We can see for miles

Clear Vision Through Glasses


I know you’ve deceived me, now here’s a surprise
I know that you have ’cause there’s magic in my eyes

I can see for miles and miles and miles and miles and miles
Oh yeah

If you think that I don’t know about the little tricks you’ve played
And never see you when deliberately you put things in my way

The Who (I Can See for Miles)

In trading, the ability to instantly understand the price and available liquidity of the asset is critical.  In crypto, however, many participants believe that the market is opaque, and that liquidity on exchanges is minimal.  It is true that many pricing sources obfuscate the situation, and the sheer number of orders on exchanges makes it difficult for traders to see where liquidity is.  At CoinRoutes, however, our aggregation technology consolidates thousands of price levels among all the exchange order books.  This data can often predict price movements, particularly at important inflection points, and also accurately show our clients the true cost of trading.  This provides our clients with a substantial advantage, both in the ability to know the cost of trading and to predict the movement of the market.  This was shown in a video produced last week, where the software was able to show how the market made a short term bottom at just over $4000.  As a result of this, we decided to do a more comprehensive study of the data.

Over the month of November, for example, we analyzed over 2.3 million observations of book data using thousands of price levels to study the cost of trading Bitcoin in each interval. We determined that the data clearly has predictive value, and that it is extremely useful in understanding the available liquidity.  For the purpose of this analysis, we included all exchanges that allow US based clients and have no restrictions on the withdrawals or deposits of US dollars.


Prediction of Future Price Movement

During the month of November, we analyzed situations where the book was imbalanced.  Specifically, we looked at situations for buying or selling 2000 Bitcoins using the CoinRoutes “Cost Calculator” to compare buy and sell trading costs.   (This software “walks” a consolidated order book, constructed of over 2000 price levels per side, aggregated across exchanges.)

Out of 2.3 million individual observations, we looked for cases where one side was either 2 or 3 times more expensive to trade due to an imbalance in the consolidated order book.   In all cases where this was the case, there was statistically significant predictive value of price movements at 1000 and 3000 seconds in the future with the most significant being an average (on over 12000 observations) of 37.77 basis points 3000 seconds after the cost to buy 2000 BTC was 3 times the cost to sell 2000 BTC.

The results are summarized in the following table:


Price Move -3 x
-3 x # of cases
Price Move -3 x
-2  x # of cases
Price Move 2 x
2 x # of cases
Price Move 3 x
3 x # of cases
1000 Seconds
-3.24 bp
-3.25 bp
3.6 bp
13.34 bp
3000 Seconds
-7.68 bp
-7.70 bp
4.42 bp
37.77 bp

As can be seen from the table, the results were consistent across every situation and time horizon.  When the book showed that it was much cheaper to sell compared to buying bitcoin, the price tended to move higher and when it was much cheaper to buy, the price moved lower.  It is important to understand, however, that this analysis required the consolidation of over a thousand price levels in every case.  Thus, systems that do not capture the full depth of consolidated order books would be incapable of providing such a signal.


Trading Cost Analysis

During the same period, across over 2.3 million time periods we analyzed the average cost of liquidity for trading 100, 200, 500, 1000, and 2000 Bitcoin at a time.  As you can see from the table below, the cost to trade such amounts is quite low on average, ranging from under 6 basis points for trading 100 to consistently under 1 percent for 2000 coins at a single instant in time.  This proves that there is substantial liquidity on exchanges, if one has the ability to aggregate and access it.

Quantity to Trade Cost to Buy in % Cost to Sell in % Average # of levels to Buy Average # of levels to Sell
100 0.0582 0.0599 46.38005 51.39254
200 0.115 0.1177 86.38609 93.35267
500 0.2435 0.2461 177.3265 193.4698
1000 0.4281 0.4279 295.2921 323.6506
2000 0.9523 0.9244 630.7214 775.7997


It is also important to note that such trading, while extremely simple for CoinRoutes clients using our Smart Order Router, is quite difficult for the average trader without such a tool.  To buy or sell 2000 coins, for example, requires the knowledge of and ability to trade over 630 or 775 price levels respectively.  Most competing systems do not capture enough data to be able to do so.



No matter what asset class is being traded, it is critical for traders to be able to quickly see and react to available liquidity.  In crypto trading, however, the market structure complicates that considerably, due to the number of price levels and fragmentation of the market.  Using the tools provided by CoinRoutes, however, can enable traders to see where liquidity is available and have an informed perspective on the market.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s