Core guide

Automated trading: what it is, what it costs, what breaks

A program that places orders by fixed rules is neither magic nor a scam. It is a trading plan that executes itself, and everything that matters about it can be checked before you spend a dollar.

Updated August 27, 2026 · 36 min read

A desk in a home office before dawn, one monitor showing a futures chart with an order ticket where entry, stop and target are already filled in, and an empty chair.

It is 8:29 AM ET on a Tuesday. Nobody is in the room. There is a cold cup of coffee on a desk in a spare bedroom and the person who owns the account is in the shower.

At 8:30:00 a US economic release hits the wire and the pre-open starts moving. Seconds later a program on that computer sends three orders in the same instant: an entry, a stop below it, a target twice as far away. Forty minutes later the position closes and the platform writes a line in a log file. The owner made no decisions this morning. He made them months ago, in writing.

That is automated trading with the marketing stripped off. Not a robot with opinions. A plan that executes itself.

The one thing to remember

Automation does not improve a trading plan. It closes the gap between the plan and what gets done. If the plan has an edge, that gap was costing you money. If it doesn't, automation just loses the money faster and more politely.

What automated trading is, and what it is not

Automated trading means writing rules a computer can execute, connecting the program to a broker, and letting it send real orders without asking you first. The rules are boring on purpose: when these conditions are true, buy this size, stop here, target there, close when either is touched.

Three things it is not. Each one costs people money every year.

What it removes is the human at the moment of execution. On the evidence, that is where most retail money dies. Not in the analysis. In the doing.

The chain inside the machine: data, decision, order, log

Every automated system, from a hedge fund's to the one in that spare bedroom, is the same four-link chain.

  1. Data in. The platform takes the exchange stream and hands each tick to the program. Delayed or dirty data poisons everything downstream.
  2. Decision. The code checks its conditions. Almost always the answer is "not now," and doing nothing is correct. A system that must trade daily was built to look busy.
  3. Order out. Orders go to the broker's API. A serious system sends the protective orders in the same breath as the entry, so the worst case is defined before anything happens.
  4. Log. Every decision, fill, rejection and error is written down: the link that tells you whether the system still behaves like its own history.
Diagram of the four stages of an automated trading system: an exchange data feed flowing into a decision block, the decision block sending an entry order with stop and target attached to the broker, and every stage writing into a log file.
The four stages every automated system runs through. Break one link and the other three stop meaning anything.

The chain contains no opinion, no hesitation, no revenge. You can see it applied to one specific system on how it works.

From an institutional desk to a spare bedroom

Rule-based trading is not new. Floor traders ran mechanical systems on paper in the 1970s. What was expensive then was not the idea. It was the plumbing: a data feed, a line to the exchange, and programmers.

Three things collapsed those costs. Exchanges went fully electronic, so an order from a laptop reaches the matching engine the way an institution's does. Retail platforms shipped full programming environments. And in 2019 the CME launched micro contracts, cutting trade size by a factor of ten and making a $10,000 account a legitimate place to run a system.

What did not collapse is the hard part. Institutions still have better fills, cheaper commissions and full-time researchers. The gap narrowed on plumbing and stayed exactly as wide on thinking, the subject of algorithmic trading.

The tools got cheap. The discipline never went on sale.

The four things sold under one name

"Automated trading" sells four products with almost nothing in common. Most of the money lost in this field comes from buying one while believing you bought another.

FamilyHow it makes moneyWhat you must trustHow it fails
Rule-based system on a retail platformA small edge repeated over hundreds of tradesThat the rules weren't fitted to the pastThe edge was never there
Grid and martingale botsSelling oscillations, adding size against the moveThat the market keeps oscillatingOne trend empties the account in a day
Signals and copy tradingSomebody else's decisions, mirroredA person you cannot auditThey change rules or quit, and you learn late
Machine-learning black boxesA model fitted to historical patternsA process nobody can explainIt degrades silently, and no one knows why

None is fraudulent by definition. What separates them is how much you can inspect before paying: a rule-based system runs in your own platform over dates you choose, and a black box does not. More in best trading bots and best trading systems.

By hand or by system: what actually changes

Automation is not about speed. A system trading one window a day competes with nobody on microseconds. What changes is consistency, and consistency has a dollar value.

 By handBy system
Rules on a bad dayNegotiable, and you will negotiateIdentical to a good day
Stop placementOften mental, often movedLive at the broker from the first instant
Missed setupsWhenever you're away or unsureNone, if the machine is on
Testable in advance?No. You cannot backtest a personYes, day by day, over years
What limits the resultYour nervesThe quality of the rules
How it goes wrongYou break your own planYou override it, or the plan was bad

That last row is the honest one. Automation moves the failure point from your discipline to your judgment about what to run.

Everything you need to run one

The inventory is short, and every line is checkable before you commit.

A platformNinjaTrader 8 for US futures. Free for charts, history, backtesting and simulation.
A strategyThe rules, compiled as code. You write one, or you buy one with a reproducible record.
Real-time dataA CME non-professional subscription. Backtests run on history; live trading needs live prices.
Capital and an accountYour own futures account, or a funded account from a prop firm.
A machine that stays onA Windows PC during the window, or a small VPS so nothing depends on your power company.
A habit of checkingTen minutes a week reading logs against the system's own history. This is the actual job.

None of it requires a technical background. The install and first simulated session are in get started; the index futures specifics are in futures trading bot.

What it costs, line by line

Most pages about automated trading talk about returns and go quiet about costs. Costs are the part you can be certain of, so they come first.

Line itemTypical costAvoidable?
Platform: analysis and backtesting$0, no time limitAlready free
Platform: live automationLease, one-time license, or higher per-contract ratesNo, but you choose the shape
Real-time CME data, non-professionalRoughly $10 to $15 a monthNo; sometimes bundled
The strategyOnce. See pricingFree systems exist, usually for a reason
VPS, optionalA few dollars a monthYes, if you leave a PC running
Commissions≈$1 per micro contract, every tradeNo. Budget 5% of gross
Funded evaluation≈$100 for 50k, ≈$250 for 100k, ≈$400 for 250kOnly on your own capital
SlippageSmall per trade, real in aggregateNo; reduced by liquid contracts
Cost breakdown for one year of running an automated futures system on one micro contract, with commissions as the largest slice, then evaluation fees, market data and the VPS, and the one-time platform license and strategy purchase shown separately below the recurring costs.
Where the money goes in a year on one micro contract. The only cost that scales with activity is commissions.

Two lines get forgotten. Commissions shave roughly 5% off gross at micro volumes. And an evaluation fee is not a one-time entry ticket: accounts get consumed and re-bought.

Two routes to the capital, worked out in dollars

There are two places the money can come from, and the choice changes almost everything.

Route one, your own account. You post margin and keep every dollar. The catch is sizing: you fund the drawdown, not the margin. The backtest shows a maximum drawdown of $4,379 on one micro contract, and the worst drawdown in any record is only the worst one so far. Two to three times that on top of margin puts a one-contract account near $10,000 to $15,000. That is money that can go to zero without changing how you live.

Route two, a funded account. A prop firm sets an evaluation on a simulated account. Hit the target without breaking the drawdown or daily loss rules and they fund you, then you split profits. A 50k evaluation costs roughly $100, which caps a single loss at $100 rather than at whatever is in your account. In exchange you accept their rulebook, and a trailing drawdown that can end a profitable account. We support Apex Trader Funding, Lucid Trading, My Funded Futures and Tradeify, meaning the system is built to run inside their rules. You buy the evaluation yourself and pay the firm directly. We do not sell, fund or provide accounts, and no account is included with the software. See funded capital.

The worked example, dollar by dollar

The last seven months of the published backtest: 50k funded account, one micro contract. Simulated figures.

  1. Gross: $43,322 over seven months.
  2. Commissions at ≈$1 per micro take about 5%, leaving roughly $41,000 net.
  3. Accounts consumed: about seven, ended by trailing drawdowns. At ≈$100 each, ≈$700, about 1.7% of the net.
  4. Data and VPS at ≈$27 a month: ≈$190.
  5. Before the split: roughly $40,100. Firms keep a share and cap early payouts. Apply their terms to that number, never to the gross.

Now the part the pitch skips. That stretch ran about $5,900 a month against a full-backtest average of ≈$2,900 net per month, or ≈$35,500 a year. Roughly double the norm. A good run, not a typical one.

Hypothetical performance. Every figure above comes from a backtest, a simulation over historical data. No real money was at risk, and simulated results are prepared with hindsight and cannot fully reflect execution, slippage or liquidity. Past performance, real or simulated, does not guarantee future results.

Numbers you can reproduce beat numbers you have to trust

The complete report: 4,557 trades, every losing month, the deepest drawdown, and how to run it yourself.

How you measure a system, and the disease that fakes it

Total profit is the least informative number in a track record: the destination, nothing about the road.

MetricWhat it tells youIn the Rentabilio backtest
Number of tradesWhether the sample means anything4,557 over 88 months
Win rateHow often it is right. Alone, nearly meaningless46.2%; it loses more often than it wins
Average win / lossWhether the wins pay for the losses$354 / $193, a ratio of 1.84
Profit factorDollars won per dollar lost; under 1.0 it loses1.58
Maximum drawdownThe deepest hole you had to sit through$4,379 on one contract
Test periodWhether it saw more than one market7+ years, day by day
Net after costsThe only profit number that is real≈$260,700, from $274,406 gross

A 46.2% win rate reads as a defect until you do the math. With the target at twice the risk, breakeven sits near a 33% hit rate. Right 46% of the time with wins 1.84 times the size of losses is what a durable edge looks like: unimpressive per trade, positive in aggregate.

Now the disease. A backtest is the most useful tool in this field and the easiest to poison. Curve fitting is a developer nudging parameters until the historical equity curve looks beautiful; the result is a system tuned to noise that no longer exists. Invisible in the report, obvious in live trading three weeks later. Defenses: a long test period, few parameters, and out-of-sample data the developer never saw. See overfitting and curve fitting.

Everything that can break

Only one of those is a software problem. The rest are human. What a normal bad stretch looks like is on the risk page.

Automating orders in your own account is legal in the US. What is regulated is trading other people's money or advising for compensation, which requires registration with the CFTC and NFA membership. Buying software and running it on your own account is neither. A prop firm evaluation is different again: you trade the firm's simulated capital under a contract, not client funds. The rule that binds everyone: nobody may promise you returns, and any figure from a simulation must be labeled as one.

On taxes, US futures generally fall under Section 1256 and its 60/40 rule: 60% of the gain or loss treated as long-term, 40% short-term, whatever the holding period. Open positions are marked to market at year end, a non-issue for a system that holds nothing overnight. Funded account payouts are treated differently (often ordinary income on a 1099) because a firm paid you rather than you trading your own contracts. Background in futures taxes under Section 1256. None of this is tax advice.

Warning signs, wrong fits, and real timelines

The warning signs when somebody sells you a system

Account screenshots instead of a reproducible report. Monthly percentage promises, which no fund on earth delivers reliably. A win rate above 80%, which almost always means tiny targets and an enormous hidden stop. No maximum drawdown published. Testimonials with stock photos. Countdown timers. And the clearest tell: you cannot run the thing yourself, before paying, over dates you choose. Checklist in trading bot scams and red flags.

Who this is not for

Anyone trading money they need. Anyone who wants excitement: a good system is deliberately dull, and boredom makes people meddle. Anyone who cannot watch an account fall for six weeks without touching it. Anyone expecting to replace an income in the first quarter.

How long it really takes

About a week to install the platform and get comfortable. One or two more to reproduce a backtest and sit through simulated sessions. Then an evaluation, which takes as long as it takes: sometimes two weeks, sometimes three failed attempts across two months. Then the part nobody markets: a quarter or more of live trading before you have enough trades to say anything statistical about your own results.

What Rentabilio is, and the road to a first trade

Rentabilio is the system sold on this site: a strategy for NinjaTrader 8 trading US index futures on micro contracts. It does four things.

  1. It trades one window a day, at 8:30 AM ET, when US economic data is released and the pre-open starts moving. The rest of the day it is asleep. A system awake 24 hours mostly finds 24 hours of commissions.
  2. It reads order flow before entering. Session bias, real volume, where institutional pressure shows up. If the read is not clean it does not trade that day.
  3. It places stop and target in the same instant as the entry, target at 2× the risk. The worst case is known before anything happens.
  4. It never holds overnight. Target or stop, the trade ends and the day is over.

The published backtest covers 88 months day by day on a $50,000 funded account, one micro contract: $274,406 gross, ≈$260,700 net, 4,557 trades, a 46.2% win rate, profit factor 1.58, maximum drawdown $4,379. Simulated results over historical data; past performance does not guarantee future results.

From curiosity to a first trade, five steps, none irreversible:

  1. Read the backtest and the risk page: drawdowns first, profits second.
  2. Install NinjaTrader 8, connect data, reproduce the report over dates you pick.
  3. Run it in simulation through several live 8:30 AM ET sessions.
  4. Choose a capital route: your own account sized off the drawdown, or a ≈$100 evaluation.
  5. Start at one contract, keep the logs, and change nothing for at least a quarter.

If something is unclear, the contact form opens a support ticket with a reference. It is the only door in. No email address, no chat.

Frequently asked questions

Is automated trading legal in the United States?

Yes. Running a program that places orders in your own account is legal, and it already accounts for much of the volume on US futures exchanges. Registration with the CFTC and NFA applies to people managing other people's money, not to a self-directed trader using software. Your results are still reportable to the IRS, and futures fall under Section 1256.

How much money do I need to start automated trading?

On the funded route, roughly $100 for a 50k evaluation, plus the platform, the strategy and about $12 a month for data. Budget three or four evaluations across your first months, because accounts get consumed and re-bought. On your own capital the figure is closer to $10,000 to $15,000 for one micro contract, because you fund the drawdown, not the margin.

Does my computer have to be on all day?

Only during the window the system trades. A strategy that works one session a day needs the platform running through that session and nothing more. A small virtual private server keeps everything up around the clock for a few dollars a month.

Can an automated system lose money?

Yes, and it will, regularly. The published Rentabilio backtest loses on 53.8% of its trades and went through a maximum drawdown of $4,379 on one contract. A system makes money by winning bigger than it loses across many trades, not by avoiding losses.

How do I know whether the system I bought is still working?

You compare it against its own history rather than against your hopes. Four or five losses in a row are normal at a sub-50% win rate and prove nothing. What matters is whether the win rate, the average win and the drawdown stay inside the range the backtest showed.

In short: automated trading is a written plan executed by something that cannot get scared. The machinery is simple and the costs are knowable. The hard parts are human: choosing a system whose numbers you can reproduce, funding it so a normal drawdown does not end it, and leaving it alone long enough to find out. If you cannot verify it before you pay for it, you already have your answer.

See a real system before you decide anything

The full Rentabilio backtest (88 months, 4,557 trades, every losing month, the deepest drawdown) sits on one page, with the steps to reproduce it yourself in NinjaTrader 8.