Automatically Managing Cryptocurrency Portfolios - A Real World Example

Buy Early and Hold (HODL)

Both retail and professional traders may buy an asset for a long-term horizon, expecting upside.

Day-Traders and Swing-Traders

These Traders use what is known as Technical Analysis, using a wealth of indicators to assign probabilities of a move in the price of an asset.

They may, develop their own analytics.

Set up Dashboards on a variety of platforms to have a market and asset view of their positions.

Most Traders will manually trade.

Professional Traders and Portfolio Managers

These Traders rely more upon advanced mathematical, statistical and cognitive (AI/ML) techniques to model statistical outcomes. Typically, these solutions will be linked to trade order engines, to automate trading.

Many will use High-Frequency Trading (HFT) to execute as close to current market information - The Information Snapshot.

In other circumstances, Analytics produced by Quants are provided to Traders and Portfolio Managers to determine position changes at a more human and strategic level.

The Key difference between Retail Traders and Professional Traders

If we think in terms of how the professional stock-market traders work, they perform market-close operations and use a host of indicators to manage risk.

We feel that the fundamental difference between cryptocurrency markets and traditional markets is that cryptocurrency markets are 24 x 7.

Our Working Model

After spending significant time in the Cryptocurrency space, we observed a number of heuristics:

• Volatility and Mean Reversion seems to permit accumulation of assets if limit orders are created appropriately
• In many circumstances, the market moves against your current Limit Orders, resulting in orders that don't fire
• Sometimes, we see consecutive hourly, daily percentage moves to the upside or downside
• Significant moves, for us, can signify a good entry and exit point
• We focused less on charting and more on moves and recent trends
• We spent minimal time day-trading and more, reacting to executed trades determining whether to reverse the limit orders or do elsewise

Manually Trading on our Heuristics

This mental model largely worked for us, day-trading was not a significant drain on time.

Our Working Conceptual Model

Once we have our conceptual model, we can decide when we want to:-

• Intervene Manually

• Rely upon Automation

Furthermore, we can

• Set up automation to work on Market Events

• Set up monitoring and alerts to aid manual intervention

Portfolio Types

Trading Portfolio

Where our specific assets and rules are executed to our schedule

Banking Portfolio

Where assets are moved to on a set schedule, potentially based upon market conditions via our rules

Funding Portfolio

Under certain market conditions, we may move certain assets from Trading to Funding or from Funding to Trading

Assets and Rules Selection

We decide:

• The assets we trade between
• The conditions these trades can execute
• The assets that are funding and banking assets

What are we looking for when testing our strategies?

This will be covered more extensively in a focused document.

In principle:

• Do trades fire based upon expected market conditions?
• Are swing trades occurring?
• Are balances moving?
• Are our limits (blocking) preventing over-firing of trades?
• Are metrics and audits confirming our Trading API and DataMart is acting as expected?
• Do we need to refine our rules?

Testing the Strategy

It is about seeing things happen in the real world.

We:

• We configure distinct sub-portfolios directly inside the exchange and permission these portfolios
• We place small amounts of assets into the portfolio
• We create our strategies and trade rules within our software
• We test, analyse and evaluate outputs
• We check the Trading Exchange