This article describes bot strategies that I have attempted or plan to attempt with regards to trading cryptocurrencies in the automated sense.
- “MoonRise” – Moonrise is one of the simplest strategies that I have attempted and is my only currently active strategy. It is based on the understanding that the market only offers us ONE future-looking signal: The Order Book balance. The Order Book, as we know, is essentially a bulletin board that lists who is willing to buy and sell and at what prices.Using the order book balance, we can simulate what would happen if we took a considerable amount of funds… say the typical volume traded in 30-minutes and ran it through the order book as it currently is. In doing this, we can guess whether the price would be forced up or down a greater percentage.In the MoonRise strategy, that is exactly what happens.
We perform a “slippage test” using a a 30-minute (subject to change) typical trading volume by first buying, then the selling the amount we were able to buy. Then we weigh the slippage against each other to create a signal from the ratio with possible values from 0.0 to 1.0. For example, If buying the asset drives the ask price up 5% but selling it drives down the bid price by 15%, then the signal is 0.25, because of the total movement (20%) just a quarter of it was in the upward direction. If the books are totally balanced, then the value is 0.5, if the sell price goes down 0%, then the signal is 1.0.
But there’s more to this. It turns out that some coins don’t respond as well as others to the order books. High capital coins like ETH might not be phased much by order books at all. This is why we keep a price and signal history of EACH coin individually, and we train the computer to learn which signals yield the best results. For small-cap coins, the computer will learn that a signal of 0.6 is likely to yield maybe 1-3% over 10 minutes, where as large-cap coins like ETH might only yield 0.1%, not even covering commision. The computer will learn to stay away from ETH automatically, while chasing down gains from smaller cap coins.
The second part of this test is an “action” test, which is basically a volume test but a little different. The action test basically answers the question, “is this coin being actively traded right now, or has everyone left the room?” Some times coins see bursts of action for 2 minutes, 30 minutes, or 3 days… and then eventually all the players leave the room. The action test tells us when the fun is starting and when the fun is over. Right now I use the benchmark that 100 trades must have been made in the last 10 minutes. Subject to change, obviously, but it is an important thing to look at, because the primary signal, the slippage test signal, is better correlated when the action is hot… and when people leave the table, they tend to sell their chips, signalling that it is time for us to get out too.
- “MoonDip” – Is a strategy I once tried, but never quite got working right. Although now that the “MoonRise” strategy is more or less working, MoonDip is really just a slight variation of this. MoonDip relies upon a sudden upward price jump as its go signal. These signals are easy to pick out, and crypto-trading slang, this is referred to as “Mooning”. Many small-cap coins are subject to the infamous “pump and dump” schemes. The “problem”, if you consider it a problem, is epidemic. This is where online gangs of traders get together to promote “pump” a coin that they’ve bought into just so they can “dump” the coin onto unsuspecting sheep. Some of those sheep win enough that they keep coming back for more. A “pump” signal is very visible on ticker charts. After a “pump” occurs, the stragglers and latecomers typically “buy the dip” that occurs as the price falls back to normal. The trick is, what part of the dip should you buy? And when should you sell what you bought? Typical signals often show a recovery after the price dips below 30% of the peak, with a price rise to 60%, but it tends to be different depending on the influence and buying power of the gang. Some pump and dumps have been seen to drive the price up 100% only to crash immediately back down to 0% in as little as 2 minutes. These characteristics make this kind of strategy risky, but if that 30-60% range covers a substantial profit, it might be worth it.
- “Churn and Burn” – Is a strategy I was running for a while that seemed pretty safe, but ultimately I stopped using. The idea behind this strategy is to buy a hugely diverse portfolio of coins that are at a low (I use a 30-day low) and sell them at a high (I used a 4-day high). I first started by buying super-small stakes in any coins that met the criteria, then once I determined how many coins would typically be open at one time, I optimized it to spend a bit more to accelerate growth. The problem with this strategy is that your funds are tied up in assets for a long time that ultimately are toxic. Virtually every alt coin on the market is a toxic asset by nature compared to bitcoin. Although those coins have good days, good weeks, good months, or even good years… their price increases are not likely to consistently outpace the price increase of bitcoin itself, meaning that virtually all of them will fall over time… and when you make bets in this manner, you’re essentially betting on the long game. Furthermore, when bitcoin has a GOOD day… you lose. There were days when bitcoin had a good day, and every alt-coin asset I had bought took a small dive. It would have been better for me to be holding bitcoin instead. Eventually I put in a rule that would ban new trades if bitcoin was “mooning”, but when I looked at the aggregate interest rate I was achieving under this scheme, it was low enough that I wanted to try something new to accelerate profits. Furthermore, without stop losses, losing assets would start to pile up, limiting the funds available for new trades. This is one reason I moved forward with the “MoonRise” strategy, because quick ins and outs meant that funds would not be tied up for long, so I could bet the entirety of my portfolio and make a few percent in a few short minutes.
- “Split Shift” – is a strategy where you look for gaps in the orderbook that you can immediately exploit. When the orderbook spread often 4-5%, for example, you might be able to get away with buying up new offers that are in the middle of that spread and then selling them at the just below or at the current ask that results after you bought it. For example, if the orderbook in the market for Oranges has one guy offering 50 oranges at $1 while all the other guys are selling their oranges for $2 or more, then buy all the guy’s 50 orange at $1 and offer them at $1.99! The risk is that there’s a chance nobody will buy them at $1.99. These kinds of holes don’t open up in the market very often however and usually they are open for a split second.
- “Zig Zag” – is a strategy that I plan to implement that has a human aspect to it. Essentially an administrator, analyst, maybe one from the community, submits buy and sell targets together and the bot simply watches the market for the right time to buy, then subsequently puts in an immediate sell order. This way a human can go through dozens of coin charts, look at all the analysis and plan buys and sells ahead of time.
- “Pick of the Moment” – Essentially the bot could be tied to a twitter account which announces its picks AFTER it buys a particular asset. If the twitter account gets enough followers, the followers are surely to drive the price up, at which time the bot sells it.