Updated 13 October 2026
The most common algo trading myths are that algorithms are only for banks, that they always make money, that a good backtest proves a system works and that you can switch them on and walk away. Our own live account, run since December 2023, shows that none of these hold, and that losing months are part of the job.
Key facts
- Our live account went from 35,000 kr in December 2023 to 218,747 kr after fees on 30 September 2026.
- 14 of its 34 months were negative, and September 2026 alone was −11.8 % (−29,230 kr).
- Of the 12 algos in our live data, 7 are positive in total and 5 are negative per 30 September 2026.
- ESMA found in 2018 that 74 to 89 % of retail CFD accounts typically lose money.
Is algo trading only for banks and hedge funds?
No. A private trader can run a rule-based trading system on their own broker account with retail platforms such as ProRealTime and MetaTrader 5. The large players were first, but the tools to run automated rules are now part of ordinary trading platforms.
With ProRealTime at IG, a system runs in ProOrder on ProRealTime's servers, which ProRealTime describes as fully server side. With MetaTrader 5, an expert advisor runs inside a terminal that has to stay online. Our own live account started with 35,000 kr, not with institutional capital.
What separates a bank from a private trader is not access to automation. It is capital, cost and the risk you can carry. That is why position size matters more than the platform you choose.
Are trading algorithms always profitable?
No. A trading algorithm follows its rules in good and bad market conditions, and the rules will lose in some periods. Our own live account shows this clearly.
Between December 2023 and September 2026 the account had 20 positive and 14 negative months. The worst month was April 2024 at −13.9 %, and the largest decline from a peak was −23.5 % between February and April 2024. A second decline of −20.3 % ran from the peak in May 2025 to February 2026, including four losing months in a row from September to December 2025.
Not every algo pulls its weight either. Per 30 September 2026, Nasdaq M3 pt.1 had made 175,646 kr in total, while Guld M2 had lost 22,649 kr. A win ratio of 69 % on the live results page also means that roughly three trades in ten are losers.
Past results, live or backtested, do not predict future returns. A losing month like September 2026 is a normal part of trading with rules, not a sign that something is broken.
Does a good backtest prove that an algo works?
No. A backtest is a simulation that runs the rules on historical prices. It shows how a system would have behaved, not how it will behave with real orders.
A live account meets things a backtest leaves out. Spread and overnight financing vary, orders can be delayed or rejected, and the broker can go offline. In September 2025 IG had major IT problems, and according to our live results page our algos largely did not trade after 10 September that month. No backtest would have shown that.
That is why we publish live results after fees every month next to any simulated figure. Our article on how a trading bot works walks through the steps from rule to executed order.
Can you switch an algo on and stop watching it?
No. The algo makes the trading decisions, but you still choose the position size, keep enough margin on the account and follow what happens.
ProRealTime, for example, can stop a system that runs out of margin or has repeated order rejections, and open positions are then closed at market. A broker outage can also leave a system without trades for days. Someone has to notice that and decide what to do.
The human role moves from pressing buy and sell to setting the frame. That frame is mainly risk, which our guide to risk management in algo trading covers in more detail.
Is algo trading about being faster than everyone else?
Not for the kind of algos a private trader runs. Our systems are rule based and most work on price bars of one to five minutes, so their edge, if there is one, comes from the rules and the risk control, not from speed.
Most of our algos are trend followers, and Nasdaq M1 and DJ30 M1 are momentum breakouts. Our live account has averaged 7.7 orders a day, according to the live results page on 1 October 2026. That is a different activity from the speed race that large firms run on their own infrastructure.
Does an algo take the risk out of trading?
No. An algo removes the emotional decision, not the market risk. CFDs are leveraged, so a small price move can give a large gain or loss relative to the margin you put up.
When ESMA restricted CFDs for retail clients in 2018, it reported that 74 to 89 % of retail accounts typically lose money. It also capped leverage at 20 to 1 on major indices and gold for retail clients. Rules in code do not change those numbers on their own.
What rules can do is apply the same limits every time. Our live results page states a risk per trade under 2 %, and that limit is set in the system, not decided in the moment.
| Myth | What is true | Our live data (per 30 September 2026) |
|---|---|---|
| Only banks can use algos | Retail platforms run automated rules | Live account started with 35,000 kr |
| Algos are always profitable | Rules lose in some periods | 14 of 34 months negative |
| A good backtest proves it works | A backtest is a simulation | Few trades after 10 September 2025 due to IG outage |
| You can stop watching | Size, margin and outages need a human | Largest decline −23.5 % (February to April 2024) |
| It is all about speed | Our edge, if any, is in the rules | 7.7 orders a day on average |
| Algos remove risk | CFDs stay leveraged products | September 2026 −11.8 % |
Every algo and every year is broken down on A+ Algos in numbers. If you want to see how one of our algos behaves on your own account, you can try one algo free for 14 days, with no card needed.
FAQ
Can a beginner use algo trading?
Yes, if they understand what they are running. You do not need to write code to use a ready-made algo, but you do need to understand CFDs, leverage, margin and position size. Start small, follow the account every week and read what the algo does. A beginner who treats an algo as a money machine is the one most likely to be disappointed.
How often does an algo have a losing month?
It depends on the algo and the market, but losing months are normal. Our live account had 14 negative months out of 34 between December 2023 and September 2026, and the longest losing run was four months, from September to December 2025. Anyone running an algo should plan for runs like that before they start.
Is algo trading legal for private traders in Sweden?
Yes. Running a rule-based system on your own account with a licensed broker is ordinary trading. Retail CFD trading in the EU is regulated, with leverage limits, a margin close-out rule and negative balance protection from ESMA's 2018 measures, later adopted by national supervisors. The algos we sell are software licences, not advice or asset management.
Do algos work on an ISK?
No. The algos we sell trade CFDs on Nasdaq 100, Dow Jones 30, DAX 40, CAC 40, gold and EUR/USD, long and short, on your own broker account. They do not trade individual shares and cannot run on an investment savings account (ISK).
Summary
Algo trading is open to private traders, but it is neither always profitable nor hands off. A backtest is a simulation, and only live results after fees show how rules meet a real market. Our own account has had 14 losing months out of 34 and a largest decline of −23.5 %, and we publish every one of them.
See the results live. Every month since December 2023 we publish the result of our own algos after fees on the live results page. Our most traded algo is Nasdaq M3 for ProRealTime.
Trading CFDs involves high risk and you can lose money. Past results, live or backtested, do not predict future returns.