You can run the same strategy logic against a crypto exchange or a regulated futures market, and the code will look almost identical. Everything that decides whether it survives sits outside the code: who holds your money, what happens when the venue goes down, what a round turn costs, and how deep the book is at the exact moment you need out.
This is not an argument that crypto is bad. It is a list of structural differences that decide where automation is worth the trouble.
What the two have in common
More than the tribal arguments suggest. Both are leveraged, both run electronically with API access, both let a program place and manage orders without a human present, and both punish an untested strategy at the same speed. A bot is a bot in either venue: rules that execute without hesitation, for better or worse.
The differences are in the plumbing. In trading, the plumbing is most of the risk.
The one thing to remember
In futures, your counterparty is a clearinghouse and your risk is the strategy. On a crypto exchange, your counterparty is the exchange itself, so you carry the strategy risk and the venue risk. Automation multiplies whatever risk is already there.
Who is on the other side of your trade
When you buy a contract on the CME, the trade is novated to CME Clearing, so from that instant your counterparty is the clearinghouse, not the anonymous seller. Your broker is a futures commission merchant regulated by the CFTC and an NFA member, and customer funds must be segregated from the firm's own money. None of that makes losses impossible. It makes one category of loss extremely unusual: the venue keeping your money.
On a crypto exchange, you deposit assets into the exchange's custody and trade against its internal order book. The exchange is your broker, clearinghouse, custodian and price source at once. Every one of those roles is a place things can go wrong, and the history is not theoretical: Mt. Gox in 2014, FTX in 2022, and a longer list of smaller venues that stopped processing withdrawals and never restarted.
For a bot this matters more than for a manual trader. Automation means capital sits on the venue continuously so the program can act. Leaving it there only while you are actively trading is not an option when the program trades on its own schedule.
Hours: a session with an end, or no end at all
CME equity index futures trade close to 23 hours a day, Sunday evening through Friday afternoon, with a daily maintenance break and an official settlement price. There is a defined close, and that gives automation something it genuinely benefits from: a natural point to be flat.
Crypto never closes. No weekends, no settlement, no maintenance window. That sounds like an advantage, and for a market maker it is. For a retail bot it mostly means the position that goes wrong does so at 3:00 AM on a Sunday, when depth is thinnest and you are asleep.
A market that never closes is a market where nobody is watching at the worst possible moment.
It also changes what risk management has to mean. A system that trades one window a day and closes before it ends (the approach behind the 8:30 AM ET window) has a defined maximum exposure per day by construction. A continuously running crypto bot holds whatever its last decision left it, indefinitely.
What each one costs to run
Futures commissions are charged per contract, flat, regardless of price. Roughly $1 per micro contract per side is a realistic figure, so a round turn is on the order of $2. A Micro E-mini S&P contract is $5 per index point, which at an index near 5,000 is about $25,000 of notional exposure. Two dollars on $25,000 is about 0.008% of notional per round turn.
Crypto fees are charged as a percentage of notional. Taker fees on major venues commonly run somewhere around 0.02% to 0.10% per side depending on your volume tier, so a round turn near 0.10% is ordinary. On the same $25,000 of exposure that is roughly $25, more than ten times the futures cost for the same position size.
Perpetual swaps add a cost with no futures equivalent: the funding rate, exchanged between longs and shorts at regular intervals to hold the perp near spot. In calm conditions it is small. In a directional market the crowded side pays it every day, purely for holding. A bot that carries positions for days is paying rent no naive backtest shows.
Trade frequency is the hidden variable. In the Rentabilio backtest, commissions took about 5% off the top across 4,557 trades: $274,406 gross became ≈$260,700 net. Those are hypothetical, simulated figures rather than a live account. Run that trade count at ten times the cost per turn and the arithmetic stops working.
Depth: what your stop actually gets filled at
A stop order is a promise to exit, not a promise about price. What you get depends on the book at that instant.
Index futures are among the deepest markets in the world, and the front-month contract typically shows a one-tick spread with real size behind it, even during data releases. In practice a small order fills where you expected.
Major crypto pairs are liquid too, and on a normal day the difference is not the point. It shows up in a cascade. Leveraged liquidations trigger more liquidations, the book empties, the spread widens, and stops fill far from their trigger. Add exchange outages during exactly those minutes (a recurring pattern across venues) and you get the scenario a bot handles worst: it wants to act, the API is unreachable, and the position keeps moving.
Grid bots and the trend that does not come back
The most popular crypto automation is the grid bot, and it deserves an honest explanation because its failure mode is arithmetic, not bad luck.
A grid places buys below and sells above at fixed intervals. Price oscillates, each round trip books a small profit, and in a sideways market it works exactly as advertised. The equity curve is a gentle staircase, which is what makes it feel safe.
Now take a real trend. An asset at $100 falls to $50, and a grid buying one unit every $5 fills at 95, 90, 85, 80, 75, 70, 65, 60, 55 and 50. That is 10 units at an average cost of $72.50, or $725 spent, now worth $500. An unrealized loss of $225 while the grid, having done nothing wrong by its own rules, is still buying. The profits it harvested were capped at a few dollars per step. The loss is not capped by anything.
With leverage it ends sooner and worse: liquidation converts the unrealized loss into a realized one at the least favorable price of the move. That is the structural point: a grid bot is bounded upside against unbounded downside, running in a market that trends hard. A system that places a stop and a target in the market at the instant of entry, target at twice the risk, has the opposite shape. Neither is magic; they fail differently, and only one fails in a size you chose in advance. That is how the risk model works.
The funded-account gap
This is the difference nobody mentions, and it may be the largest.
In US futures there is a mature ecosystem of proprietary trading firms that will let you trade their capital after you pass an evaluation. Apex Trader Funding, Lucid Trading, My Funded Futures and Tradeify are the ones we support, and you buy the evaluation from the firm yourself. Topstep, Take Profit Trader, Earn2Trade and Bulenox are other well-known names in the same space. An evaluation for a $50,000 account typically costs around $100, a $100,000 account around $250, and a $250,000 account around $400. Pass, and you trade firm capital under firm rules.
The economic consequence is that your downside per attempt is the evaluation fee, not the account. In the strongest stretch of the Rentabilio backtest ($43,322 gross in the last 7 months, roughly double its own historical monthly average), getting there hypothetically would have consumed about 7 funded accounts along the way, roughly $700 in fees. Simulated, not a live account. But that structure does not exist in crypto at comparable scale or price: a handful of firms offer crypto evaluations, the ecosystem is far smaller and less standardized, and the rules vary enormously. In practice, automating crypto means risking your own money. The futures side is covered in funded capital and how funded accounts work.
| US index futures (CME) | Crypto exchange | |
|---|---|---|
| Counterparty | Central clearinghouse | The exchange itself |
| Customer funds | Segregation required by CFTC rules | Varies by venue; often exchange custody |
| Hours | About 23 hours, Sunday to Friday, defined close | Continuous, no close, no settlement |
| Cost per round turn | Flat per contract, near 0.008% of notional on a micro | Percentage of notional, commonly near 0.10% |
| Holding cost | None intraday; quarterly roll | Funding rate on perpetuals, paid repeatedly |
| Depth under stress | Deep, tight spread, rarely halted | Thins fast in cascades; outages documented |
| Funded capital | Mature prop firm market, evaluation from about $100 | Thin and inconsistent; usually your own money |
| Typical retail bot | Rule-based, stop and target on entry | Grid or DCA, frequently without a stop |
| Tax treatment | Section 1256 contracts | Property; every disposal is an event |
Where crypto automation genuinely makes sense
Being specific cuts both ways. Crypto has real structural advantages.
- Small size. You can trade a few dollars of exposure. Even a micro futures contract carries thousands of dollars of notional.
- Genuine inefficiency. Fragmented venues, uneven fees and unequal information leave opportunities a mature market arbitraged away decades ago.
- Open access. No brokerage approvals, an API key in minutes, and markets that do not care what time zone you live in.
- An asset class futures does not cover. Wanting to hold the asset is a different goal from trading a price series.
What those advantages share is that they favor small size and self-funded capital. What they do not do is fix venue risk, funding costs, or the missing evaluation market that caps your downside per attempt.
See what the futures side looks like with real numbers
More than seven years of backtest on a $50,000 funded account: 4,557 trades, 46.2% win rate, 1.58 profit factor, $4,379 max drawdown, plus the steps to reproduce it in NinjaTrader 8.
Hypothetical performance. The Rentabilio figures in this article come from a backtest over historical data, not from a live account. Simulated results are prepared with hindsight, carry no financial risk, and cannot fully reflect real execution, slippage or liquidity. Past performance, real or simulated, does not guarantee future results.
Frequently asked questions
Are crypto trading bots profitable?
Some are, in the same limited sense any strategy is: over a specific period, in specific conditions, after costs. The headwinds are higher percentage fees, funding costs on perpetuals, venue risk, and the fact that the most common design has unbounded downside in a trend. That does not make profit impossible, but an edge has to be larger to survive the friction.
Is a grid bot safe if I do not use leverage?
Safer, because you cannot be liquidated, but not safe. The failure mode changes from a forced exit to an indefinite hold: capital tied up far underwater while the grid keeps buying and no rule ever says stop. That is an allocation problem rather than a blowup, which is better, but it is still an unbounded loss against a capped gain.
Can I use the same bot on crypto and futures?
The logic sometimes transfers; the risk parameters almost never do. Different volatility, tick and fee structures, depth and session behavior mean a stop distance tuned for one venue is arbitrary in the other. Anything that moves between markets has to be retested from scratch on the new venue's data, with that venue's real costs included.
Why do prop firms exist for futures but barely for crypto?
Because the infrastructure a funding model needs is already in place: a central clearinghouse, standardized contracts, a regulated brokerage layer and risk software that enforces a daily loss limit reliably. Crypto's fragmented venues and inconsistent custody are much harder to underwrite, which is why the ecosystem described in prop firm trading is far more developed on the futures side.
Which should I automate first?
If the goal is trading a system with known costs, deep liquidity and the option of using someone else's capital, regulated futures are the more forgiving place to start, and how to trade futures covers the mechanics. If the goal is exposure to crypto assets themselves, or testing ideas at very small size, crypto is the right venue and the job is to respect venue risk rather than pretend it is absent.
In short: the code is the easy part. Futures give you a clearinghouse, a defined close, flat per-contract costs, deep books and an evaluation market that caps what a failed attempt costs. Crypto gives you access, tiny size and real inefficiency, in exchange for venue risk, percentage fees, funding costs and a market that never lets you be flat by default. Automate where the structure works with you.
