Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Crypto Wash Trading

arxiv.org

Post reply on HN
Print view View original

221–230 of 306 posts

Re: Crypto Wash Trading

#221

Post by bluecalm »

solveit wrote: post #156
dustingetz wrote: post #141

Earlier quoted context omitted.

You don't even need to detect it. Name a single instance in human history where an unregulated and unaccountable industry didn't instantly degenerate into fraud. "2.5T" "dollars" (majority held by early insiders) is an incredible moral hazard

"I'm sure there is a lot of fraud" is very different from "This is the type of fraud that is going on and here is the evidence".

The difference matters for legal proceedings but not for much else, surely not for personal investment decisions.

Re: Crypto Wash Trading

#222

Post by bduerst »

JumpCrisscross wrote: post #158
tw600040 wrote: post #146

Earlier quoted context omitted.

Huh. What if 2 people decide to do it?

> What if 2 people decide to do it? Two accounts trading back and forth will light up anti-spoofing tech from the 1980s. Keep in mind that the AML regulations Coinbase follows are specifically designed to catch fake money movement.

Anti-spoofing tech for regulated securities. There's no rules for crypto to suggest that Coinbase needs to (or would) flag a group of individuals doing this.

Re: Crypto Wash Trading

#223

Post by bluGill »

MuffinFlavored wrote: post #208
gillesjacobs wrote: post #118

This paper jumps the gun. Detecting wash trading by examining distributions over rounded order prices is a strong and dubious claim for which they provide little evidence. The author's equate wash trading to non-rounded, clustered prices which really just indicates automated trading. Now automated ("bot") trading is a technology needed for exchanges wash trading sure, but not exclusive evidence of it. Automated tradi…

> Automated trading strategies (e.g., "grid trading") are really popular Why? After buying and selling side fees, is it easy to make a profit in an automated way with crypto?

I don't know, but it seems to be. In general though, if there are easy profits to make it is because someone on the other side is willing to lose "a small amount of money" for a long time knowing that when things change they will make a ton of money fast, on the backs of all the people they lost to.

For starters, if you can make money in automated trading, why would you not make all the possible money yourself instead of telling other people (or writing the software) and splitting the profit? This question is one I ask anytime someone mentions easy trading, and because I've never heard a good answer I stay away.

To do better than average is possible, but it requires a lot of deep study.

Re: Crypto Wash Trading

#224

Post by wombatpm »

MrMan wrote: post #179
ur-whale wrote: post #167

Earlier quoted context omitted.

> Benford's Law is a pretty established method of detecting fraud in forensic accounting Benford law never had to deal with exchanges where any customer can write a python bot and start trading at sub-second latencies via the exchange API's. I'd be very surprised if whatever statistical model they're relying on is in any way a match for what real (as in: legitimate orders from actual customers) trading goes on on cry…

Python ... Subsecond latencies Ok

Bitcoin transactions with sub second latency

Double ok

Re: Crypto Wash Trading

#226

Post by capableweb »

greatjack613 wrote: post #127

I have used sniper software such as https://cmcsnipe.com/ and the ease of use of web3 has allowed automated trading to be taken to the next level. Not surprised that so much fake volume exists when it is so easy to create.

This sounds like spam, but is somehow not flagged (yet?). This is from the website:

> CMCSnipe uses insider information to know minutes before a coin is going to be listed on CoinMarketCap or CoinGecko.

How is this website relevant to wash trading?

Re: Crypto Wash Trading

#227

Post by JumpCrisscross »

bduerst wrote: post #222
JumpCrisscross wrote: post #158

Earlier quoted context omitted.

> What if 2 people decide to do it? Two accounts trading back and forth will light up anti-spoofing tech from the 1980s. Keep in mind that the AML regulations Coinbase follows are specifically designed to catch fake money movement.

Anti-spoofing tech for regulated securities. There's no rules for crypto to suggest that Coinbase needs to (or would) flag a group of individuals doing this.

> Anti-spoofing tech for regulated securities. There's no rules for crypto to suggest that Coinbase needs to (or would) flag a group of individuals doing this.

It's the same stuff that catches money laundering. Coinbase isn't exempt from anti-money laundering laws. (With respect to Coinbase not being subject to the Exchange Act, that's very much an open, if irrelevant to this discussion, question.)

Re: Crypto Wash Trading

#228

Post by Cypher »

70% seems a tad bit high.

I think their expectation that real traders would use rounded numbers overlooks that crypto is hyper fractionalized. If someone is exiting their Doge position they're not going to use a rounded number as fee's are paid in a % of that crypto.

Re: Crypto Wash Trading

#229

Post by krisrm »

KarlKemp wrote: post #203
gillesjacobs wrote: post #118

This paper jumps the gun. Detecting wash trading by examining distributions over rounded order prices is a strong and dubious claim for which they provide little evidence. The author's equate wash trading to non-rounded, clustered prices which really just indicates automated trading. Now automated ("bot") trading is a technology needed for exchanges wash trading sure, but not exclusive evidence of it. Automated tradi…

Their idea does show a pretty stark difference b/w "regulated" and "unregulated" exchanges, however, as would be expected if that activity is fraudulent in some way.

I'm no fan of crypto currencies (as my comment history will show), but does "different" imply "fraudulent" necessarily? or are there other factors at play? eg. easier APIs, more variance in crypto prices so that an automated trading strategy becomes potentially more lucrative?

Re: Crypto Wash Trading

#230

Post by dqpb »

sonthonax wrote: post #20

Basically the lesson here is: don't underestimate how stupid HFT algorithms can be, especially when there isn't really a penalty for doing this. I've worked at an above board HFT with a big crypto desk, and this happened constantly.

> especially when there isn't really a penalty for doing this

Executing unprofitable trades is its own perfect penalty, is it not?

Post reply on HN