How Token Metrics uses machine learning for crypto investment analysis

0:01I'm Ashlyn Addison from event chain for investmentpitch media and FinTech news network and today on blockchain interviews we have Ian Polina the CEO and founder of token metrics Ian thank you so much for being on the show it's a pleasure to have you here and it's a pleasure to see you again Thank You Justin happy to be here you're very welcome I'm excited to learn

0:20a lot about token metrics today I know it's been growing rapidly and when I first started following you back in 2017 you had quite the crypto run which seems to have led into what is now the formal creation of token metrics could you give a little bit of background to the viewers on how to 2017 go how did this all start and how did you get to this point today yes absolutely so token

0:43matrix was born from my earlier ico analytics spreadsheet basically bringing a data-driven approach or what we call the Moneyball approach to IC investing and we learned that was definitely a big meet in the market because at the peak prior to me taking down my spreadsheet in January 2018 the spreadsheet was being viewed over 1 million times a month

1:04by investors all over the world from retail investors to hedge fund managers and people at well-known crypto funds to us that really showed us there was a big need in the market to help people with investment research but we also realized that the spreadsheet cannot scale there was just a simple Google sheets spreadsheet it cannot scale to past a million people so we said you know what

1:26let's now build an actual application so the first step was putting a web application then down the road we plan to build a mobile application so we've spent the last year and a half building token metrics but also taking everything that we've learned in 2017 and 2018 and saying how can we make this even better how can we make this 10 times better hundred times better and to us the

1:48obvious answer was we have to get look at a lot more data pumps not just what humans can manage but we have to leverage machine learning and that's really the next evolution of human metrics is now we're building a platform that uses machine learning to help investors find profitable investments and also filter out scams in the crypto space okay great and yeah machine

2:08learning and data is key and I know that those are two of the you know foundations of your competitive Vantage with token metrics and if I recall you know when I was following you when you had the spreadsheet you know you were using some of your expertise that you would learn in IBM and some data analytics to look at you know standard deviations and and certain the

2:28supply of coins and things like that is this like now flushed out into a machine learning algorithm that you and your team have created specifically for token metrics yes correct so we're taking that same approach but scaling this up on a whole different levels so the first step is starting with human capital so we take my expertise but we also find other experts from Goldman

2:51Sachs JP Morgan to bring in their entire brain trust and we codify that into software and then we take that software and we add data we had tons and tons of data from different data sets both quantitative and qualitative data so think data from coin market cap but we also have our experts going through and manually scoring every single country currency based on fundamental analysis

3:14based on code reviews based on technical analysis and we take all this data and we ingest it into the machine learning model and we're creating special custom tailored models for every single cryptocurrency so right now we have over 100 custom killed' machine learning models for every 50 kind seat and every cryptocurrency has four different models so in total we're running over 20

3:37million machine learning models a day and we don't think anywhere out there anything quite like this and meat is accessible to the average investor because our belief is that the future of investing is with machines whose machine learning we've seen the same thing happened on Wall Street with on funds and for the average investor to compete there isn't really a tool out there or

3:57platform out there unless you are hedge fund with lots and lots of money to going to invest into a hog fund a research you cannot compete so our idea is to build your platform who spend millions of doubt about on money building us from experimentation to bringing the best team around the world and making this available to the average investor so they have a chance to

4:16compete in the crypto investing mmm-hmm wow that's great and you've been refining this for you know two almost three years now just on the ico part but at the same time the landscape has sort of changed in there used to be a ton of icos in and startups in 2017 and it seems that there aren't as many now or maybe there aren't as many good ones or maybe you're just

4:38not hearing about them as much because the hype can you talk a little bit about the differences between when you first started and those investment thesis is that you made then and looking at it now you know I'm not sure if this is just my perspective but it seems like more people are investing in long-standing cryptocurrency projects that have been

4:56around since before 2017 that more have you know stronger fundamentals to coins that have been around since 2014-2015 versus coins that are just coming out whereas it seemed like everything could do well for new coins in 2017 it's been a little bit tough for now would you agree yes yeah totally agree I mean I SEOs are no longer in favor anymore and we'll all

5:18submit that that evolution as well so for instance token matrix is not just for I see us now rather we're looking at cryptocurrencies that are already trading on exchanges because we answer the question of what coin should somebody invest in at this point in time so we factor in coins trading on exchanges like Bitcoin aetherium top 20 top 100 in multi market cap plus also

5:41those new cryptocurrencies coming in so for example right now toka matrix has our team has gone through and being able to find projects that have potential before anybody else finds them so for instance we found synthetics synthetics has been one of the best performing coins of this year about our development team told us that's the top 10 best technology right now in terms of any

6:03cryptocurrency so being able to find early picks like that is something that we think is something else we're bringing value to investors even those whom who aren't really pro ICO because IC o—- IES are definitely out of favor and for us to really have long-term success you have to expand an entire system to looking at tokens already trading and that was really the hard

6:25part that's why it took us over a year just putting out the system but now we think anybody out there even traders can take advantage of this platform that's great and with the machine learning algorithms that you guys have created you mentioned you know hedge funds and really large companies may have some products that are sort of competitive like this but in the

6:45cryptocurrency industry and with products that lower the barriers to entry for people that aren't you know necessarily accredited investors or they're just into the crypto space are there is there other projects that are doing things similar to token metrics right now um we haven't really seen anything quite like this I mean we have seen a few other competitors but they

7:06aren't really doing anything in terms of machine learning like we are either they're doing it like by having nine young people go through and rate coins but for us I haven't seen anybody who's doing this level of detail and also doing this level of back tests and making everything a fully transparent so for example if you're going to Tokyo metrics we have 30 day price predictions

7:25for it cryptocurrencies and we display the accuracy of each model over the course of time so when we're trying to hide for example our model for predicting bitcoins price over the last 30 days is about 90% accurate well I'm not of a predicting aetherium is about so many something percent accurate so our goal is to be better than the average investor or trader if the

7:46average investor trader has a success rate of about 50 something percent something even as low as 70% can help so the average investor have to have that edge definitely and you mentioned that you're almost competing with with quant trading or you've integrated that into the platform can you talk a little bit about for the newer investors what is what is quarter quant funds what is

8:07quant rating why is it so successful and how have you integrated that into the platform so quantitative trading or investing is essentially just using algorithms and mathematics think of it as Wall Street is not is no longer really depending purely on just regular traders or investors they're bringing in PhDs from Harvard in mathematics and engineering and physics

8:29to find patterns in data and for example the best-performing Wall Street fund of all time at this point called the medallion fund is a quantified Jim Simons fund what was recently featured in the book called the madness of the market his outperformed warren buffett in the last 30 years his is had an are alive about almost 40% while Warren Buffett and analyze growth about about 20% still

8:59his beat run Watford by two times and he's not a traditional Wall Street guy he was he was working in Austin former coal cracker the NSA he said you know what we can apply mathematics to investing because people think markets are efficient we're done really efficient and he said there's somewhat efficient there's semi efficient but there's still patterns in the data very very small

9:24patterns and he's able to find those patterns using math and Wall Street has now shifted towards being almost purely quantifying and we think the same thing is happening in crypto so according to PwC in 2018 well Bitcoin went down 72% the only crypto funds that made money we're quad funds one fund had eight percent ROI last year this is in a bear market

9:45so you think this trend is now coming to Couture's crypto and the average investor is out of luck unless they can evolve and adapt and we think we've got the platform top them with that mm-hm that's great and just to be clear your platform is more of an educational platform you guys don't have any licenses to hold people's money or to make trades on their behalf or financial

10:06advice is it strictly education or what to what extent do you go to helping the members oh oh right now we basically built a data analytics platform which should think of it as monix time for crypto and this is an ax best research platform to help investors with their due diligence and research of projects now we're in the process of applying to become a sec registered investment

10:29advisor and that does some EC not ready yet but that's that's that's on our roadmap so we think sometime next year hopefully not pending SEC approval the idea is to become a sec license to Robo advisor so being able to take users funds through a qualified property custodian approved by the SEC and almost like a trading bud but regulated by the SEC mm-hmm that's interesting especially

10:52with you know the SEC coming up with all these different litigations for different crypto companies for different reasons and they're still sort of exploring you know what's the proper way to go about with startups versus investing in major coins like Bitcoin and etherium so all the best to you guys in growing that and I'm looking forward to seeing it happen in 2020 and I hope that

11:13the market gets a little kick back into it in 2020 for all of us as well so for the people that are already members of token metrics can you just explain a full picture of you know what is it exactly they're seeing price predictions they're seeing a few other things right now yes so when you're looking to talk metrics calm the homepage we give you the best cryptocurrencies at this point

11:38in time looking at several data points looking at fundamental analysis looking at code reviews our team goes there and manually checks the code of each project doing automated technical analysis and we take all these different data points and create one single holistic break so for example in our top 10 we may have projects like Bitcoin its premium Cardno

11:58Manero what what have you but it's kind of like women market cab but we're giving you the highest-rated cryptocurrencies with a highest retention for growth so we come from this from the investor perspective when I just giving you price action yeah they were giving you as an investor in our opinion based on the data what projects have the best chance for growth

12:19definitely and you know I recall in your 2017 reviews you know looking into the team a little bit and then of course looking into the github an actual project code it seemed like a lot of these projects had you know a billion dollar market cap and like a non-existent github yes yes so that's really one of the things we had to cream with your metrics we brought in a former

12:41IT executive for Goldman Sachs who was part of the global investment research division a developer also worked with that JP Morgan at Barclays with their payments platform to come in and he helps us to do the due diligence on all these products code so because as investors as somebody myself who's made lots of investment in the space and that's what successes but also lots of

13:03failures more importantly being able to learn from those failures and bring in other experts around me and our team to help us not make those same mistakes again so we think we book that same platform that's all we have for each cryptocurrency we're looking at over 74 different data points and that's just from the manual from our team going through and manually

13:23everything then was added data points from the machines and ap is and then using all of that to have a better perspective on what projects can outperform in the long term that's great and now we're heading into 2020 what are you and your team working on to for future integrations and upgrades for the token metrics platform so right now we're working on optimizing website for

13:49mobile so being able to because over half of the world is now mobile so creating a platform that can scale globally to Mumbai and that's just some people it's not anybody has a computer and then we're also working on creating an affiliate program because lots of people have we talked to us saying hey we like the platform want to share this insane way for us to really kind of

14:08share in the success of the platform so we plan to launch an affiliate program in about a month or so and just really getting the word out there helping people leverage the machine learning to become better investors creating educational content for everybody