In most asset classes, the larger the order size, the more market impact. This has been true in crypto more than most, with some spectacular examples that have caused huge price moves in the Bitcoin chart. Part of the reason for this has been the structure of the crypto markets.
Over the past 3 years, we have often asserted that crypto trading resembles the way that equities traded in the City of London or on the Nasdaq desks in the late 1990s, with most trades consummated by calling dealers. Admittedly, in crypto, rather than the phone, the communication method has been Telegram/Skype/Etc., but that is finally starting to change. More investors are trading electronically, using algorithmic tools on multiple venues. CoinRoutes specifically, has seen significant increases in adoption as proof of that trend. CoinRoutes clients have executed over $2 billion in notional value year-to-date with over $430 million traded in July alone.
Of particular significance, however, is the performance of CoinRoutes algorithms, particularly recently. We studied our clients algorithmic trades and found that the performance was extremely good, with data that shows almost no market impact. The following table shows the execution performance vs the consolidated midpoint price on all orders during the June/July study period in Bitcoin-USD that were executed via our “SmartPost” algorithm without time constraints. The sample includes 4135 client orders that executed 27,346 Bitcoin with a notional value of over $270 million.
The headline numbers, which show that the algorithms traded better than the midpoint price before fees and only 3.4 basis points worse than the midpoint after paying exchange fees are impressive. More impressive, however, is that the slippage numbers were consistent across size buckets (proving that size doesn’t matter), showing a standard deviation of roughly 10 basis points across the 4135 orders. It is impressive, because the only way such a result is possible is if the market was completely unaware of the CoinRoutes orders. Considering the negligible market impact created, it seems pretty evident that the anti-gaming logic embedded in the algorithms worked well, without sacrificing its ability to find the necessary liquidity to complete those orders. To underscore that point, consider the following table which compares the same orders to hypothetical “smart routing” against the same exchanges used by the algorithm.
This shows that net of fees, CoinRoutes algorithms saved our clients roughly 27.8 basis points, which translated into over $750,000 in savings, for this particular sample that includes only Bitcoin-USD. It is worth noting that CoinRoutes algorithms tend to show larger savings for less liquid crypto pairs, so these numbers understate the potential benefit of algorithmic trading for those investors that trade a more diversified portfolio.