Live Trading
Also known as: Real Money Trading, Live Deployment, Real Account Trading, Going Live
What is it?
Live trading is running a strategy on a real-money account, where orders reach the market, fills are real, and the result is your actual capital rather than a simulated balance. It is the last of four stages, and the only one where costs are fully truthful. A backtest replays historical data with assumed spread and no market impact.
- 1Backtest - historical data, assumed costs The rules are replayed over past data with an assumed spread, no slippage and no requotes. This stage tests the logic and nothing about execution.
- 2Forward test - live prices, simulated money The same rules run on a demo account against live prices. Timing becomes honest, but fills are still simulated, so execution cost remains understated.
- 3Live at reduced size - real fills, small exposure Real money at a small fraction of intended size. The purpose is cost discovery: confirm that live spread, slippage and swap match what the model assumed.
- 4Live at full size Intended size, entered only after the live numbers hold at reduced size. Nothing about the rules changes here - only the amount of capital exposed to them.
- !Kill criteria - written before stage 3 The drawdown and losing streak at which the strategy stops, decided in advance. Written during a drawdown instead, they are always too generous. Your capital is at risk.
A forward test runs on live prices with simulated money, so timing is honest but fills are not. Live trading adds everything the first two stages had to assume: real slippage, requotes, partial fills, swap charges and the broker's actual execution behaviour during news. A system showing a 62% backtested win rate on 1.2-pip assumed spread can look materially different once a live 1.9-pip average spread and 0.4 pips of slippage are subtracted from every trade.
The standard practice is therefore to go live in stages rather than at full size: a small allocation first, purely to confirm that live costs match the model, then an increase only once the live numbers hold. Written kill criteria - the drawdown or losing streak at which you stop - belong in place before the first live order, not after one. Your capital is at risk.
Why it matters: Live trading is the first stage where real costs and real fills apply, which is why results often differ from a backtest that had to assume them.
This is the only stage where losses are real, so sizing, cost verification and written kill criteria matter more here than any rule in the strategy.
Real-world example
A scalping bot backtested on an assumed 1.2-pip spread went live into a 1.9-pip average with 0.4 pips of slippage - roughly 1.1 pips of extra cost per trade, enough to change the outcome of a short-hold strategy.
How SignalBots handles it
SignalBots supports a demo and real account toggle so the same signals can be run on a simulated account first and moved to live only once the cost model has been confirmed on your own broker. Your capital is at risk.
See /risk-warning.
Pro tip
Write the kill criteria - the drawdown and losing streak that stop the strategy - before the first live order. Written after a drawdown, they are always too generous.
Common pitfalls
Going live at full intended size straight from a backtest, so the first real measurement of live costs happens with the maximum amount of capital exposed.
Frequently asked questions
How long should I forward test before going live?
Long enough to see a losing streak comparable to the backtested maximum, which is usually two to four weeks for an active system. Trade count matters more than calendar time.
Why do live results differ from the backtest?
Because a backtest assumes spread, has no slippage and no requotes, and fills every order at the modelled price. Live trading applies real spread, slippage, swap and the broker's actual behaviour in fast markets.
What size should I start live trading with?
Small enough that the first weeks are a cost measurement rather than a position. A common approach is a small fraction of intended size, increased only once live spread and slippage match the model.
Does a demo account behave the same as a live one?
Not entirely. Demo servers usually fill at the quoted price with no requotes or partial fills, so a demo tests the logic and the timing but understates execution cost.
When should I stop a live strategy?
At the criteria you wrote before starting - typically a drawdown beyond a multiple of the backtested maximum, or a losing streak longer than the test produced. Deciding mid-drawdown is how small losses become large ones.
Trading involves substantial risk of loss. Historical and backtested results do not guarantee future performance. Read the full risk warning.
