Insights ยท RCF Research

Tick Data vs Bar Data: Why Two Backtests of the Same Strategy Disagree

Run the same strategy on bar data and on tick data and you often get two different answers. The difference is not a rounding error. For some strategies it decides whether there is any edge at all.

By RCF Research Desk, Rushali Consulting Firm FZE LLCPublished 8 min read

What a price bar hides

Most charting and backtesting starts with bars: for each minute, hour or day, four numbers, the open, high, low and close (OHLC). Bars are compact and easy to work with. But a bar records where price went, not the order it went there.

Take a simple example. A strategy buys WTI at 70.00 with a stop-loss at 69.80 and a profit target at 70.25. The next one-minute bar opens at 70.05, reaches a high of 70.30 and a low of 69.75. Both the stop and the target were touched inside that minute. Did the trade win USD 0.25 or lose USD 0.20 per barrel? The bar cannot tell you. The backtester has to assume a path, and that assumption becomes part of your "result".

Tick data records every price change, in order, with its time. With ticks, the question has a real answer.

How MetaTrader 5 tests your strategy

The MT5 Strategy Tester offers several modelling modes, and the choice changes results:

  • Open prices only: fastest, uses just the bar open. Only suitable for systems that act once per bar.
  • 1 minute OHLC: builds a few synthetic price points from each one-minute bar. Faster, but the path inside each minute is guessed.
  • Every tick: generates synthetic ticks between one-minute bar points. More detailed, but still generated, not recorded.
  • Every tick based on real ticks: replays the actual tick history, including real bid and ask prices and therefore real spreads. This is the most realistic mode, and only as good as the tick data behind it.

The test report also shows a "history quality" figure. A low figure means the tester filled gaps with generated data, and the result deserves less trust.

Spreads are not constant

Bar-based tests usually assume a fixed spread. Real spreads in oil widen at exactly the moments many strategies want to trade: at the daily rollover, in thin overnight hours, and around news. For WTI, the US Energy Information Administration's weekly petroleum inventory report, normally released on Wednesdays at 10:30 a.m. New York time, can move prices sharply in seconds, and spreads widen with it.

A test that charges a fixed, narrow spread at those moments is quietly giving the strategy better prices than it would ever get live.

Slippage and order fills

Even with perfect price data, a backtest has to decide how orders fill. Most testers fill a market order at the next available price and a stop order exactly at its level. Live trading is less kind. When a stop triggers in a fast market, the order fills at the next price a counterparty will accept, which can be several cents per barrel worse. Limit orders have the opposite problem: in a backtest, a limit at 70.00 often counts as filled the moment price touches 70.00, while live, it may not fill at all unless price trades through it.

Tick data does not remove these effects, but it lets you model them honestly. You can see how far price moved in the seconds after a stop level was hit, and you can require price to trade through a limit before counting it as filled. Bar data hides both.

Where tick data comes from, and why sources differ

There is no single "true" tick history for a CFD on oil or a currency pair. Each broker or liquidity provider produces its own stream of quotes, with its own spreads and its own gaps. WTI CFDs are usually priced off the front-month futures contract, and the way a broker rolls from one contract month to the next creates price jumps that a careless test can mistake for real moves.

Three practical rules follow. First, test on data that is as close as possible to the feed you will trade on. Second, check how the data handles contract rollovers and missing periods. Third, when a strategy's edge is thin, test on more than one data source. If the result only survives on one of them, it is probably an artefact of that data, not an edge.

Which strategies are most affected

A daily trend-following system that holds positions for weeks barely notices the difference between bar and tick testing. The strategies most at risk are those where small price differences decide the outcome:

  • Short holding times: scalping and intraday systems.
  • Tight stops and targets, often both inside a single bar.
  • High win rates with small average wins.
  • Entries on limit orders or breakouts of exact levels.

A worked example: our own system

Our Capital Converter backtest on WTI is exactly the kind of strategy where testing accuracy matters. Over 5,513 trades from May 2020 to December 2025 it won 76.5% of the time, but its average win (USD 18.00) was much smaller than its average loss (USD 48.70) at a fixed 0.10 lot.

The edge per trade, the expectancy, was USD 2.28. Put differently: if costs on each trade were USD 2.28 higher than the test assumed, through wider spreads, slippage or a few unlucky intrabar paths, the entire six-year profit would disappear. That is why we test on tick-level data, why we publish the losing years, and why we are running the system on demo accounts before reporting live results.

Any strategy with a similar profile, high win rate and thin margin per trade, should never be judged on a bar-based backtest alone.

A checklist for a backtest you can trust

  1. Use real tick data with real bid and ask prices, not generated ticks.
  2. Check the history quality and the date range. Gaps are common in older data.
  3. Include commissions, swap (overnight financing) and realistic slippage.
  4. Test through known stress periods, not only calm ones.
  5. Keep part of the data out of sample. Build on one period, test on another.
  6. Compare the backtest with a forward test on a demo account. If they disagree, believe the forward test.
  7. Look at the full distribution: drawdown, losing streaks and the worst months, not only total return.

Test it yourself or let us do it. We sell tick-level data downloads for major currency pairs, gold and WTI, from USD 250 per instrument. Or send us your rules and our backtesting service runs them on our data and reports the results, from USD 350.

Important: This article is general education, not personal investment advice. Backtest figures are hypothetical, produced with hindsight, and do not reflect real execution, spreads, slippage or swap costs in full. Past performance, real or simulated, does not guarantee future results. Trading crude oil and other leveraged products can lose you money. Read our risk disclosure.