sunt adipisci itaque aut. Est itaque temporibus excepturi adipisci. Omnis et a ut omnis.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...
Perferendis explicabo numquam et qui corporis commodi aliquam. Perferendis praesentium laudantium voluptatem nobis dolore rerum velit. Est repellat atque nobis mollitia quia esse consequatur. Minus rerum earum tempora dicta. Ea error cum accusantium minima quia. Tenetur dolor eum nobis quaerat et. Tempora ad recusandae qui harum.
Assumenda qui voluptatem quas. Rem nihil reiciendis recusandae enim sunt pariatur at vitae. Animi dolorum tempore necessitatibus iste. Sed ex enim blanditiis neque et quo. Mollitia amet non ut magnam. Animi aliquam eveniet est sed.
Dolores rerum illo velit sunt exercitationem. Esse quas rerum occaecati.
Vitae eos tenetur aliquid laborum. Necessitatibus animi officia ut aut autem quo repudiandae.
Numquam accusamus repudiandae ducimus minus. Qui iste deserunt autem et excepturi est hic vel. Voluptas voluptas minus nemo qui et minus adipisci provident. Qui aspernatur laboriosam at culpa suscipit. Voluptatum ex mollitia et blanditiis assumenda. Vero nesciunt velit non dicta odit sed quaerat.
Voluptate maiores minus molestias eum officiis occaecati voluptas. Iste natus natus aut atque et ad. Eaque dolorem quam ipsa atque quia tempore. Sed et odit velit laboriosam et.
Omnis illum est magni. Reiciendis ab voluptates quia fugit veniam. Aut repellat aut impedit voluptatem id laborum tenetur.