There are “pump-and-dump” manipulations in cryptocurrencies. The issue is actually widespread. In just the first 7 months of 2020, two academic have documented and studied over 350 such cases.
Cryptocurrency pump-and-dumps differ from the old one: they are brazen, fast, executed by groups of actors. The organizers front run their groups, making great returns for themselves. The rest of the participants lose money on average, but keep on participating.
Academics are focusing on the topic. Here is the current state of the art.
A bit of history
On February 6th, 2018, Christopher Giancarlo, the Chairman of the CFTC, testified in front of the Senate’s Banking Committee. His opening statement explained the role of the CFTC in regulating crypto currencies. He reminded us all of their benefits, but also of their potential dangers, notably manipulation. A specific note focusing solely on pump-and-dumps events is actually available on the CFTC’s website. This short document explains how manipulators work in online groups. The organizers inform group members of the next manipulation event in public chats:
The Wall Street Journal reported on the issue in August 2018, and had identified 175 such operations. Explanations:
- “A pump-and-dump scheme is one of the oldest types of market fraud: Traders talk up the price of an asset before dumping it for a profit and leaving fooled investors with shrunken shares“
- In the crypto era, boiler rooms are replaced by “chat groups“. The biggest chat is “Big Pump Signal”. With 74,000 followers on Telegram, it eventually had to expand into Discord to manage its growth.
- The process is repetitive. Ahead of time, the organizers announce the exchange, and the time at which the pump will take place.
- At the last second, they announce the coin, the only missing information. The effect are drastic – the asset jumps within seconds, reaching a peak within minutes, a return of 50-75%.
- The WSJ simulated one of these schemes from intraday data:
- Such practice have been forbidden since the 1930’s, but kept on proliferating, notably during the dot-com boom. The current scale is surprising.
- Members actually pay $50-250 to belong to a more exclusive group, which may be given preferential information on the name of the asset in exchange for the requirement to advertise the pumps.
- The moderators are anonymous, hidden behind legal and technical screens.
The latest academic article
Multiple academic articles have started analyzing this phenomenon. The most recent is A new wolf in town? Pump-and-dump manipulation in cryptocurrency markets, by Anirudh Dhawan (University of Technology Sydney) and Tālis J. Putniņš (Stockholm School of Economics in Riga). They have reviewed hundreds of crypto currency manipulations, and analyzed the motivations of the participants.
A few facts
There are currently 7,000 cryptocurrencies, which have raised ~29 bn of capital in ICOs, and now represent $100 trn of trading volume per year.
Pump-and-dump manipulations in cryptocurrencies are widespread and account for a substantial amount of crypto trading. The authors identified 355 such cases just in the first 7 months of 2020, involving millions of participants and generating~$350m of trading.
~15% of the coins were affected at least once.
There are 2 pumps per day on average, much higher than in equities at any time.
The organizers pocketed an estimated $6m from other participants over the period.
Pumps-and-dumps are short-lived. It takes 8 minutes on average for the currency to reach its peak.
The asset then reverts to its original price in hours, at worse a day or two.
The volume of trading during the period is typically 13-14 average daily volume.
The currency moves are significant: typically 65% in price or four standard daily variation (of already volatile assets).
The difference with equity pump-and-dumps
- Unlike the old fashion equity pump-and-dumps, the organizers do not pretend to have any insider information (there is no “the mine has found silver!” announcement).
- They do not pretend that the crypto is undervalued whatsoever.
- The goal IS the manipulation. The organizers state this objective very clearly.
- The operation is presented as a game of skills to participants – to profit from the operation, one needs to be fast, and one shouldn’t be last… Organizers still ask participants to delay their sales, so that all have participants have a chance to win.
- Organizers buy the asset in advance of the full announcement. They sell the tokens at advantageous prices, maybe close to the peak. Their return expectancy is therefore strictly positive.
- Since the asset returns to its original levels, it means the other participants have a strictly negative outcome on average.
- Absent a speed or skill advantage, rational individuals would not participate. So why would any other participant take such odds? The authors assume, and statistically demonstrate, that participants:
- Either believe they have higher-than-average technical capacity/skills, thanks to attrition, or are unaware of their weakness. Either way, it is over-confidence.
- or find satisfaction in the thrills of gambling. They may maximize a utility function based on average gains over multiple participations (trading amounts, target gains, stop losses…). Estimates of that function are modelized in the paper and lead to statistical considerations on the scheme’s outcomes.
- Despite the gaming aspects, the scheme is undoubtedly market manipulation; the schemes and the organizers unduely influence the price of securities.
- Their overall outcome is a decay in the integrity of & the public confidence in markets, which will surely inhibit institutional acceptance.
- To optimize their profits, organizers need to strategize:
- Liquidity is both an asset, but mostly a detriment. Too much liquidity and the asset will not move. Not enough liquidity, and the organizers will not be able to buy ahead of the flock.
- Therefore currencies of lower-liquidity become the principal targets.
- Organizers have to be moderate in their pre-pump purchases, so that the other participants have a chance of winning and may want to come back. The organizer’s profits come from the repetition of the operations, not the maximization of each operation’s profits. Repeated participation is key.
- Pump-and-dumps seem to have accelerated over time – earlier peaks, earlier reversals.
About exchanges and enforcement
- Crypto exchanges, still in their infancies, have little to no technology to monitor and eliminate those schemes. They rarely have anti-fraud policies. None have market surveillance. A couple have rudimentary technology.
- They actually have an incentive not to be pro-active – they collect commissions on atypically large trading volumes. See the NY AG’s Virtual Markets Integrity Initiative Report on the many conflicts of interest in the area.
- There is virtually no action from regulators or enforcement agencies to counter these pump-and-dumps., except that the CFTC has staked its jurisdiction on derivatives, manipulations and offered whistleblower rewards.
- Unless the regulators intervene, the manipulations have no reason to decrease. They will actually prevail, threatening the overall benefit of the token-based ecosystem.
- The two factors which will further increase pumps-and-dumps are the participation of the general public in cryptos, and a strong bullish sentiment in the asset class.
The paper’s contributions
Here are the main contributions of the academic paper.
- There is no spreading of false information, like in equity pump-and-dumps. Actually, participants know that there is no information asymmetry. The revealed information IS the information needed for the manipulation. As a result, a lot of research from the equity space is lost.
- Crypto exchanges fill orders in a first-come-first served. Speed of order entry therefore dictates the price at which you execute your trades. Buying and selling latencies are uncorrelated.
Result 1: Rational individuals with correct beliefs do not participate in cryptocurrency pump-and-dumps.
- Additional points: the higher the self-confidence, the less market impact is needed.
- The more self-confidence, the more organizers can load ahead of the pump.
Result 2: Sufficiently overconfident individuals participate in cryptocurrency pump-and-dumps.
- Indeed, the faster the participants, the more profits for them.
- Profits are transferred from slow to fast participants.
- This does not explain why low-skilled or skill-unaware participants still participate.
- Perception of skills is relevant for participation; actually, with a model of overconfidence, the expected outcome increases with over-confidence.
- Less liquid assets tend to attract more overconfident individuals.
- Pumps with more pre-purchase from organizers tend to have less participation from overconfident players; they are attracted to pumps with less pre-pump purchases.
Result 3: Individuals with preferences on cumulative prospects participate in cryptocurrency pump-and-dumps as a form of gambling.
- Gamblers can be modeled as maximizing a utility pay-off function with target gains and stop losses. They are rational under that angle.
- They notably prefer multiple small bets than concentrated large bets.
Result 4: Pumps with more participants and pumps in less liquid coins have higher peak prices and earn manipulators higher profits.
- Participants’ outcome is increased when assets are less liquid, and therefore prefer illiquid currencies.
- Their outcome is reduced when organizers buy more currencies ahead of the pumps.
- The organizer’s outcome is increased when there are more participants.
Result 5: Pumps with more manipulator participation are less attractive to non-manipulators and therefore have less non-manipulator participation.
Result 6: Non-manipulators are more attracted to pumps in relatively illiquid coins.
Result 7: Participation in pump-and-dumps through time increases when the general level of interest in cryptocurrencies increases, when the returns of past pumps are higher, and when there is an increase in market-wide gambling activity.
Other pump characteristics
- In the 355 pumps studied, participants have traded $350m of assets.
- The organizers have pre-purchased about $24.5 m of assets in the two hours prceeding the pump.
- Based on the typical moves, they have made ~6m profit, or a 24.8% return in the space of minutes or hours.
- On average, pumps take 8 minutes to reach their peak (median 1.5 minutes), and generate an average return of 65.5% return (way more than BTC or SPX ever returned for such a short time)
- The average pump return is ~4 StDev of the asset’s average return.
- The pump-and dump represents 13.5 ADV, and 40% of that is in the pre-peak period (~5 AVD)
- Manipulators earn 49% on the average pump.
- Here is the average asset return profile
- Here is the average cumulative traded volume
The interest in cryptocurrencies pumps increases, when the returns of past pumps are higher, and when there is an increase in market-wide gambling activity.
- A 1% increase in participation impacts the organizers’ profits by 0.96%.
- The data is inconclusive on the impact of liquidity on organizers’ profitability.
- Organizers tend to take large pre-pump positions when they anticipate a higher rate of participations – a 1% increase in participation is associated with a 0.44% increase in pre-pump inventory positions.
- A 1% increase in participation is associated with a 0.24% higher pump return.
- Organizers select less liquid pumps, because they are more attractive to participants.
- A good metric for liquidity is the coin market capitalization.
- Pumps are not the major driver for market volatility. The impact of pumps on volatility actually remains small.
Welfare implications
Pumps cause wealth transfers from the least to the most sophisticated players – manipulators.
Gambling is regulated so that gamblers are not abusively taken advantage of by gambling services providers. There is no such control in crypto currencies.
Unfortunately, the least sophisticated players are also the poorest.
Pump-and-dumps harm price accuracy and informativeness of the markets.
You could compare those pump-and-dumps, where the last one in gets hurt, to new micro-Ponzi schemes as well.
Widespread pump-and-dumps damage market integrity. Financial institutions, which are concerned about their reputations, may chose to not associate themselves with such schemes and digital assets as a result.
Regulators also restrict the growth of crypto assets (ETFs) because of the risk of manipulation.
Conclusion of the conclusion
Institutional investors will be wary of being associated with such assets. They will therefore avoid less liquid assets, hence keeping those in the less liquid area and concentrating liquidity in few coins.
Hedge funds could probably take advantage of pump-and-dumps. They could monitor chat rooms and /or detect pump names ahead of the crowd. They would have an execution advantage (there are now institutional quality execution platforms).
But following the pump-and-dump direction would generate legal & compliance risk, without even mentioning headline risk.
On the other hand, they could easily sell the peak of the pump. Since the assets reverts in hours typically, it is less likely that you will remain short at the end of the day, so you could buy-back your naked short during the day. Now we need the confirmation that naked sales are possible in cryptos… Borrows are expensive.
References
- Written Testimony of Chairman J. Christopher Giancarlo before the Senate Banking Committee, Washington, D.C., February 6th, 2018
- CFTC’s advisory resource on bitcoin and crypto currencies: https://www.cftc.gov/Bitcoin/
- CFTC Customer Advisory: Beware Virtual Currency Pump-and-Dump Schemes, February 15, 2018
- Wall Street Journal, August 5, 2018: Traders Are Talking Up Cryptocurrencies, Then Dumping Them, Costing Others Millions
- Office of the New York State Attorney General, September 18, 2018: Virtual Markets Integrity Initiative Report.
- A new wolf in town? Pump-and-dump manipulation in cryptocurrency markets, by Anirudh Dhawan, University of Technology Sydney (UTS), and Tālis J. Putniņš, University of Technology Sydney (UTS); Stockholm School of Economics, Riga, November 17, 2020