- Smart Crypto DCA, A Proof of Concept
Dollar-Cost-Averaging (or DCA) has enjoyed a growing popularity among retail crypto investors. And why wouldn’t it? After all, for not being necessarily optimal as we will see it, this investment scheme offers at a minimum a frame of reference for the casual or undisciplined investor to grow a stake with consistency. However, my quant mind couldn’t help but suspect this method to be overwhelmed by randomness, potentially opening much room for improvement — which indeed there was.
As it turns out, I found a way to generate substantial savings i.e. decrease the weighted average realized cost basis, while respecting the spirit of DCA in terms of simplicity and low involvement. On what basis I found it is beyond the scope of this Medium article. But for those quant buffs out there, more trading-related information can be found in the following original video published on YouTube on October 25th, 2021 titled “Smart Bitcoin DCA: Beyond Random.”
In a nutshell, I leveraged my proprietary mathematical framework to narrow down those fields of motion dynamics that best describe the price movements of cryptocurrencies. I then attempted to capture this motion statistically with a handful of relevant metrics and indicators. The goal was to materially beat the average cost basis that a regular/blind/random DCA would generate. But most importantly, its parametric rules should not be optimized and should be portable with similar or greater average success across various crypto instruments without requiring any further tweaking. And lastly, because filtering those specific buying opportunities would by definition reduce the number of investments versus a regular DCA, a specific money management scheme would have to be devised so that, in the end, the exact same amount of dollars would be invested with our Smart DCA.
Once again, a core requirement was that the Smart DCA model should never become a frantic convoluted system that would completely depart from the DNA of DCA which in its purest form is mostly a hands-off approach. For that reason, I stuck to a weekly time frame, with close-to-close readings. I started with Bitcoin (BTC) for its longer price history and overall best relative liquidity. I then expanded the model to a few altcoins which I laid out in a second original video published on YouTube on November 18th, 2021 titled “Smart Bitcoin DCA: Expanded to Altcoins.” The crypto market being fairly young, I had to restrict this expanded analysis to a limited number of altcoins that offered a sufficiently decent statistical history and reliable liquidity, namely and in no particular order: Ethereum (ETH), Litecoin (LTC), Monero (XMR), Ripple (XRP), Cardano (ADA) and Binance Coin (BNB).
With three simple degrees of freedom, the final model managed to generate staggering savings on BTC well in excess of 40% by itself, and easily topping 80% when combined with the money management over its entire available price history. By blindly porting the exact same parameters over to altcoins in order to avoid curve-fitting, comparable average proportions were observed with some coins such as ETH and XMR even surpassing the BTC benchmarks.
It quickly became clear, however, that even though Smart DCA was a tremendous improvement over regular DCA, the magnitude of these savings was a direct positive function of the length of the price history and the appreciation of these coins. For that reason, and to remove any remaining doubt of a lucky outcome due to sampling bias, I normalized these coins’ time series by starting from December 2017. Why December 2017? Because starting to DCA in right before a decline is mathematically the best timing for a regular DCA, with ‘best’ being defined as ‘favoring a minimum expected realized cost basis’. My intention here was to stack the odds against Smart DCA by having it compete with regular DCA when the conventional method is at its performance peak. The results were equally impressive. Over this shorter study period, Smart DCA generated average savings of 36% by itself and of 50% when combined with its specific money management. Here is a recap table of the results.
Copyright 2021 © FX Physics. All Rights Reserved.
Limitations obviously apply to this model, if only for the facts that:
(1) its continued validity would assume that the price motion dynamics of cryptocurrencies would not change, most notably as relates to their volatility but irrespectively of their rate of appreciation.
(2) the weekly time frame limits the purely statistical significance of this model, at least if we omit its being grounded in fundamental mathematics.
(3) its logic is rooted in the ‘HODL’ cult which underlying ‘buy-and-pray’ investment logic assumes zero risk management throughout the entire life cycle of the investment.
It is worth noting though that unlike most so-called ‘trading strategies’ presented on social media, this one has the merit to be objective, quantified, replicable and trustless. In fact, you are strongly invited to verify it yourself. You may find it interesting that Reddit demoted my model post by tagging it as “Low Quality” (see below), all the while letting clickbait subjective nonsensical analysis such as price action and chart drawings proliferate and prosper while these methods have been consistently proven to drive retail off a cliff. If anything, this should serve as a wake-up call that these platforms are incompetently discriminating content at the expense of their users.
The only takeaway is this: if you can’t test it quantitatively, then it does NOT work.
You were warned. Trade carefully.
- Date of publication:
- Thu, 11/18/2021 - 16:09
Click on the link - it will be copied to clipboard