Benford's Law is a pretty established method of detecting fraud in forensic accounting.
Imagine you have a completely manual market which follows normal statistical distributions, including Benford's Law. Now you introduce one grid-trading bot with a large amount of capital. (A grid trading bot is basically a piece of software that automatically buys when the price falls below a certain level and then automatically sells when it goes above a certain range.) That one bot is going to make up the vast majority of transactions, because it's effectively "clamping" the price within its trading range. When random fluctuations take it below, it buys and sets a floor on the price. When random fluctuations take it above, it sells and sets a ceiling on the price. If you make the range small enough that most ordinary trades would end up occurring with the bot, it's going to take up the vast majority of volume.
There's nothing illegal about grid trading. They work to dampen random price fluctuations in a market. In exchange, they take on the risk that fundamental supply & demand might shift enough that they're left holding the bag, eg. they run out of inventory to sell and then the price jumps sharply higher, or they collect all the inventory and then the price drops.
But because they're non-human and take one side or another of most trades, they are going to account for a disproportionate amount of volume. This isn't fraud, it's that you don't understand the structure of the market.
"Abstract. Is Benford's law a good instrument to detect fraud in reports of statistical and scientific data? For a valid test, the probability of ‘false positives’ and ‘false negatives’ has to be low. However, it is very doubtful whether the Benford distribution is an appropriate tool to discriminate between manipulated and non-manipulated estimates. Further research should focus more on the validity of the test and test results should be interpreted more carefully."
1. https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1468-0475....
Here [2] is Google Scholar on Benford's Law. Of the 36,000 papers it pulled very, very few claim the law is not useful and valid. The vast majority (actually, every one of the first many pages) show how it's useful and demonstrate uses of it. If you want only recent papers, select from the left panel. Same result.
In fact, the first several pages of results contain many papers showing empirical validation of the usefulness of the law.
[1] https://www.degruyter.com/document/doi/10.1111/j.1468-0475.2...
[2] https://scholar.google.com/scholar?hl=en&as_sdt=0%2C14&q=ben...
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 crypto exchanges.
Ok
Lol, fair enough.
But I believe my point stands even if you go to python-level latencies.
Double ok