Crypto-ML has a Market Index which is key to understanding the general direction and sentiment in the cryptocurrency markets. For those who like a little more behind-the-scenes view of how it works and how machine learning is used to optimize it, this post will dive into some details.
This post comes at the request of users on our Crypto-ML Community. If you haven’t joined in yet, please do so–we greatly appreciate feedback, ideas, and questions.
Prior to going through the details below, I encourage you to read our Overview of the Market Index.
The goal of Crypto-ML trading signals
Before diving into the Market Index, it’s good to understand the other information Crypto-ML provides. Our trade signals are designed to provide entry and exit points (buy and sell) for a variety of cryptocurrency. These signals are generated by machine learning. While not bound by certain time parameters, they tend to be swing trading signals.
Crypto-ML provides both long and short signals.
As most experienced traders know, signals matter in context. That is, it’s good to know the big picture. Specifically, it is good to trade with the market. Or as we say, you should swim with the current.
The goal of the Crypto-ML Market Index
The Market Index seeks to provide this big-picture view. Trades operate in isolation and don’t really tell you where price may be going in the long term. But the Market Index value and trend can help provide valuable insight into these types of questions:
- Are we quantitatively in a bull market or bear market?
- Should I be weighing more heavily toward long trades or short trades?
- As an investor, should I increase or decrease my long-term holdings?
- If price is relatively flat, is there any change in overall sentiment?
Getting access to the Market Index
Data utilized by the Market Index
To help answer these questions, we didn’t have to look far. In order to have the trade signals, we built our own proprietary data set and data aggregation method. When framed in a different way, this same data set is effective at answering big-picture questions too.
While we can’t give away everything here, we can say that price is heavily weighted. As we discussed in posts on our machine learning methods, our research has shown that price effectively encompasses many sentiment factors.
In addition to price and a handful of other data points, we apply complex calculations akin to technical analysis including many basics such as moving averages. In fact, machine learning was heavily used to uncover which data points matter, which technical indicators matter, and what values should be applied to the indicators.
Last, since this is a picture of the market, we blend data from Bitcoin, Ethereum, Bitcoin Cash, and Litecoin weighted by market capitalization.
The end result is a highly-tuned soup of calculations that give our picture of sentiment.
How machine learning is used for the Market Index
Apart from utilizing machine learning to create the initial data set and tune the equations, there is an additional step to formalize a goal and have it optimize all variables to best achieve that goal.
So here is how that plays out. If the Market Index correctly predicts the 30-day direction of the market, it receives a biscuit. If it doesn’t, no biscuit.
The machine learning models then have one simple goal: get as many biscuits as possible.
In reality, we’re giving it a “1” every time it’s right, but we call the database field “biscuit” because personification is fun.
So let’s say the equations decide to increment (increase) the Market Index. If the market moves up over the course of 30 days, then it made the right choice and receives a cookie.
With this mechanism in place, the equations can self optimize and adjust to create a great predictive model
How to read the Market Index
There are a few different ways to get meaningful insight from the Market Index.
Is the value positive or negative? This tells you if we’re in a bull or bear market. This is a lagging indicator. But, if you take long trades during bull markets and short trades during bear markets (combined strategy), you will experience greater trading results (see Trade History).
Is the value extreme? A very high or very low number can be an indication that a reversal is pending. Much like overbought/oversold technical indicators, extreme numbers on the Market Index are difficult to hit. This means a strong positive is indicative of mania and strong negative is indicative of despair. Long-term investors typically want to buy and sell counter to the general market.
What is the slope? If the Market Index is strongly trending, that is an indication the market dynamics are changing. Right now, we are seeing sideways price action but the Market Index is strongly increasing (though still negative). We also have seen areas of support and resistance develop within the Market Index.
Putting this all together, we can look at a couple of situations. The ultra-fast bull market in 2017 resulted in an extreme reading on the Market Index of +69. This occurred about 2 weeks before Bitcoin peaked at $19K and was followed by a sharp slope change.
This extreme reading is indicative of a pending reversal. Having a moderately positive reading during a bull market is a much healthier picture of growth.
With this information in hand, while most people were buying, you could have sold and protected your cash.
Likewise, right now we are coming off of an extremely negative Market Index that peaked around -94 in mid-December. That shows the bear market was, in fact, a more extreme move than the preceding bull market was.
This reading was followed by a slope change. For long-term investors, that may have been a good accumulation point. As traders, we can grow our confidence in taking long trades. As crypto enthusiasts, we can grow our confidence that the value will begin increasing. And we can hope that increase will come in a more sustainable fashion this time around.
Where is the market headed?
Alright, that is the million-dollar question that machine learning seems to be doing a great job of helping us answer.
All indications point to us nearing the exit of a long bear market. While this is not confirmed, the Market Index is trending in that direction strongly. Additionally, on its uptrend (making higher highs and higher lows), it has broken out of the channel to the upside. This would seem to indicate even stronger forces are at play.
Let us know where you think we’re headed in the comments below or in the Crypto-ML Community Forums.
Not a member yet? Sign up for free today.