I Built a Crypto Strategy SHITCOINER - currently in production on very large capital.
BACKGROUND
I’m a Python programmer by profession. Eight years ago, I quit my job and started building crypto arbitrage bots. In the beginning, even the most primitive script running across exchanges could be profitable. Then the market became increasingly professional, and I had to look for more and more complicated and sophisticated algorithms.
About a year ago, I shut down all my arbitrage bots because they simply stopped working, and I was no longer able to come up with anything new. I did make good money from them, though, and essentially without taking much risk (delta-neutral arbitrage).
I started looking for a different approach, this time with actual market risk on perps. I combined several ideas that had occurred to me over the years.
I should point out upfront that I don’t use, or even believe in, technical analysis - especially when it comes to crypto.
EDGE
The main thesis behind the strategy is that ~100% of tokens are shitcoins with no fundamental value beyond speculation, driven almost entirely by retail greed.
Over the long term, every shitcoin will lose most of its ATH - it’s simply a matter of time. The downward gravitational force on price is further amplified by continuous inflation (emissions, unlocks) and market dilution from constantly listing new shitcoins.
ALGORITHM
In very simple terms, we take the top 50–200 tokens by market cap, with a primitive filter requiring them to be at least 1.5 years old. The goal is to minimize the risk of a speculative 10x–50x pump in a newly launched asset.
The position is built using limit SELL orders, but the grid is dynamic and designed to target mean reversion.
Each token has a predefined maximum dollar position size up to which the strategy is allowed to average in. During each loop cycle (typically every 1–2 hours), the algorithm places a SELL order for each token representing a fraction of its target position size.
Typically, we place the order at a price 3–6% above the current market price. If the order doesn’t fill, it is cancelled during the next cycle and replaced based on the current market price and current total equity.
In other words, we are trying to capture local spikes and betting on a retracement.
The basket of shitcoin shorts is hedged with a BTC long at roughly a 1:1 ratio. The BTC long is primarily there to maximize margin efficiency and mitigate the effects of altseasons.
Profit is realized gradually whenever reducing the token position plus the corresponding portion of BTC results in a profit. Typically, we start reducing the position once the combined P&L reaches around +2%.
If a token position exceeds its target value, it is cut using a market order at a loss.
The parameters are slightly adjusted in real time based on a hype_score (Google Trends, volume, price momentum). The algorithm switches into defensive mode as the market starts heating up.
The strategy currently runs across 20 instances, each slightly differently tuned for diversification.
RESULTS / RISKS
The strategy will blow up during an extremely strong altseason - something like 2021: COVID + stimulus + a young market, where literally every shitcoin was doing 10x every month.
Individual token explosions will reduce profitability, but they are not necessarily a threat given how broad the basket is.
I built a very detailed exchange-engine model to simulate execution down to 1-minute candles.
Based on numerous simulations, I’m targeting a CAGR of >150–200% with a maximum drawdown of 50–60%. I put $2.1M into production (21 instances per 100k).
Obviously, the potential risk/profit profile can be reduced, but personally I believe crypto altseasons will become weaker over time, so YOLO.
The main backtests start from 2024. Going further back doesn’t make much sense because there simply aren’t enough old tokens to construct the basket properly.
Funding fees eat up around 10–20% of the profit.
The high drawdown is unfortunately an expected consequence of the strategy during altseasons (12/2024, 07/2025).
Under the fundamental assumption that we don’t include very young shitcoins, it is difficult to find a set of parameters under which the strategy generates a loss.
I also ran stress tests by drastically reducing the token-age filter to just a few months. The strategy still produces high returns, but with an even deeper maximum drawdown.
So far, the live production results are consistent with the model, although the program has only been running for a few months.