The Strange Case of ICO Returns

2017, Jul 25    

2017 seems to be, along with perhaps a few political contenders, the year of the ICO (or Initial Coin Offering). The potential of cryptocurrencies is beginning to take off, and we are seeing this not only in valuations but in the strength of the developer communities that are forming. It is an exciting time! But along with the innovation enabled by the new economy, there are a raft of risks associated with the lack of standards and regulations. Here at Sangus we are working hard to develop a new form of due diligence marketplace and rating agency to help address the information disparities faced by ICO investors.

The ICO space is still relatively new, making a detailed analysis of historical returns difficult. But even with the data dataset available from CoinMarketCap we are able to see some interesting trends, particularly when it comes to the ratio of “flops” to huge successes. You can find the code used to generate the results in this article here. All data comes from historical snapshots provided by CoinMarketCap from April 28, 2013 through June 18, 2017. Please comment below with thoughts about the analysis and we’ll try to get back to you as soon as possible (also happy to share the dataset if need be).


The dataset includes two types of coins; regular cryptocurrencies (Bitcoin, Ethereum, Litecoin, etc) and derivatives based on Ethereum (Gnossis, Golem, etc). We’ll go ahead and ignore the distinction but it is worth noting that you might obtain somewhat different results if you zero in on coin offerings as opposed to token offerings.

CoinMarketCap only offers weekly snapshots, so for a given token our dataset will only reveal its price in the first week, rather than at the ICO itself. In addition, any coins that first became available prior to April 28, 2013 are excluded in our analysis (to keep comparisons between tokens fair). Fortunately this only applies to 7 cases (BTC, DVC, LTC, NMC, NVC, PPC, TRC) including the usual culprits BTC and LTC. Any coins that disappear from the dataset were delisted and so are assumed to have fallen to value of $0 (according to Gliss, delistings are typically the result of the coin being removed from all monitored exchanges). This applies to a rather disappointing set of 765 cases.


A total of 1,434 coins are considered, of which 7 were released prior to the start date and 765 were delisted at some point. The chart below shows the number of new ICO’s each year. Evidently 2014 was quite popular at almost 700, and at the current rate 2017 is on track for almost 400.

Annual ICOs

More interesting, when it comes to scale, is to focus on the total initial market caps. Note that CoinMarketCap calculates this as (circulating supply * price) which, while analogous to calculating public float, makes it tricky to compare values. Some founders choose to sell the majority of the coins upfront (so that they are circulating), while others either hold back a fraction for later funding, or more commonly, use a vesting system that prevents coin sales for long periods of time. For example Steem implements a fairly complex economy in which the main currency (Steem) can be converted to Steem Power (removing it from circulation) and only transferred back to Steem over a period of two years.

Annual Initial Coin Cap

By this metric, 2017 already trounces every other year, and remember this only goes through June. At well over $2 billion in initial coin cap in the first six months, we are at almost four times the annual number for 2014. And in terms of the mean coin cap these effects compound such that the average initial coin cap in 2017 works out to over 10 times that of 2014.


With capitalizations skyrocketing, one might wonder what kind of downside investors are looking at. And that is where the story becomes a little more interesting. Solely based on delistings, 765 coins fell to $0 from their release (over 50%). Naturally this metric is biased against offerings that happened longer ago (as they had more time to fail), so instead we can break down coin performance based on weeks since ICO. The table below shows the return quantiles (as a ratio, i.e. (final_price - initial_price) / (initial_price)). We exclude coins that are too young to have the relevant return and report the distribution of returns using percentiles (i.e 0.1 corresponds to the 10th percentile). Surprisingly, even by the first week, the 50th percentile corresponds to a return of -0.42 or -42%, so investors in the typical ICO lose around half their value by the end of the first week. Looking farther out, by the 96th week (around two months short of two years) everything through the 70th percentile has lost 100% of its value, and even the 90th percentile has lost almost 90% of its value. Let’s call these “flops”.

ICO Returns Percentile

Given that the market has substantially heated up, it makes sense that returns through week 24 look stronger (simply because they include more 2017 coins). But the main takeaway here is to see just how quickly coins die out; after around a year virtually all of the coins have lost 3/4 of their value.


Much of modern investment strategy comes down to diversification. So let’s suppose you took that approach to Initial Coin Offerings. Even accounting for the vast majority of coins collapsing, if you had invested $1 in each token or coin offering beginning April 28, 2013 (for a total of $1,427) you would now have $417,617 (292 times your initial investment). If instead you chose to put all your money into the baseline Ethereum platform (again $1,427) you would have now have $747,718.

Unfortunately that is not quite reflective of the experience of average investors in this space. The chart below shows a quantile breakdown of coin returns; the lion’s share of the profits go to that top 10th percentile and more or less everyone else loses. This sort of payout distribution is fairly reminiscent of early stage venture investing; just keep in mind that the sums we are talking about here (tens to hundreds of millions of dollars) are far larger than seed or Series A investors typically participate in.

Token Return Distribution

Value Destruction

From a long-term market perspective what is most shocking about this funding system is the amount of capital that is lost in the “flops”. We can take a conservative estimate of value lost by summing the initial coin market caps for all coins that had an ultimate return of -95% or worse. This works out to $881,512,081; almost a billion dollars. In the context of international financial markets this is hardly a drop in the water, but relative to the global cryptocurrency economy this amounts to almost 1% over 4 years (based on the peak estimate of $100 billion in June 2017).

We believe that with the growing community of professional developers, the raw number of individuals dependent on cryptocurrency markets for their livelihoods will continue to tick up. And if there was one lesson to be had from tech in 2001 it is that friendly funding environments can quickly turn into a playground for pump-and-dump schemes, or worse a race to raise as much as possible before burning out. Perhaps 1% lost to failed projects is within reason but there is a real possibility that another Enron could do irrevocable damage.

About Us

Sangus is a one-stop, neutral site to help find ratings, details, and links for every Initial Coin Offering (ICO). Our mission is to help the ICO evolve into a trustworthy and respected fundraising mechanism, starting with providing would-be token buyers with all the information they need.

ICOs don’t pay a fee to list on our site and we don’t charge for reviews or ratings. Our client is the public, not ICOers. If you’d like to learn more, read about us here or jump straight into the site.