Crypto algo trading

Crypto algo trading: strategies, platforms and how to start

Crypto algo trading means trading crypto by exact written rules that a computer checks and acts on, instead of by hand. It is also called algorithmic crypto trading, and the rule set is the trading algorithm. Keel is a crypto algo trading platform for Hyperliquid perpetuals: you build a strategy from blocks you can read, backtest it with fees, slippage and funding, paper trade it, then run it from your own account.

By Keel Research Team
Updated November 17, 2026
An algo strategy you can read
Momentum Strategy on Hyperliquid
  • Once a day, rank the 50 most traded Hyperliquid perps by their 14-day return.
  • Keep the top 25% of them, and only coins whose volume is running in the upper half of the group.
  • Hold nothing while the wider market is trending down.
  • Size each coin by its volatility, cap any one coin at 15% of the account, and never borrow.
Sharpe
1.17
Return
+151%
Worst dip
−39%

Backtest on Hyperliquid, Aug 2024 to Sep 2026, after fees, slippage and funding.

The basics

What crypto algo trading is

Algo is short for algorithm. In trading, an algorithm is a set of exact rules: which coins to look at, what makes you buy or sell, how big each position is, and how often to check. A computer follows those rules on a schedule and places the orders. You decide the rules once, and the program applies them the same way every time.

You will see several names for the same thing. Algorithmic crypto trading and crypto algorithmic trading are the long forms. Systematic trading stresses that the rules are fixed in advance. Quant trading usually means the rules come from statistics on market data. A crypto trading algorithm is the rule set itself, and crypto algo trading software, or an algo trading platform, is the program that tests the rules and runs them.

Crypto suits rules for plain reasons. The market trades all day, every day, so nobody can watch it by hand. Exchanges publish prices and take orders through public APIs. Most active trading is in perpetual futures, which let a strategy go short, and which pay or charge funding that a rule can measure and trade on.

Rules don't make a strategy profitable. An algorithm applies an idea faster and more consistently than a person would, so a bad idea loses money more consistently too. That is why testing comes before trading: a backtest on past data with full costs, then a paper run on live prices.

The four parts of any algo trading setup

  1. 01

    Rules

    The strategy, written so that nothing is left to judgment on the day: the coins, the entry, the exit, the size and the schedule.

  2. 02

    Data

    Price history for every coin the rules touch and, for perpetual futures, the funding history. The same data feeds the test and the live run.

  3. 03

    A backtest

    The rules replayed over past data, bar by bar, with trading fees, slippage and funding charged. It shows the drawdowns as well as the return.

  4. 04

    Execution

    The program that turns each decision into orders on the exchange, keeps running when you are away, and lets you stop it.

Two words

How an algo strategy differs from a trading bot

The two words overlap, and many people use them for the same thing. In practice, a trading bot usually means a ready-made program you switch on, such as a grid bot or a DCA bot. You pick a coin and fill in a few settings, and the logic is the template's.

An algo strategy starts from the other end. You write the rule yourself, test it on history, and only then let a program run it. The program is the smaller part. The work is in the rule, and in the evidence that it held up after costs.

How an algo strategy differs from a trading bot
A template botAn algo strategy
Where the logic comes fromA fixed pattern someone else wrote. You set its parameters.A rule you wrote or changed, which you can read line by line.
What it tradesUsually one coin or one pair at a time.Often a basket of coins that are ranked, filtered and sized together.
What you know before it tradesSometimes a short simulation. Often only how the pattern is meant to behave.A backtest of the exact rule with fees, slippage and funding, and a paper run.
When it stops workingHard to tell a rough patch from a broken bot when the logic is hidden.You can compare live behaviour with the backtest of the same rule.

If a ready-made bot is what you want, see how to pick the best crypto trading bot, or what it takes to build a Hyperliquid trading bot.

The strategies

Common crypto algo trading strategies

Most rule-based strategies belong to one of four families. Each links to a backtest on Hyperliquid data with fees, slippage and funding charged, including the runs that did badly.

Trend following

Hold a coin while its price trend is up, and get out or go short when it turns down. A common version compares a fast moving average with a slow one.

Works in

Moves that last: long rallies and long sell-offs.

Struggles in

Sideways markets, where each signal reverses before it has paid for its costs.

The EMA crossover backtest

OKOur grade for this rule on a basket of Hyperliquid perps.

Momentum

Buy what has gone up over a set lookback and sell or short what has gone down. Across many coins it becomes a ranking: hold the strongest and skip or short the weakest.

Works in

Stretches where the winners keep winning.

Struggles in

Turning points, when the strongest coins fall the hardest.

The momentum backtest

OKOur grade for this rule on a basket of Hyperliquid perps.

Mean reversion

Buy a coin after it drops far below its recent average and sell when it comes back. The bet is that a stretched price snaps back.

Works in

Ranges, and the days after an overreaction.

Struggles in

A drop that is the start of a trend, where the rule keeps buying into the fall.

The mean reversion backtest

WeakOur grade for this rule on a basket of Hyperliquid perps.

Funding carry

Perpetual futures pay or charge funding between longs and shorts, every hour on Hyperliquid. A carry strategy holds the side that gets paid: short the perps where funding is highest, long the ones where it is lowest or negative.

Works in

Periods when the funding gaps between coins persist.

Struggles in

Sharp price moves. The positions still move with price, so a rally in the coins you are short can cost more than the funding earned.

The funding carry backtest

Sharpe 1.39 after costs on Hyperliquid, Aug 2024 to Sep 2026.

These four are a start. Every rule we have backtested is graded the same way, and the strategy library holds full strategies you can fork. Grades run Strong, OK, Weak, Poor.

Compared

Crypto algo trading platforms and software compared

Six products people use for algo trading in crypto, each described from its own site. They are listed in no order of preference, with no prices.

Crypto algo trading platforms and software compared
ProductWhat it isWhat it runsMarketsSource
TradetronA hosted platform where you build a trading algo without code, with a strategy marketplace and an MCP connectorBacktests and live trading in your own broker account100+ brokers, per its sitetradetron.tech/crypto-trading, read 17 November 2026
CoinruleA hosted, no-code builder of if-then trading rulesRules you set in its builder, with a backtest25+ exchanges, including Hyperliquidcoinrule.com, read 17 November 2026
NautilusTraderAn open-source trading engineEvent-driven backtests and live trading from the same codeMany venues, including Hyperliquidnautilustrader.io, read 17 November 2026
FreqtradeFree, open-source trading software written in Python, which you host yourselfBacktesting, parameter search and live tradingCrypto exchanges, including Hyperliquidfreqtrade.io/en/stable, read 17 November 2026
HummingbotAn open-source frameworkMarket-making and other strategiesCentralized and decentralized exchangeshummingbot.org, read 17 November 2026
QuantConnectA cloud platform for people who write codeResearch, backtests and live tradingEquities, options, futures, FX and crypto, live through 20 integrationsquantconnect.com, read 17 November 2026

Read from each product's own site on 17 November 2026. Products change, so check the source before you decide. We left out anything we could not confirm there.

The products fall into three groups. Tradetron and Coinrule are hosted and need no code: you set rules in a browser. NautilusTrader, Freqtrade and Hummingbot are open source: the code is public and the strategy is yours to write. QuantConnect is a cloud platform for people who code across many asset classes.

Three questions sort them quickly. Do you want to write code? Where should the strategy run, on your machine or theirs? And which market do you trade, since each product reaches different exchanges?

Keel is a hosted platform too, for Hyperliquid perpetuals only. What it does and what it doesn't is further down this page.

Getting started

How to start crypto algo trading

Two routes cover most people: write the strategy in Python, or describe it to an AI agent. Both begin with one rule written down exactly.

The Python route

Write the strategy in code

Python is the usual language for algo trading. Open-source software such as Freqtrade and NautilusTrader gives you a backtester and a live runner, and Hyperliquid publishes an official Python SDK for placing orders and reading markets. The rest is yours: the price and funding history, the cost model, and a server that stays up.

On Keel, a strategy is a short file in Python syntax, built only from Keel's blocks. You can edit it in the app's Code view, or write it in your own editor and use the keel-trade command-line tool to check it and run backtests on Keel against Hyperliquid history.

Fits if you program and want the strategy in a file you can keep under version control.

The agent route

Describe it to an AI agent

Add Keel's hosted MCP server to Claude, ChatGPT, Cursor or Claude Code and sign in to Keel in your browser. The assistant gets 33 tools: it can search Keel's 262 strategy blocks, build a strategy from your description, run a backtest and read the result back to you.

Then describe the idea in plain words, read the strategy it built, and ask for changes. The agent writes and tests the rules. It never places trades itself: the strategy you approved trades on the schedule it was built with.

Fits if you think in ideas more than in code, and already work in an AI assistant.

The order is the same on either route

  1. 01

    Write one rule down exactly

    Which coins, what makes you enter, what makes you exit, how big, how often. If you can't say when it sells, it isn't ready to test.

  2. 02

    Backtest it with full costs

    Charge trading fees and slippage on every trade and funding on every open perp position. Read the worst drawdown before the return.

  3. 03

    Paper trade it on live prices

    Run the same rule forward with simulated fills. No money is at risk, and you see whether live behaviour matches the test.

  4. 04

    Go live small

    Start with an amount you can afford to lose, check on it regularly, and stop it if the market no longer looks like your test.

Keel

What Keel does

Keel is an algo trading platform for Hyperliquid's crypto perpetuals. One strategy goes from idea to backtest to paper to live without being rebuilt.

01

Build

A strategy is built from 262 building blocks: market data, indicators, filters, ranking and position sizing. Describe the idea and Keel's AI picks the blocks, or place them yourself. Every setting shows on a canvas, and there is a code view of the same strategy.

02

Backtest

Each backtest runs on Hyperliquid's price and funding history. By default every fill pays a 4.5 bps fee and 4.5 bps of slippage, and every open position pays or earns the funding Hyperliquid actually charged.

03

Paper trade

A saved strategy runs forward on live Hyperliquid prices with simulated fills. It needs no wallet, and no money is at risk.

04

Run it live

The strategy you tested trades from your own Hyperliquid account, through an agent wallet that can place trades but can never withdraw. Keel runs it on its servers, so nothing has to stay open on your computer.

What Keel doesn't do

  • Keel trades Hyperliquid's native crypto perpetuals only. Stock and index perps are HIP-3 markets, and Keel doesn't support them yet. Support is coming soon.
  • A strategy decides at the close of each bar, on 5-minute to daily bars. It can't leave orders at price levels it chooses, which rules out grid bots, high-frequency scalping and arbitrage between exchanges.
  • Strategies are built from Keel's blocks, not arbitrary code. If you need a custom model the blocks can't express, a code-first framework is the better fit.

On cost: Building, backtesting and paper trading start free. The free plan paper trades one strategy and runs one live on a timeframe of 1 hour or slower, with a 5 basis point fee on live trades. Paid plans start at $49 a month, run more strategies and lower that fee. Hyperliquid's own trading fees apply as usual. The pricing page has the full limits. See pricing.

Here is one algo strategy from Keel's public library, a momentum strategy across many coins. You see the rules in plain words, the same rules as blocks, and how they did on Hyperliquid. You can fork it into your own account and change it.

1

The rules

  • Once a day, rank the 50 most traded Hyperliquid perps by their 14-day return.
  • Keep the top 25% of them, and only coins whose volume is running in the upper half of the group.
  • Hold nothing while the wider market is trending down.
  • Size each coin by its volatility, cap any one coin at 15% of the account, and never borrow.
2

The same rules, as blocks

Read it from the top. Daily prices come in for each coin. One branch ranks the coins by recent return and keeps the strongest, a second keeps only coins with strong volume, and a third checks the trend of the whole market. The last blocks hold the selected coins only while that market trend is up, and size each one by its volatility. Switch to Code to see the same strategy as text.

Configuration

Globalsconfig
target_timeframe1dbar_offset12h
UniverseassetsTop by Volume · HL Perps · 50 assets50 resolved
Executionorders
rebalanceBufferedbuffer_threshold0.2buffer_moderelative+1

Factories

selectionfactory
Pipeline4 steps
PriceDataLoaderData Loader
Parallel3 branches
selection
selectioncall
market_regime
EWMAIndicator
window2
MarketTrendRegimeFilterRegime Detector
fast_period5slow_period17
vol
ReturnVolatilityIndicator
window36d
return_volstore
RegimeGateForecast Composer
signal_keyselectionindex_keymarket_regimeconditionindex > 0
VolWeightSizerPosition Sizer
vol_slotreturn_voltarget_leverage1max_weight0.15
3

The backtest

EquityDrawdown
Sharpe
1.17
Total return
+151.0%
Worst drawdown
−38.8%
Trades
1,243

Hyperliquid, Aug 2024 to Sep 2026, after trading fees, slippage and funding, on the default settings. Look at the drawdown before the return. The strategy only buys, so it takes the first part of every sell-off before the market check moves it to cash.

FAQ

Questions about crypto algo trading

What is crypto algo trading?

Crypto algo trading is trading crypto by exact written rules that a computer checks on a schedule and turns into orders. It is the same thing as algorithmic crypto trading. The rule set is the trading algorithm, and the software that tests and runs it is the algo trading platform.

Is crypto algo trading profitable?

It can be, and many strategies lose money. An algorithm only applies a rule consistently, so the result depends on the rule. Judge a strategy by a backtest that charges fees, slippage and funding, then by a paper run on live prices, and expect live results to be worse than the backtest. Keel's backtested strategy pages show the losing runs too.

What is the best crypto algo trading platform?

It depends on how you work. If you write code and want full control, open-source software such as NautilusTrader or Freqtrade fits. If you want no code, hosted builders such as Tradetron and Coinrule let you set rules in a browser. If you want to describe a strategy, read it as blocks and backtest it on Hyperliquid with fees, slippage and funding before it trades, that is what Keel is built for.

Do I need to know Python for algo trading?

No. Python is the common language for writing your own, and the open-source frameworks use it. On Keel you can describe the rules in plain words or build them from blocks. A Keel strategy is also a short file in Python syntax built from those blocks, which you can edit if you want to.

What is the difference between algo trading and a trading bot?

A trading bot usually means a ready-made program with a fixed pattern, such as a grid or DCA bot, where you choose a coin and a few settings. Algo trading starts with your own rule, which you test on history before a program runs it. Many people use the two words for the same thing.

Can an AI agent do crypto algo trading?

An AI assistant such as Claude or ChatGPT can write and test the rules. Connected to Keel, it builds a strategy from Keel's blocks and backtests it on Hyperliquid data. The model never places orders: the strategy you approved trades on the schedule it was built with.

Which markets can Keel trade?

Keel trades Hyperliquid's crypto perpetual futures, such as BTC, ETH, SOL and HYPE. Stock and index perps are HIP-3 markets, and Keel doesn't support them yet. Support is coming soon. Your money stays in your own Hyperliquid account, and Keel can place trades there but can never withdraw.

What does crypto algo trading on Keel cost?

Building, backtesting and paper trading start free. The free plan paper trades one strategy and runs one live on a timeframe of 1 hour or slower, with a 5 basis point fee on live trades. Paid plans start at $49 a month, run more strategies and lower that fee. Hyperliquid's own trading fees apply as usual. The pricing page has the full limits.

Build an algo strategy you can test first

Starting is free. You see the rules and the backtest before anything trades, and your money stays in your own account.