Why Algorithms Decide How Crypto Prices Move
Cryptocurrency prices don't just move randomly. Every trade and exchange impacts price on a 24/7 global market. Increasingly this process is being driven by algorithmic strategies. Algorithms aren't just setting prices, they're deciding how those prices are set.
Market-making bots change prices every millisecond. Arbitrage bots connect prices between exchanges. DeFi liquidations occur when oracles refresh. The point is that today's cryptocurrency prices are formed by instantaneous automated feedback loops. Human manipulation isn't always the cause behind price swings, but computers now dictate market volatility and liquidity.
Here's how the system functions and what it means to traders, long-term investors and anyone else asking why the market can be tranquil one minute and turbulent the next.
Crypto Pricing Is Fragmented and Algorithms Are the Glue
Instead of trading on one exchange like traditional stocks, cryptocurrencies trade on multiple exchanges, each with their own distinct order book, rules, and users. A Bank for International Settlements study finds that cryptocurrency markets are fragmented and demonstrates how price discovery can vary across exchanges due to information flow and frictions.
This fragmentation creates both opportunities and needs for algorithms:
- Arbitrage bots look at the same asset on two different exchanges and trade when there's a pricing opportunity.
- Routing systems choose where to execute based on liquidity, fees, latency, and slippage.
- Cross-venue market makers hedge inventory risk by hedging against one venue and quoting on another.
What appears to be one market price is actually consensus derived across thousands of smaller markets. If that consensus collapses due to an outage, liquidity shock or rapid risk-off moves, prices can move quicker than human reaction time.
Market-Making Algorithms Tighten Spreads Until They Stop
Market makers are firms, increasingly bots, that continuously quote buy and sell prices. Under typical circumstances automated market making provides more efficient trading through:
- Tighter bid-ask spreads
- Deeper order books
- Faster execution
True, but only under normal conditions. Algorithmic liquidity will often only arrive when the model's risk limits allow. Many systems will respond similarly to market stress: wider spreads, smaller orders or no quoting at all.
The Liquidity Mirage
The term "liquidity mirage" explains sudden drops or spikes in crypto prices. It happens when prices change rapidly due to a thin order book right when a lot of traders want to buy or sell.
A 2024 paper examining fragmentation and liquidity in Bitcoin finds that liquidity and market structure can interact in destabilizing ways and can lead to markets suffering from episodes consistent with flash-crash dynamics in fragmented venues.
Translation: if you're watching just the last traded price you're only seeing part of the picture. Often liquidity (depth and spread) will erode before a significant move shows up on the chart.
Volatility Clusters Are Often Algorithmic Feedback Loops
Cryptocurrency is famously volatile. Less obvious is how often volatility arrives in clusters: long periods of relative quiet interrupted by frenzies.
Algorithms can exacerbate this effect because many are programmed, explicitly or implicitly, to respond to similar signals:
- Order-book imbalance
- Short-term momentum
- Volatility breakouts
- Funding-rate shifts in perpetual futures
- Liquidation heatmaps
Enough systems see the same "go" signal, and they all buy. Enough risk engines see the same "stop" signal, and they all sell. This is why cryptocurrencies can sometimes experience large price movements seemingly absent of new market or fundamental news. The catalyst can be structural, reaching a certain level in positioning, liquidity, volatility, or similar metrics.
Reinforcement Learning Delivers a New Twist
Algorithmic trading strategies need not be rule-based. Increasingly firms apply machine learning techniques such as reinforcement learning (RL), allowing agents to learn over time to optimize strategies that will yield the most reward.
A July 2025 NBER working paper provides insights into the relevance of today's markets. RL driven trading agents are capable of independently producing supra-competitive collusive outcomes without communication or explicit collusion agreement. These outcomes harm competition and price efficiency, as demonstrated in simulations.
This doesn't necessarily imply crypto markets are constantly "rigged by AI." It does highlight a new form of manipulation:
- Harmful outcomes don't always require a mastermind.
- Interaction rules and incentives can produce system-level behavior that looks coordinated.
- This differs greatly from traditional manipulation which most often presupposes known actors with intent.
Derivatives and Liquidations Turn Small Moves Into Big Ones
Spot markets also play a major role but much of crypto's short term trading volume is dominated by perpetual futures and other derivatives. Leverage is commonly used by both humans and trading strategies. Leverage against crowd favorite positions can cause dramatic runaway price moves in the opposite direction through forced liquidations.
This is where algorithms play two key roles:
- Positioning and risk engines automatically reduce exposure when volatility rises or margin tightens.
- Exchange and DeFi liquidation bots automatically sell or buy upon reaching a collateral ratio.
In other words there is built-in "if-then" selling pressure in the market, and it can transmit rapidly.
Oracles Are Not Just Observers, They Trigger Events
Prices don't just signal information in DeFi, they act on it. Lending protocols, liquid staking solutions, and derivatives exchanges all rely on price feeds to gauge collateralization levels and trigger liquidations.
Oracle network Chainlink is one of the largest in existence. Their Data Feeds work through mechanisms such as deviation thresholds and heartbeat time. Price updates therefore occur in intervals, not continuously.
That design makes sense for reliability and cost, but it also affects the market:
- If price updates lag fast-moving markets, liquidations may bunch up.
- When updates hit, they can trigger a wave of on-chain actions at once.
- Traders anticipating these mechanics may front-run expected liquidation zones.
This means that spot price is not a monolithic term. Spot price, index price, oracle price and exchange mark can all vary, and algos trade on that spread.
So Is This Manipulation
Sometimes there is obviously illegal activity such as wash trading, spoofing, or organized pumps. More frequently today's anomalous price movements are better attributed to market structure:
- Fragmented venues plus arbitrage
- Conditional liquidity provision
- Leverage plus liquidation rules
- Shared signals and automated risk controls
- Machine-learning strategies with emergent behavior
Issues around market structure in the crypto-assets markets are beginning to attract regulatory attention, including the operation of trading platforms and activities by market participants to manipulate outcomes. The unpleasant reality is that markets can fail even when nobody is ripping anyone off.
What This Means for Traders and Investors
To thrive in algo-fueled markets, the edge comes from reading structure, not predicting headlines:
- Watch liquidity, not just price. Spreads, depth, and sudden thinning matter.
- Respect leverage. Liquidation cascades are algorithmic accelerants.
- Expect system changes. Calm can flip to chaos when models de-risk together.
- Know the reference price. Exchange last price is not the same as mark price, which is not the same as oracle price.
Crypto remains a human-driven market of fear, greed, story, and adoption. However, intraday price action is becoming dictated by code more and more.
When you understand this, most of the perceived randomness in volatility seems contrived, not by conspiracy but by the invisible hand of algorithmic markets.
