AmiBroker Backtesting

What it is

AmiBroker is a technical analysis and testing program. Programs of this kind are known for a formula language that works on whole arrays at once: instead of looping through bars one by one, a line of code describes a calculation for every bar in the history in a single statement. That style is compact and fast for backtesting a rule across many symbols, and it has a built-in portfolio-level backtester and tools for testing a rule on rolling windows of data.

It is mainly a testing and analysis tool. Sending live orders usually goes through add-ons or bridges to a broker that you or others supply, so the live link is a separate thing from the test. What those links are, and what they cost, is for the vendors to say. Check the current pages. This page explains how it works in general terms. We have not tested it for this page. Check the vendor's current terms.

How a bot connects here

The connection style is built-in scripting for the test and a bridge for live orders.

• Array thinking. A signal is an array of true and false, one per bar. Shifting an array forward or backward in time is easy to do the wrong way round.
• Trade timing settings. A backtest can be told whether to trade on the signal bar or later, at the open or close. This setting is as important as the rule.
• Portfolio backtest. Position limits, ranking and costs apply across symbols.
• Bridge to live. A separate piece of software translates signals into orders. It is a different program with different failure modes.

Worked example: one signal, two timings

A made-up rule: if a bar closes above its own open, it was an up bar, and the rule goes long for one bar. That fact is known only at the close. Version A buys at that bar's open and sells at its close, so it trades a bar it could only recognise after the bar had ended. Version B waits and trades the next bar, open to close, which is what a real order can do. Both use the same 9 made-up bars and were checked in node.

Version A adds up to 6.79% and version B to -4.79%. The only difference is when the trade is allowed to happen. Version A uses the close of a bar to decide a trade that began at its open, which no one can do. This is look-ahead bias, covered in its own guide. In an array language the same mistake can hide in a single reference to a future bar. If a backtest looks too good, the first thing to check is whether any value used on a bar could not have been known at that bar.

What to verify before trusting a backtest here

These checks apply to any platform of this kind. The answers come from the vendor's current documentation and from your own tests, not from this page.

How bots break on this kind of platform

• Look-ahead. The test knows something live cannot.
• Bridge failure. Signals do not reach the broker, or arrive twice.
• Mismatch between test and live logic. The bridge version is a rewrite.
• Data differences. Adjusted data in test, raw prices live.
• Overfitting by speed. Fast tests invite thousands of tries. See overfitting.

A safe order of work

Backtest with honest costs, forward test on new data, run on a simulated account for weeks, then trade a very small size and compare each step with the one before. See forward testing and paper trading.

What this page does not tell you

It does not tell you whether the product is suitable, what it costs, or how it behaves today. Those change and belong to the vendor. It does not give a strategy to trade. This page explains how it works in general terms. We have not tested it for this page. Check the vendor's current terms.