Introducing the Fishiness Indicator for Derivative Trading Volume

Perpetual and Futures are an integral part of today’s crypto markets and most serious spot exchanges by now offer these products as well. Just like with spot markets, traders tend to look at 24h volumes as a proxy for liquidity. While the claimed trading volumes for derivatives seem to be more honest than claimed spot volumes, the reason might simply be that complete fake exchanges still mostly focus on spot. On a second look, claimed derivative volumes seem quite odd, too. I hereby introduce the **Fishiness Indicator**for derivative trading volume as a metric to measure if claimed 24h trading volumes look reasonable or odd and define it as the **claimed 24h trading volume divided by the open interest**. The rationale behind this is that while it is quite easy to fake trading volume, open interest is harder to fake (although far from impossible). Not surprisingly, the Fishiness Indicators of claimed volume to open interest differ strongly between exchanges from 0.19 for Bitfinex to 8.19 for Binance (see figure). There are **three legitimate reasons** for why such ratio might differ between exchanges: 1) **Fees**: The lower the fees, the higher the trading volume. The less it costs to trade, the more is traded, since traders can chase smaller edges. The best example for this when “back in the days” trading on Chinese exchanges was charged 0% both for maker and for taker trades. CNY trading volume was then reported to be 98% to 99% of global BTC trading volume (exchanges made their money by lending fees for margin trading). Some derivative exchanges have lower fees than others; this potentially explains a part of the difference in ratios. 2) **Presence of gamblers**: Gambling traders trade with a higher leverage, get liquidated more often, and more often switch between positions. This causes, on average, more trading volume per open interest than professionals, who just open one position to hedge and then hold it until expiry or drastic changes in market conditions. 3) **Volatility of premiums and funding rates**: The more volatile premiums on futures and on funding rates are, the higher the incentive to switch positions back and forth to react to these changes. Since the volatility of premiums and funding rates differs across exchanges (something to look at in more detail in a future pulse) this can likely explain part of the observed difference. While the difference in the ratio can certainly be explained in part by these legitimate reasons, I strongly believe that other, “fishy”, reasons also explain part of the differences. These relate to either faked or manipulated trading volume. Let’s first look at **naked volume faking**. The most basic way to fake volume is to simply report more volume than is reported in the trade history. A little bit more elaborate, and much more common, is to print false trades between the spread. This can be done in a stupid way (steady over time, identical trade sizes or identical trade size distribution) or clever (mimicking actual trading volume in regards to time and order sizes). Basic and stupid volume faking can easily be observed by watching the order books and trade flow. Spotting clever volume faking is harder, but can be detected by looking more closely and using some tools (e.g. [https://github.com/Whalepool/Bitcoin-Volume-Validator](https://github.com/Whalepool/Bitcoin-Volume-Validator)). I thus believe that naked volume faking does not play much of a role anymore (except for pure fake exchanges). More likely seems to be **manipulation of trading volume. **This is actual trading volume but conducted** **via **Internal Trading Desks** (ITC) that pay zero fees (either directly or indirectly by being owned by the exchange). These desks create a liquid market with many executed trades and much liquidity in the order books. While this looks and sounds good it is actually quite bad for traders and much worse than naked volume faking? To see why, let’s break such volume down into two categories: a) trades between internal trading desks and b) trades between an internal trading desk and a customer. Trades of category a) might look similar to naked fake volume, but are worse: they take liquidity from a regular trader that now either cannot execute his trade or only at a worse price. Such volume thus does not only create the false impression of liquidity in the trade history but also in the order books. Trades of category b) actually look positive on the surface: Traders get executed at potentially better prices than what other regular traders would accept (since ITCs effectively pay 0% fee on their trades and can thus offer better quotes). Traders should be happy about ITCs – after all, they give them liquidity, no? No. ITCs do not exist to lose money to traders. Instead, they are armed with the best tools and best traders available and trade at the lowest latency. They will only execute trades that have a positive expected value – which means that the other side has a negative expected value for his trade. Thus, if a trader gets executed against an internal trading desk, he can expect it to be a losing trade. But such desks do not only execute the losing orders of regular traders but also prevent their winning orders from being executed: since ITCs pay zero fees, a trader can expect such desks to snap any profitable trading opportunity before him. Think of ITCs as the odds maker in sports betting: you might win against them in the short term, but will always end up **losing against the house in the long run**. I want to explicitly exclude any exchange activity that is not aimed at profitable trading by exploiting their customers but simply at channeling or mirroring liquidity across books. This cannot, by definition, exploit traders to make extra money, but is instead designed to return zero profits. Manipulation by ITCs cannot be spotted by observing order books or the trade history. Instead, other metrics are needed. One idea would be to track external indicators like website traffic. While this approach is promising, it comes with some limitations and a lot of workload. I thus propose the Fishiness Indicator as a shortcut to identify potentially manipulative exchanges where traders should be cautious. Knowing that such an indicator is not precise science I suggest to be cautious, when an exchange has a Fishiness Indicator value of 3 or larger and be at the lookout at a value between 2 and 3. The limitation of the Fishiness Indicator is that it assumes that volume can be easily faked but open interest not. This is a strong assumption, since open interest can also be faked, either directly by reporting a wrong figure or by having internal trading desks build positions that offset each other. Hence, a low value of the indicator does not imply that there is no volume manipulation. Actually, with so such an indicator becoming public knowledge, I predict that we will see much lower values in the future, when manipulative exchanges decrease it by increasing their reported open interest. This logic could actually be turned upside down: If, such the Fishiness Indicator of an exchange is high today but low in, say 6 months, without the reported volume having changed, it is a good sign that the exchange is manipulating the trading volume. One thought on the very low FI of Bitfinex: It is quite possible that their derivative products are mostly used by professionals that want to hedge positions but do not change these once built. The very low volatility of the funding rate (usually it is 0%) supports this view since these traders aim to keep their hedging costs close to zero. ​